Skip to content

Unify Operators using Python Interface, Remove CMake#37

Merged
andrej merged 23 commits intoamd:develfrom
andrej:refactor
Dec 3, 2025
Merged

Unify Operators using Python Interface, Remove CMake#37
andrej merged 23 commits intoamd:develfrom
andrej:refactor

Conversation

@andrej
Copy link
Collaborator

@andrej andrej commented Nov 26, 2025

  • Drop the CMake-based structure of the existing operators in favor of the Python interface that we devised for the llama application, i.e. move from CMakeLists.txt to wrapping all operators in a Python AIE base operator.
  • Port all test.cpps to test.pys, aiming to factor out shared/previously duplicated code.
  • Restructure into an operators directory, where each operator has a subdirectory with a design.py (IRON code), op.py (operator wrapper/Python interface), reference.py and test.py.
  • Add a "dry run" facility to the Python interface that allows printing the compilation commands for artifact generation. I added this primarily because it was useful to verify against the CMake compilation commands. Removed, commits were 0fa2fca, 350dd5f, 2250e79 and 69994ff.
  • Small cleanups as I touched operators, e.g. disambiguate AIE (hardware) columns and matrix columns with clearer variable names.
  • Add test discovery script that works without CMake to run the test.pys in CI.

Remaining to do:

  • WIP: Add support for data types other than bf16 in the Python operator interface -- needed for the dequant kernel.
  • Port remaining two operators (memcpy, dequant).
  • Remove all remnants of CMake.
  • Make the CI call the new Python test discovery script rather than CMake.

@andrej andrej changed the title Unify Python operator abstractor with CMake operators Unify Python operator abstraction from Llama with CMake operators Nov 26, 2025
@andrej andrej changed the title Unify Python operator abstraction from Llama with CMake operators Unify Operators using Python Interface, Remove CMake Dec 1, 2025
@andrej andrej requested a review from hunhoffe December 1, 2025 23:30
Copy link
Collaborator

@hunhoffe hunhoffe left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think this is a great step forward!

I'd recommend subsequent PRs to:

  • Simplify test setup/config, probably using a library such as pytest
  • Simplify the python import/path setup in this repo

However, this PR is already large/chaotic, so I'd review my comments but push these tasks to future work.

@@ -0,0 +1,177 @@
#!/usr/bin/env python3
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

this could be a good intermediary step, but is there a reason we can't use something like pytest to find tests within a directory? This seems a little like reinventing the wheel when there are mature python testing libraries.

Another option could be to find classes with tests during module load time in the operator top level init file (e.g., do introspection to find which of the loaded things are testable).

I'm fine with this for now -- just brainstorming :)

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Claude whipped this together with me so these tests could fit into the existing testing framework that is a weird handmade solution of CMake+Python. Once it's all migrated to Python, I agree that we should move to something more established. Created issue #42 for this which I can tackle after merging this.

from operators.common.test_utils import run_test


regular_test_cases = [
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This could (in a later PR probably) be addressed with something like a pytest.parametrize with M, K, etc. parameters.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Agreed. This is ugly and for now just mirrors what was in the CMake directly and I didn't like doing it but as you already pointed out, the PR is already getting large.


def set_up_artifacts(self):
operator_dir = Path(__file__).parent
file_name_base = f"relu_{self.num_aie_columns}c_{self.num_channels}ch_{self.size}_{self.tile_size}t"
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I wonder if both file_name_base and mlir_artifact could have a default impl in the operator class that simplifies this setup for most operators (could. be overridden if needed).

I also wonder if the set_up_artifacts should be part of initialization, but that's probably a question for another day.

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It'd be great if there was some customization capability to be able to change the name / directory of object files and MLIR files.

python_cmd += "print(str(ctx.module))"
else:
python_cmd += "print(str(mlir_code))"
self.dry_run.append(f"python3 -c '{python_cmd}' > {artifact.path}")

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You can get the python executable with sys.executable

class AieccCompilationRule(CompilationRule):
def __init__(self, build_dir, peano_dir, mlir_aie_dir, *args, **kwargs):
self.build_dir = build_dir
self.aiecc_path = Path(mlir_aie_dir) / "bin" / "aiecc.py"

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

)

def compile(self, artifacts):
clang_path = Path(self.peano_dir) / "bin" / "clang++"

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@ypapadop-amd
Copy link

It's kind of big to review exhaustively, but I think it's a good step towards a reusable IRON framework. We have a number of functions for compilation, linking etc. in MLIR-AIE that can be reused.

@andrej
Copy link
Collaborator Author

andrej commented Dec 2, 2025

Thanks for the comments @ypapadop-amd. I agree it's large. The goal was to limit the scope of this PR to only removing the duplication of build scripts for the operators. Instead of having the CMake build scripts under example and the Python build scripts in application/llama, this moves everything to use the Python-based solution.

The compilation.py etc. were mostly just moved from where they currently sit in applications/llama to operators/common.

Erika and I are planning for this to be a multi-step refactor. Reusing existing code from elsewhere (more code deduplication) and coming to a clean, reusable operator interface are the next steps. Perhaps addressing things like operator artifact naming, and reusing existing IRON and JIT code should be part of those next steps.

@andrej andrej marked this pull request as ready for review December 3, 2025 18:54
@github-actions
Copy link
Contributor

github-actions bot commented Dec 3, 2025

📊 Test Results for Small Benchmark/Test Suite

6b841b9 (2025_12_03_21_43_03)

IRONCLAD

Tested on 2025_12_03_21_43_03 at commit 6b841b9.

Test Checks Latency (mean)Bandwidth (mean)GFLOP/s (mean)
axpy_1_cols_2_channels_2048_tile_2048_3.0 ✅ 5/5 129.84 0.10 n/a
axpy_2_cols_2_channels_2048_tile_1024_3.0 ✅ 5/5 117.16 0.11 n/a
axpy_4_cols_2_channels_2048_tile_512_3.0 ✅ 5/5 123.66 0.10 n/a
axpy_8_cols_2_channels_2048_tile_256_3.0 ✅ 5/5 136.40 0.09 n/a
dequant_1_cols_1_channels_2048_tile_2048 ✅ 5/5 116.86 0.05 n/a
dequant_1_cols_2_channels_2048_tile_1024 ✅ 5/5 109.40 0.05 n/a
dequant_2_cols_1_channels_2048_tile_1024 ✅ 5/5 114.08 0.05 n/a
dequant_2_cols_2_channels_2048_tile_512 ✅ 5/5 158.28 0.03 n/a
dequant_4_cols_1_channels_2048_tile_512 ✅ 5/5 140.38 0.04 n/a
dequant_4_cols_2_channels_2048_tile_256 ✅ 5/5 134.44 0.04 n/a
dequant_8_cols_1_channels_2048_tile_256 ✅ 5/5 154.26 0.03 n/a
dequant_8_cols_2_channels_2048_tile_128 ✅ 5/5 129.72 0.04 n/a
eltwise_add_1_cols_2_channels_2048_tile_2048 ✅ 5/5 111.68 0.11 n/a
eltwise_add_2_cols_2_channels_2048_tile_1024 ✅ 5/5 128.94 0.10 n/a
eltwise_add_4_cols_2_channels_2048_tile_512 ✅ 5/5 116.12 0.11 n/a
eltwise_add_8_cols_2_channels_2048_tile_256 ✅ 5/5 143.18 0.09 n/a
eltwise_mul_1_cols_2_channels_2048_tile_2048 ✅ 5/5 104.40 0.12 n/a
eltwise_mul_2_cols_2_channels_2048_tile_1024 ✅ 5/5 109.60 0.11 n/a
eltwise_mul_4_cols_2_channels_2048_tile_512 ✅ 5/5 112.98 0.11 n/a
eltwise_mul_8_cols_2_channels_2048_tile_256 ✅ 5/5 128.10 0.10 n/a
gelu_1_cols_1_channels_2048_tile_2048 ✅ 5/5 140.22 0.06 n/a
gelu_1_cols_2_channels_2048_tile_1024 ✅ 5/5 129.80 0.06 n/a
gelu_2_cols_1_channels_2048_tile_1024 ✅ 5/5 123.56 0.07 n/a
gelu_2_cols_2_channels_2048_tile_512 ✅ 5/5 120.24 0.07 n/a
gelu_4_cols_1_channels_2048_tile_512 ✅ 5/5 107.14 0.08 n/a
gelu_4_cols_2_channels_2048_tile_256 ✅ 5/5 126.64 0.07 n/a
gelu_8_cols_1_channels_2048_tile_256 ✅ 5/5 153.92 0.05 n/a
gelu_8_cols_2_channels_2048_tile_128 ✅ 5/5 164.68 0.05 n/a
gemm_2048x2048x2048_64x64x64_2_cols_0_bcolmaj_0_ccolmaj_0 ✅ 5/5 24536.28 1.03 n/a
gemm_2048x2048x2048_64x64x64_2_cols_0_bcolmaj_1_ccolmaj_0 ✅ 5/5 25829.18 0.97 n/a
gemm_2048x2048x2048_64x64x64_2_cols_1_bcolmaj_0_ccolmaj_0 ✅ 5/5 27516.14 0.91 n/a
layer_norm_1_cols_1_channels_2048_tile_2048 ✅ 5/5 128.18 0.06 n/a
layer_norm_1_cols_2_channels_2048_tile_1024 ✅ 5/5 129.78 0.07 n/a
layer_norm_2_cols_1_channels_2048_tile_1024 ✅ 5/5 124.30 0.07 n/a
layer_norm_2_cols_2_channels_2048_tile_512 ✅ 5/5 130.64 0.06 n/a
layer_norm_4_cols_1_channels_2048_tile_512 ✅ 5/5 126.48 0.07 n/a
layer_norm_4_cols_2_channels_2048_tile_256 ✅ 5/5 129.36 0.07 n/a
layer_norm_8_cols_1_channels_2048_tile_256 ✅ 5/5 110.72 0.08 n/a
layer_norm_8_cols_2_channels_2048_tile_128 ✅ 5/5 153.84 0.05 n/a
matrix_vector_mul_128x128_32_1col ✅ 5/5 n/a 0.25 n/a
matrix_vector_mul_2048x8192_1_1col ✅ 5/5 n/a 13.13 n/a
matrix_vector_mul_2048x8192_1_2col ✅ 5/5 n/a 23.97 n/a
matrix_vector_mul_2048x8192_1_4col ✅ 5/5 n/a 40.28 n/a
matrix_vector_mul_2048x8192_1_8col ✅ 5/5 n/a 45.54 n/a
matrix_vector_mul_8192x2048_4_1col ✅ 5/5 n/a 12.81 n/a
matrix_vector_mul_8192x2048_4_2col ✅ 5/5 n/a 24.34 n/a
matrix_vector_mul_8192x2048_4_4col ✅ 5/5 n/a 40.03 n/a
matrix_vector_mul_8192x2048_4_8col ✅ 5/5 n/a 45.14 n/a
mem_copy_16_cores_2_chans_2048_tile_128_False ✅ 5/5 131.42 0.06 n/a
mem_copy_1_cores_1_chans_2048_tile_2048_False ✅ 5/5 106.66 0.08 n/a
mem_copy_2_cores_1_chans_2048_tile_1024_False ✅ 5/5 123.68 0.07 n/a
mem_copy_2_cores_2_chans_2048_tile_1024_False ✅ 5/5 114.86 0.07 n/a
mem_copy_4_cores_1_chans_2048_tile_512_False ✅ 5/5 123.82 0.07 n/a
mem_copy_4_cores_2_chans_2048_tile_512_False ✅ 5/5 138.74 0.06 n/a
mem_copy_8_cores_1_chans_2048_tile_256_False ✅ 5/5 125.44 0.07 n/a
mem_copy_8_cores_2_chans_2048_tile_256_False ✅ 5/5 115.20 0.08 n/a
mha ✅ 5/5 47283.32 0.18 n/a
relu_1_cols_1_channels_2048_tile_2048 ✅ 5/5 106.64 0.08 n/a
relu_2_cols_1_channels_2048_tile_1024 ✅ 5/5 120.44 0.07 n/a
relu_4_cols_1_channels_2048_tile_512 ✅ 5/5 118.06 0.07 n/a
relu_8_cols_1_channels_2048_tile_256 ✅ 5/5 166.86 0.05 n/a
rms_norm_1_cols_1_channels_2048_tile_2048 ✅ 5/5 148.70 0.06 n/a
rms_norm_1_cols_2_channels_2048_tile_1024 ✅ 5/5 128.96 0.07 n/a
rms_norm_2_cols_1_channels_2048_tile_1024 ✅ 5/5 125.20 0.07 n/a
rms_norm_2_cols_2_channels_2048_tile_512 ✅ 5/5 104.20 0.08 n/a
rms_norm_4_cols_1_channels_2048_tile_512 ✅ 5/5 140.84 0.06 n/a
rms_norm_4_cols_2_channels_2048_tile_256 ✅ 5/5 133.58 0.06 n/a
rms_norm_8_cols_1_channels_2048_tile_256 ✅ 5/5 153.92 0.05 n/a
rms_norm_8_cols_2_channels_2048_tile_128 ✅ 5/5 129.10 0.07 n/a
rope_1_cols_2_channels_4096_tile_4096_0 ✅ 5/5 133.82 0.19 n/a
rope_2_cols_2_channels_4096_tile_2048_0 ✅ 5/5 132.70 0.19 n/a
rope_4_cols_2_channels_4096_tile_1024_0 ✅ 5/5 121.76 0.20 n/a
rope_8_cols_2_channels_4096_tile_512_0 ✅ 5/5 127.22 0.20 n/a
sigmoid_1_cols_1_channels_2048_tile_2048 ✅ 5/5 124.68 0.07 n/a
sigmoid_2_cols_1_channels_2048_tile_1024 ✅ 5/5 144.76 0.06 n/a
sigmoid_4_cols_1_channels_2048_tile_512 ✅ 5/5 139.34 0.06 n/a
sigmoid_8_cols_1_channels_2048_tile_256 ✅ 5/5 135.42 0.07 n/a
silu_1_cols_1_channels_2048_tile_2048 ✅ 5/5 134.02 0.06 n/a
silu_2_cols_1_channels_2048_tile_1024 ✅ 5/5 141.98 0.06 n/a
silu_4_cols_1_channels_2048_tile_512 ✅ 5/5 142.38 0.06 n/a
silu_8_cols_1_channels_2048_tile_256 ✅ 5/5 136.00 0.06 n/a
softmax_1_cols_2_channels_4096_tile_2048 ✅ 5/5 118.06 0.14 n/a
softmax_2_cols_2_channels_4096_tile_1024 ✅ 5/5 109.42 0.15 n/a
softmax_2_cols_2_channels_4096_tile_512 ✅ 5/5 127.18 0.14 n/a
swiglu_decode_1x2048x2048 ✅ 5/5 3893.34 0.00 n/a
tanh_1_cols_1_channels_2048_tile_2048 ✅ 5/5 130.60 0.06 n/a
tanh_2_cols_1_channels_2048_tile_1024 ✅ 5/5 127.02 0.07 n/a
tanh_4_cols_1_channels_2048_tile_512 ✅ 5/5 94.90 0.09 n/a
tanh_8_cols_1_channels_2048_tile_256 ✅ 5/5 137.18 0.06 n/a
transpose_2048_M_64_N_1_cols_1_channels_64_m_64_n_8_s ✅ 5/5 135.60 3.91 n/a
transpose_2048_M_64_N_1_cols_2_channels_64_m_64_n_8_s ✅ 5/5 136.14 3.92 n/a
weighted_rms_norm_1_cols_2_channels_2048_weights_2048 ✅ 5/5 116.22 0.07 n/a
weighted_rms_norm_2_cols_2_channels_2048_weights_1024 ✅ 5/5 126.94 0.07 n/a
weighted_rms_norm_4_cols_2_channels_2048_weights_512 ✅ 5/5 136.56 0.06 n/a
weighted_rms_norm_8_cols_2_channels_2048_weights_256 ✅ 5/5 132.80 0.06 n/a
📈 Trends (vs main branch) for Small Benchmark/Test Suite

6b841b9 (2025_12_03_21_43_03)

IRONCLAD Trends

axpy_1_cols_2_channels_2048_tile_2048_3.0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.11 (+32.42%)0.10 (+37.92%)0.09 (+32.06%)0.09 (+68.79%)0.01 (-13.28%)141.90 (-36.37%)129.84 (-23.62%)137.40 (-18.70%)111.90 (-18.91%)12.83 (-57.74%)
3a053d3 — 2025-11-26 01:12:160.08 (n/a)0.07 (n/a)0.07 (n/a)0.05 (n/a)0.01 (n/a)223.00 (n/a)170.00 (n/a)169.00 (n/a)138.00 (n/a)30.37 (n/a)

axpy_2_cols_2_channels_2048_tile_1024_3.0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.13 (+11.04%)0.11 (+23.99%)0.11 (+29.81%)0.09 (+37.97%)0.02 (-14.95%)133.90 (-22.15%)117.16 (-14.98%)116.60 (-17.30%)91.90 (-3.26%)15.65 (-36.82%)
3a053d3 — 2025-11-26 01:12:160.12 (n/a)0.09 (n/a)0.08 (n/a)0.07 (n/a)0.02 (n/a)172.00 (n/a)137.80 (n/a)141.00 (n/a)95.00 (n/a)24.77 (n/a)

axpy_4_cols_2_channels_2048_tile_512_3.0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.10 (+8.57%)0.10 (+20.46%)0.10 (+22.88%)0.10 (+36.15%)0.00 (-80.35%)127.00 (-21.12%)123.66 (-12.05%)124.10 (-12.61%)119.70 (-1.07%)2.36 (-85.62%)
3a053d3 — 2025-11-26 01:12:160.09 (n/a)0.08 (n/a)0.08 (n/a)0.07 (n/a)0.01 (n/a)161.00 (n/a)140.60 (n/a)142.00 (n/a)121.00 (n/a)16.43 (n/a)

axpy_8_cols_2_channels_2048_tile_256_3.0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.12 (+82.13%)0.09 (+56.46%)0.09 (+47.30%)0.08 (+68.74%)0.02 (+138.45%)155.30 (-36.35%)136.40 (-30.55%)139.20 (-27.12%)98.50 (-41.02%)20.49 (-20.15%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.06 (n/a)0.05 (n/a)0.01 (n/a)244.00 (n/a)196.40 (n/a)191.00 (n/a)167.00 (n/a)25.66 (n/a)

dequant_1_cols_1_channels_2048_tile_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.07 (+70.54%)0.05 (+36.99%)0.04 (+5.18%)0.04 (+69.97%)0.01 (+64.88%)144.00 (-36.84%)116.86 (-22.20%)126.60 (+2.10%)71.80 (-37.02%)24.45 (-42.40%)
3a053d3 — 2025-11-26 01:12:160.04 (n/a)0.03 (n/a)0.04 (n/a)0.02 (n/a)0.01 (n/a)228.00 (n/a)150.20 (n/a)124.00 (n/a)114.00 (n/a)42.45 (n/a)

dequant_1_cols_2_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.07 (+81.52%)0.05 (+40.24%)0.04 (+6.13%)0.04 (+15.73%)0.02 (+322.31%)144.70 (-7.24%)109.40 (-17.37%)124.40 (+1.14%)71.00 (-40.83%)30.19 (+111.23%)
3a053d3 — 2025-11-26 01:12:160.04 (n/a)0.04 (n/a)0.04 (n/a)0.03 (n/a)0.00 (n/a)156.00 (n/a)132.40 (n/a)123.00 (n/a)120.00 (n/a)14.29 (n/a)

dequant_2_cols_1_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.06 (+65.83%)0.05 (+55.12%)0.05 (+54.66%)0.04 (+53.16%)0.01 (+111.75%)128.30 (-29.89%)114.08 (-30.35%)113.20 (-30.55%)91.90 (-35.28%)13.11 (-10.44%)
3a053d3 — 2025-11-26 01:12:160.03 (n/a)0.03 (n/a)0.03 (n/a)0.03 (n/a)0.00 (n/a)183.00 (n/a)163.80 (n/a)163.00 (n/a)142.00 (n/a)14.63 (n/a)

dequant_2_cols_2_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.04 (+13.42%)0.03 (-0.20%)0.03 (-6.95%)0.03 (-2.19%)0.01 (+52.57%)184.40 (+9.76%)158.28 (+8.86%)163.90 (+15.42%)117.40 (-5.32%)22.35 (+41.51%)
3a053d3 — 2025-11-26 01:12:160.04 (n/a)0.03 (n/a)0.03 (n/a)0.03 (n/a)0.00 (n/a)168.00 (n/a)145.40 (n/a)142.00 (n/a)124.00 (n/a)15.79 (n/a)

dequant_4_cols_1_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.04 (-5.02%)0.04 (+16.32%)0.04 (+19.41%)0.03 (+20.16%)0.01 (-27.77%)167.10 (-10.64%)140.38 (-9.67%)143.90 (-10.06%)117.60 (+13.08%)19.44 (-31.08%)
3a053d3 — 2025-11-26 01:12:160.05 (n/a)0.03 (n/a)0.03 (n/a)0.03 (n/a)0.01 (n/a)187.00 (n/a)155.40 (n/a)160.00 (n/a)104.00 (n/a)28.20 (n/a)

dequant_4_cols_2_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.05 (+32.61%)0.04 (+18.17%)0.04 (+19.52%)0.03 (-1.27%)0.01 (+197.90%)167.50 (+8.77%)134.44 (-7.41%)132.10 (-10.14%)105.30 (-19.00%)19.92 (+147.98%)
3a053d3 — 2025-11-26 01:12:160.04 (n/a)0.03 (n/a)0.03 (n/a)0.03 (n/a)0.00 (n/a)154.00 (n/a)145.20 (n/a)147.00 (n/a)130.00 (n/a)8.03 (n/a)

dequant_8_cols_1_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.04 (-16.42%)0.03 (-0.05%)0.03 (-8.19%)0.03 (+23.69%)0.00 (-61.53%)178.00 (-13.17%)154.26 (+0.96%)153.20 (+16.95%)129.80 (+28.51%)15.34 (-62.48%)
3a053d3 — 2025-11-26 01:12:160.05 (n/a)0.03 (n/a)0.04 (n/a)0.02 (n/a)0.01 (n/a)205.00 (n/a)152.80 (n/a)131.00 (n/a)101.00 (n/a)40.88 (n/a)

dequant_8_cols_2_channels_2048_tile_128

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.06 (+87.64%)0.04 (+44.06%)0.04 (+36.92%)0.03 (+15.52%)0.01 (+406.87%)172.00 (-7.03%)129.72 (-22.51%)126.30 (-21.55%)90.40 (-42.78%)26.17 (+148.77%)
3a053d3 — 2025-11-26 01:12:160.03 (n/a)0.03 (n/a)0.03 (n/a)0.03 (n/a)0.00 (n/a)185.00 (n/a)167.40 (n/a)161.00 (n/a)158.00 (n/a)10.52 (n/a)

eltwise_add_1_cols_2_channels_2048_tile_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.12 (+9.50%)0.11 (+39.53%)0.11 (+49.26%)0.10 (+89.94%)0.01 (-65.10%)122.70 (-43.46%)111.68 (-27.01%)109.30 (-28.09%)103.00 (-1.90%)6.68 (-82.00%)
3a053d3 — 2025-11-26 01:12:160.11 (n/a)0.08 (n/a)0.08 (n/a)0.05 (n/a)0.02 (n/a)217.00 (n/a)153.00 (n/a)152.00 (n/a)105.00 (n/a)37.09 (n/a)

eltwise_add_2_cols_2_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.10 (+3.81%)0.10 (+9.68%)0.10 (+9.49%)0.09 (+31.47%)0.01 (-45.49%)140.50 (-18.31%)128.94 (-3.78%)123.60 (-1.90%)117.90 (+3.42%)9.22 (-56.49%)
3a053d3 — 2025-11-26 01:12:160.10 (n/a)0.09 (n/a)0.09 (n/a)0.07 (n/a)0.01 (n/a)172.00 (n/a)134.00 (n/a)126.00 (n/a)114.00 (n/a)21.19 (n/a)

eltwise_add_4_cols_2_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.15 (+64.34%)0.11 (+43.49%)0.12 (+57.39%)0.08 (+22.54%)0.03 (+173.82%)152.50 (-12.36%)116.12 (-21.22%)105.70 (-31.81%)80.40 (-34.63%)30.59 (+57.35%)
3a053d3 — 2025-11-26 01:12:160.09 (n/a)0.08 (n/a)0.07 (n/a)0.07 (n/a)0.01 (n/a)174.00 (n/a)147.40 (n/a)155.00 (n/a)123.00 (n/a)19.44 (n/a)

eltwise_add_8_cols_2_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.13 (+57.54%)0.09 (+22.51%)0.08 (+12.86%)0.07 (+5.94%)0.02 (+259.59%)179.40 (+1.36%)143.18 (-8.80%)149.40 (-4.84%)97.50 (-31.82%)28.65 (+130.00%)
3a053d3 — 2025-11-26 01:12:160.08 (n/a)0.07 (n/a)0.07 (n/a)0.06 (n/a)0.01 (n/a)177.00 (n/a)157.00 (n/a)157.00 (n/a)143.00 (n/a)12.46 (n/a)

eltwise_mul_1_cols_2_channels_2048_tile_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.13 (+25.14%)0.12 (+53.56%)0.12 (+51.74%)0.11 (+96.86%)0.01 (-49.44%)116.70 (-45.47%)104.40 (-32.99%)104.70 (-29.26%)93.50 (-14.22%)7.55 (-78.01%)
3a053d3 — 2025-11-26 01:12:160.10 (n/a)0.08 (n/a)0.08 (n/a)0.05 (n/a)0.02 (n/a)214.00 (n/a)155.80 (n/a)148.00 (n/a)109.00 (n/a)34.31 (n/a)

eltwise_mul_2_cols_2_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.13 (+45.32%)0.11 (+51.58%)0.12 (+60.15%)0.08 (+25.95%)0.02 (+99.32%)145.80 (-14.74%)109.60 (-28.27%)101.90 (-32.96%)96.80 (-26.11%)18.33 (+17.86%)
3a053d3 — 2025-11-26 01:12:160.09 (n/a)0.08 (n/a)0.08 (n/a)0.07 (n/a)0.01 (n/a)171.00 (n/a)152.80 (n/a)152.00 (n/a)131.00 (n/a)15.55 (n/a)

eltwise_mul_4_cols_2_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.13 (+56.36%)0.11 (+50.60%)0.11 (+45.58%)0.09 (+43.16%)0.01 (+138.72%)132.80 (-24.97%)112.98 (-28.04%)112.80 (-26.27%)96.80 (-31.35%)14.19 (+12.17%)
3a053d3 — 2025-11-26 01:12:160.08 (n/a)0.07 (n/a)0.07 (n/a)0.06 (n/a)0.01 (n/a)177.00 (n/a)157.00 (n/a)153.00 (n/a)141.00 (n/a)12.65 (n/a)

eltwise_mul_8_cols_2_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.14 (+36.16%)0.10 (+42.89%)0.11 (+84.66%)0.07 (+22.44%)0.03 (+64.28%)173.60 (-12.32%)128.10 (-22.55%)108.20 (-41.83%)89.10 (-21.15%)35.75 (+10.83%)
3a053d3 — 2025-11-26 01:12:160.10 (n/a)0.07 (n/a)0.06 (n/a)0.06 (n/a)0.02 (n/a)198.00 (n/a)165.40 (n/a)186.00 (n/a)113.00 (n/a)32.25 (n/a)

gelu_1_cols_1_channels_2048_tile_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.07 (-10.95%)0.06 (+5.47%)0.06 (+3.03%)0.04 (+46.25%)0.01 (-48.79%)182.60 (-31.61%)140.22 (-12.91%)133.90 (-2.97%)114.50 (+12.25%)22.79 (-60.67%)
3a053d3 — 2025-11-26 01:12:160.08 (n/a)0.06 (n/a)0.06 (n/a)0.03 (n/a)0.02 (n/a)267.00 (n/a)161.00 (n/a)138.00 (n/a)102.00 (n/a)57.94 (n/a)

gelu_1_cols_2_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.09 (-8.52%)0.06 (-2.81%)0.06 (+2.49%)0.05 (+10.49%)0.01 (-26.42%)149.30 (-9.52%)129.80 (+0.78%)132.70 (-2.43%)96.20 (+9.32%)19.81 (-25.26%)
3a053d3 — 2025-11-26 01:12:160.09 (n/a)0.07 (n/a)0.06 (n/a)0.05 (n/a)0.02 (n/a)165.00 (n/a)128.80 (n/a)136.00 (n/a)88.00 (n/a)26.50 (n/a)

gelu_2_cols_1_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.10 (+21.48%)0.07 (+4.46%)0.06 (-13.40%)0.05 (+45.19%)0.02 (+1.93%)158.40 (-31.13%)123.56 (-8.20%)128.20 (+15.50%)81.50 (-17.68%)25.27 (-47.67%)
3a053d3 — 2025-11-26 01:12:160.08 (n/a)0.07 (n/a)0.07 (n/a)0.04 (n/a)0.02 (n/a)230.00 (n/a)134.60 (n/a)111.00 (n/a)99.00 (n/a)48.30 (n/a)

gelu_2_cols_2_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.08 (+26.39%)0.07 (+33.34%)0.06 (+14.44%)0.06 (+63.32%)0.01 (-20.55%)138.40 (-38.76%)120.24 (-28.77%)127.60 (-12.60%)97.30 (-20.89%)17.69 (-62.37%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.05 (n/a)0.06 (n/a)0.04 (n/a)0.01 (n/a)226.00 (n/a)168.80 (n/a)146.00 (n/a)123.00 (n/a)47.02 (n/a)

gelu_4_cols_1_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.09 (+20.53%)0.08 (+25.87%)0.07 (+15.10%)0.07 (+67.31%)0.01 (-36.05%)118.30 (-40.25%)107.14 (-22.92%)109.50 (-13.10%)94.60 (-17.02%)9.14 (-69.73%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.07 (n/a)0.04 (n/a)0.01 (n/a)198.00 (n/a)139.00 (n/a)126.00 (n/a)114.00 (n/a)30.19 (n/a)

gelu_4_cols_2_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.08 (+25.65%)0.07 (+24.19%)0.06 (+23.71%)0.06 (+19.28%)0.01 (+37.31%)144.20 (-16.16%)126.64 (-19.34%)127.70 (-19.18%)108.20 (-20.44%)12.41 (-9.71%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.05 (n/a)0.05 (n/a)0.05 (n/a)0.00 (n/a)172.00 (n/a)157.00 (n/a)158.00 (n/a)136.00 (n/a)13.74 (n/a)

gelu_8_cols_1_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.07 (-0.28%)0.05 (-6.98%)0.05 (-16.34%)0.04 (+28.93%)0.01 (-27.86%)186.20 (-22.42%)153.92 (+3.03%)150.60 (+19.52%)115.30 (+0.26%)24.72 (-46.40%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.07 (n/a)0.03 (n/a)0.01 (n/a)240.00 (n/a)149.40 (n/a)126.00 (n/a)115.00 (n/a)46.13 (n/a)

gelu_8_cols_2_channels_2048_tile_128

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.06 (+6.46%)0.05 (+1.73%)0.05 (-1.63%)0.04 (+2.96%)0.01 (+31.48%)193.30 (-2.86%)164.68 (-1.15%)162.70 (+1.69%)139.00 (-6.08%)20.64 (+17.74%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.05 (n/a)0.05 (n/a)0.04 (n/a)0.00 (n/a)199.00 (n/a)166.60 (n/a)160.00 (n/a)148.00 (n/a)17.53 (n/a)

gemm_2048x2048x2048_64x64x32_8_cols_0_bcolmaj_0_ccolmaj_0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
3a053d3 — 2025-11-26 01:12:163.72 (-0.11%)3.55 (-2.75%)3.56 (-2.75%)3.36 (-5.33%)0.12 (+76.10%)2725.23 (-0.11%)2602.52 (-2.75%)2612.11 (-2.75%)2459.89 (-5.33%)86.27 (+76.10%)6984.00 (+5.63%)6608.60 (+2.91%)6577.00 (+2.83%)6304.00 (+0.11%)221.87 (+87.15%)
9e98049 — 2025-11-25 20:50:553.72 (n/a)3.65 (n/a)3.66 (n/a)3.54 (n/a)0.07 (n/a)2728.26 (n/a)2676.06 (n/a)2686.03 (n/a)2598.29 (n/a)48.99 (n/a)6612.00 (n/a)6422.00 (n/a)6396.00 (n/a)6297.00 (n/a)118.55 (n/a)

gemm_2048x2048x2048_64x64x32_8_cols_0_bcolmaj_1_ccolmaj_0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
3a053d3 — 2025-11-26 01:12:163.40 (-3.14%)3.38 (-0.82%)3.38 (-0.82%)3.35 (+4.93%)0.02 (-84.51%)2493.81 (-3.14%)2475.27 (-0.82%)2479.06 (-0.82%)2453.92 (+4.93%)13.33 (-84.51%)7001.00 (-4.70%)6940.80 (+0.71%)6930.00 (+0.83%)6889.00 (+3.24%)37.43 (-84.76%)
9e98049 — 2025-11-25 20:50:553.51 (n/a)3.40 (n/a)3.41 (n/a)3.19 (n/a)0.12 (n/a)2574.53 (n/a)2495.72 (n/a)2499.62 (n/a)2338.67 (n/a)86.02 (n/a)7346.00 (n/a)6892.20 (n/a)6873.00 (n/a)6673.00 (n/a)245.57 (n/a)

gemm_2048x2048x2048_64x64x32_8_cols_1_bcolmaj_0_ccolmaj_0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
3a053d3 — 2025-11-26 01:12:163.02 (+0.52%)2.87 (-2.05%)2.86 (-2.24%)2.70 (-6.16%)0.11 (+131.66%)2216.19 (+0.52%)2104.44 (-2.05%)2094.08 (-2.24%)1980.84 (-6.16%)80.25 (+131.67%)8673.00 (+6.56%)8175.60 (+2.21%)8204.00 (+2.29%)7752.00 (-0.51%)314.03 (+145.36%)
9e98049 — 2025-11-25 20:50:553.01 (n/a)2.93 (n/a)2.92 (n/a)2.88 (n/a)0.05 (n/a)2204.81 (n/a)2148.41 (n/a)2142.13 (n/a)2110.81 (n/a)34.64 (n/a)8139.00 (n/a)7998.60 (n/a)8020.00 (n/a)7792.00 (n/a)127.99 (n/a)

gemm_2048x2048x2048_64x64x64_2_cols_0_bcolmaj_0_ccolmaj_0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:571.03 (n/a)1.03 (n/a)1.03 (n/a)1.02 (n/a)0.00 (n/a)24562.90 (n/a)24536.28 (n/a)24539.30 (n/a)24511.20 (n/a)17.88 (n/a)

gemm_2048x2048x2048_64x64x64_2_cols_0_bcolmaj_1_ccolmaj_0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.98 (n/a)0.97 (n/a)0.97 (n/a)0.97 (n/a)0.01 (n/a)25915.90 (n/a)25829.18 (n/a)25891.00 (n/a)25560.00 (n/a)135.01 (n/a)

gemm_2048x2048x2048_64x64x64_2_cols_1_bcolmaj_0_ccolmaj_0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.92 (n/a)0.91 (n/a)0.92 (n/a)0.91 (n/a)0.00 (n/a)27684.10 (n/a)27516.14 (n/a)27480.50 (n/a)27390.60 (n/a)106.41 (n/a)

gemm_2048x2048x2048_64x64x64_8_cols_0_bcolmaj_0_ccolmaj_0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
31ab24e — 2025-11-22 02:01:563.65 (-1.06%)3.59 (+0.83%)3.63 (-0.26%)3.45 (+2.66%)0.07 (-44.71%)2678.08 (-1.06%)2632.13 (+0.83%)2658.60 (-0.26%)2528.31 (+2.66%)53.86 (-44.71%)6795.00 (-2.59%)6529.80 (-0.92%)6462.00 (+0.26%)6415.00 (+1.07%)137.14 (-45.27%)
d9a64c5 — 2025-11-21 20:49:353.69 (n/a)3.56 (n/a)3.64 (n/a)3.36 (n/a)0.13 (n/a)2706.77 (n/a)2610.50 (n/a)2665.61 (n/a)2462.71 (n/a)97.41 (n/a)6976.00 (n/a)6590.40 (n/a)6445.00 (n/a)6347.00 (n/a)250.58 (n/a)

layer_norm_1_cols_1_channels_2048_tile_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.07 (-22.26%)0.06 (-25.94%)0.06 (-26.24%)0.06 (-25.31%)0.01 (-15.28%)140.10 (-10.76%)128.18 (-9.86%)130.10 (-9.65%)111.50 (-14.23%)10.70 (-0.71%)
3a053d3 — 2025-11-26 01:12:160.09 (n/a)0.09 (n/a)0.09 (n/a)0.08 (n/a)0.01 (n/a)157.00 (n/a)142.20 (n/a)144.00 (n/a)130.00 (n/a)10.78 (n/a)

layer_norm_1_cols_2_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.09 (-10.89%)0.07 (-20.03%)0.05 (-34.56%)0.05 (-29.49%)0.02 (+89.90%)160.70 (-5.47%)129.78 (-12.31%)149.70 (+1.84%)92.80 (-25.16%)30.41 (+95.45%)
3a053d3 — 2025-11-26 01:12:160.10 (n/a)0.08 (n/a)0.08 (n/a)0.07 (n/a)0.01 (n/a)170.00 (n/a)148.00 (n/a)147.00 (n/a)124.00 (n/a)15.56 (n/a)

layer_norm_2_cols_1_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.08 (+5.92%)0.07 (+1.10%)0.07 (-1.03%)0.05 (+7.06%)0.01 (-2.65%)162.50 (-37.74%)124.30 (-34.44%)113.20 (-32.62%)96.90 (-37.08%)23.98 (-41.28%)
3a053d3 — 2025-11-26 01:12:160.08 (n/a)0.07 (n/a)0.07 (n/a)0.05 (n/a)0.01 (n/a)261.00 (n/a)189.60 (n/a)168.00 (n/a)154.00 (n/a)40.83 (n/a)

layer_norm_2_cols_2_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.07 (-15.89%)0.06 (-12.13%)0.06 (-24.21%)0.06 (+27.36%)0.01 (-50.04%)145.00 (-47.65%)130.64 (-27.34%)140.70 (-12.06%)112.60 (-20.70%)14.21 (-71.06%)
3a053d3 — 2025-11-26 01:12:160.09 (n/a)0.07 (n/a)0.08 (n/a)0.04 (n/a)0.01 (n/a)277.00 (n/a)179.80 (n/a)160.00 (n/a)142.00 (n/a)49.10 (n/a)

layer_norm_4_cols_1_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.08 (-36.13%)0.07 (-20.35%)0.07 (-9.80%)0.05 (-13.02%)0.01 (-55.48%)157.10 (-23.37%)126.48 (-18.82%)118.30 (-26.06%)108.50 (+4.33%)18.02 (-44.99%)
3a053d3 — 2025-11-26 01:12:160.12 (n/a)0.08 (n/a)0.08 (n/a)0.06 (n/a)0.02 (n/a)205.00 (n/a)155.80 (n/a)160.00 (n/a)104.00 (n/a)32.76 (n/a)

layer_norm_4_cols_2_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.09 (+33.54%)0.07 (+14.00%)0.07 (+8.92%)0.05 (+9.04%)0.01 (+55.93%)163.20 (-38.88%)129.36 (-40.77%)125.50 (-38.78%)93.90 (-50.05%)22.64 (-28.72%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.06 (n/a)0.05 (n/a)0.01 (n/a)267.00 (n/a)218.40 (n/a)205.00 (n/a)188.00 (n/a)31.77 (n/a)

layer_norm_8_cols_1_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.09 (-18.79%)0.08 (-4.96%)0.08 (+5.90%)0.06 (-3.48%)0.01 (-36.88%)138.10 (-30.95%)110.72 (-31.06%)103.20 (-37.07%)92.80 (-17.88%)16.15 (-44.50%)
3a053d3 — 2025-11-26 01:12:160.11 (n/a)0.08 (n/a)0.07 (n/a)0.06 (n/a)0.02 (n/a)200.00 (n/a)160.60 (n/a)164.00 (n/a)113.00 (n/a)29.10 (n/a)

layer_norm_8_cols_2_channels_2048_tile_128

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.06 (-11.47%)0.05 (-17.54%)0.05 (-22.86%)0.05 (-21.25%)0.00 (+34.95%)171.00 (-15.35%)153.84 (-18.78%)159.00 (-13.59%)133.30 (-24.69%)13.32 (+26.42%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.07 (n/a)0.07 (n/a)0.06 (n/a)0.00 (n/a)202.00 (n/a)189.40 (n/a)184.00 (n/a)177.00 (n/a)10.54 (n/a)

matrix_vector_mul_128x128_32_1col

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.32 (+538.09%)0.25 (+451.74%)0.22 (+394.73%)0.19 (+351.45%)0.05 (+1563.37%)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)
3a053d3 — 2025-11-26 01:12:160.05 (n/a)0.05 (n/a)0.04 (n/a)0.04 (n/a)0.00 (n/a)0.05 (n/a)0.04 (n/a)0.04 (n/a)0.04 (n/a)0.00 (n/a)790.00 (n/a)738.40 (n/a)749.00 (n/a)659.00 (n/a)50.18 (n/a)

matrix_vector_mul_2048x8192_1_1col

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:5713.28 (+17.57%)13.13 (+21.23%)13.21 (+22.40%)12.68 (+20.52%)0.23 (-14.56%)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)
3a053d3 — 2025-11-26 01:12:1611.30 (n/a)10.83 (n/a)10.80 (n/a)10.52 (n/a)0.27 (n/a)11.29 (n/a)10.83 (n/a)10.79 (n/a)10.52 (n/a)0.27 (n/a)3191.00 (n/a)3101.20 (n/a)3110.00 (n/a)2972.00 (n/a)74.99 (n/a)

matrix_vector_mul_2048x8192_1_2col

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:5724.51 (+39.15%)23.97 (+44.96%)24.18 (+40.23%)22.68 (+66.11%)0.66 (-55.16%)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)
3a053d3 — 2025-11-26 01:12:1617.62 (n/a)16.54 (n/a)17.24 (n/a)13.65 (n/a)1.47 (n/a)17.60 (n/a)16.53 (n/a)17.23 (n/a)13.65 (n/a)1.47 (n/a)2459.00 (n/a)2048.80 (n/a)1947.00 (n/a)1906.00 (n/a)207.97 (n/a)

matrix_vector_mul_2048x8192_1_4col

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:5742.47 (+107.69%)40.28 (+105.11%)40.05 (+107.54%)38.69 (+102.94%)1.23 (+111.65%)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)
3a053d3 — 2025-11-26 01:12:1620.45 (n/a)19.64 (n/a)19.30 (n/a)19.07 (n/a)0.58 (n/a)20.44 (n/a)19.63 (n/a)19.28 (n/a)19.05 (n/a)0.58 (n/a)1761.00 (n/a)1711.20 (n/a)1740.00 (n/a)1642.00 (n/a)49.93 (n/a)

matrix_vector_mul_2048x8192_1_8col

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:5747.47 (+147.58%)45.54 (+143.79%)45.52 (+144.70%)44.22 (+143.26%)1.13 (+205.20%)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)
3a053d3 — 2025-11-26 01:12:1619.17 (n/a)18.68 (n/a)18.60 (n/a)18.18 (n/a)0.37 (n/a)19.16 (n/a)18.67 (n/a)18.59 (n/a)18.17 (n/a)0.37 (n/a)1847.00 (n/a)1798.20 (n/a)1805.00 (n/a)1751.00 (n/a)35.70 (n/a)

matrix_vector_mul_8192x2048_4_1col

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:5713.47 (+23.10%)12.81 (+20.91%)13.34 (+24.09%)10.75 (+10.27%)1.04 (+140.41%)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)
3a053d3 — 2025-11-26 01:12:1610.94 (n/a)10.60 (n/a)10.75 (n/a)9.75 (n/a)0.43 (n/a)10.94 (n/a)10.59 (n/a)10.74 (n/a)9.74 (n/a)0.43 (n/a)3445.00 (n/a)3174.00 (n/a)3124.00 (n/a)3068.00 (n/a)137.12 (n/a)

matrix_vector_mul_8192x2048_4_2col

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:5724.67 (+47.56%)24.34 (+47.88%)24.36 (+48.95%)23.85 (+46.99%)0.27 (+32.63%)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)
3a053d3 — 2025-11-26 01:12:1616.72 (n/a)16.46 (n/a)16.35 (n/a)16.23 (n/a)0.21 (n/a)16.71 (n/a)16.45 (n/a)16.34 (n/a)16.22 (n/a)0.21 (n/a)2069.00 (n/a)2040.40 (n/a)2053.00 (n/a)2008.00 (n/a)25.37 (n/a)

matrix_vector_mul_8192x2048_4_4col

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:5741.13 (+93.41%)40.03 (+102.25%)40.00 (+94.21%)38.43 (+113.11%)0.93 (-28.84%)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)
3a053d3 — 2025-11-26 01:12:1621.26 (n/a)19.79 (n/a)20.60 (n/a)18.03 (n/a)1.31 (n/a)21.25 (n/a)19.78 (n/a)20.59 (n/a)18.02 (n/a)1.31 (n/a)1862.00 (n/a)1704.00 (n/a)1630.00 (n/a)1579.00 (n/a)115.16 (n/a)

matrix_vector_mul_8192x2048_4_8col

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:5746.32 (+140.61%)45.14 (+170.40%)44.92 (+177.22%)44.01 (+217.73%)0.79 (-60.71%)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)
3a053d3 — 2025-11-26 01:12:1619.25 (n/a)16.70 (n/a)16.20 (n/a)13.85 (n/a)2.00 (n/a)19.24 (n/a)16.68 (n/a)16.19 (n/a)13.84 (n/a)2.00 (n/a)2424.00 (n/a)2040.60 (n/a)2072.00 (n/a)1744.00 (n/a)247.94 (n/a)

mem_copy_16_cores_2_chans_2048_tile_128_False

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.07 (+43.51%)0.06 (+44.34%)0.06 (+45.28%)0.06 (+37.42%)0.00 (+72.49%)146.90 (-21.86%)131.42 (-25.50%)129.30 (-26.11%)120.50 (-25.16%)9.02 (-5.87%)
3a053d3 — 2025-11-26 01:12:160.05 (n/a)0.04 (n/a)0.04 (n/a)0.04 (n/a)0.00 (n/a)188.00 (n/a)176.40 (n/a)175.00 (n/a)161.00 (n/a)9.58 (n/a)

mem_copy_1_cols_1_channels_2048_tile_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
003b49d — 2025-11-14 19:30:540.05 (-16.99%)0.05 (-12.47%)0.05 (-7.32%)0.04 (-11.17%)0.00 (-29.61%)197.00 (+12.57%)171.00 (+13.70%)164.00 (+7.89%)153.00 (+20.47%)17.49 (-4.32%)
0999a9a — 2025-11-13 19:24:080.06 (n/a)0.05 (n/a)0.05 (n/a)0.04 (n/a)0.01 (n/a)175.00 (n/a)150.40 (n/a)152.00 (n/a)127.00 (n/a)18.28 (n/a)

mem_copy_1_cols_2_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
003b49d — 2025-11-14 19:30:540.06 (+5.56%)0.05 (-1.59%)0.05 (-1.39%)0.03 (-14.07%)0.01 (+25.03%)263.00 (+16.37%)166.00 (+4.53%)144.00 (+1.41%)126.00 (-5.26%)49.38 (+43.17%)
0999a9a — 2025-11-13 19:24:080.06 (n/a)0.05 (n/a)0.05 (n/a)0.03 (n/a)0.01 (n/a)226.00 (n/a)158.80 (n/a)142.00 (n/a)133.00 (n/a)34.49 (n/a)

mem_copy_1_cores_1_chans_2048_tile_2048_False

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.10 (+29.56%)0.08 (+36.66%)0.07 (+22.01%)0.06 (+45.86%)0.01 (+19.30%)132.50 (-26.39%)106.66 (-22.15%)110.90 (-11.98%)86.20 (-17.12%)17.06 (-34.80%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.06 (n/a)0.04 (n/a)0.01 (n/a)180.00 (n/a)137.00 (n/a)126.00 (n/a)104.00 (n/a)26.17 (n/a)

mem_copy_2_cols_1_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
003b49d — 2025-11-14 19:30:540.06 (-15.32%)0.05 (-17.09%)0.05 (-23.81%)0.04 (-15.54%)0.01 (-18.18%)193.00 (+18.40%)160.80 (+20.36%)168.00 (+31.25%)124.00 (+18.10%)22.92 (+11.39%)
0999a9a — 2025-11-13 19:24:080.07 (n/a)0.06 (n/a)0.06 (n/a)0.05 (n/a)0.01 (n/a)163.00 (n/a)133.60 (n/a)128.00 (n/a)105.00 (n/a)20.58 (n/a)

mem_copy_2_cols_2_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
003b49d — 2025-11-14 19:30:540.06 (-23.02%)0.05 (-6.98%)0.05 (-2.13%)0.04 (-15.31%)0.01 (-33.29%)196.00 (+18.07%)147.00 (+6.52%)141.00 (+2.17%)126.00 (+29.90%)25.27 (+8.13%)
0999a9a — 2025-11-13 19:24:080.08 (n/a)0.06 (n/a)0.06 (n/a)0.05 (n/a)0.01 (n/a)166.00 (n/a)138.00 (n/a)138.00 (n/a)97.00 (n/a)23.37 (n/a)

mem_copy_2_cores_1_chans_2048_tile_1024_False

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.08 (+22.79%)0.07 (+22.50%)0.07 (+11.32%)0.06 (+55.25%)0.01 (-29.24%)139.00 (-30.85%)123.68 (-14.70%)123.50 (-3.52%)104.10 (-12.52%)11.94 (-60.67%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.05 (n/a)0.06 (n/a)0.04 (n/a)0.01 (n/a)201.00 (n/a)145.00 (n/a)128.00 (n/a)119.00 (n/a)30.35 (n/a)

mem_copy_2_cores_2_chans_2048_tile_1024_False

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.08 (+31.64%)0.07 (+30.45%)0.07 (+33.91%)0.06 (+21.57%)0.01 (+69.34%)134.30 (-11.64%)114.86 (-17.37%)109.80 (-19.85%)102.80 (-18.41%)11.08 (+14.09%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.06 (n/a)0.06 (n/a)0.05 (n/a)0.00 (n/a)152.00 (n/a)139.00 (n/a)137.00 (n/a)126.00 (n/a)9.72 (n/a)

mem_copy_4_cols_1_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
003b49d — 2025-11-14 19:30:540.07 (+11.97%)0.05 (+3.09%)0.05 (+2.10%)0.03 (-11.35%)0.01 (+41.65%)229.00 (+12.81%)156.00 (-0.51%)143.00 (-2.05%)117.00 (-10.69%)38.21 (+48.71%)
0999a9a — 2025-11-13 19:24:080.06 (n/a)0.05 (n/a)0.05 (n/a)0.04 (n/a)0.01 (n/a)203.00 (n/a)156.80 (n/a)146.00 (n/a)131.00 (n/a)25.69 (n/a)

mem_copy_4_cols_2_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
003b49d — 2025-11-14 19:30:540.06 (+9.02%)0.05 (+5.59%)0.05 (+0.00%)0.04 (+3.63%)0.01 (+25.48%)193.00 (-3.50%)147.40 (-4.66%)147.00 (+0.00%)122.00 (-8.27%)25.24 (+7.48%)
0999a9a — 2025-11-13 19:24:080.06 (n/a)0.05 (n/a)0.05 (n/a)0.04 (n/a)0.01 (n/a)200.00 (n/a)154.60 (n/a)147.00 (n/a)133.00 (n/a)23.48 (n/a)

mem_copy_4_cores_1_chans_2048_tile_512_False

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.08 (+19.25%)0.07 (+17.87%)0.07 (+34.33%)0.05 (+8.75%)0.01 (+56.93%)155.00 (-1.27%)123.82 (-7.46%)111.10 (-20.07%)97.20 (-10.00%)22.66 (+34.72%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.05 (n/a)0.05 (n/a)0.01 (n/a)157.00 (n/a)133.80 (n/a)139.00 (n/a)108.00 (n/a)16.82 (n/a)

mem_copy_4_cores_2_chans_2048_tile_512_False

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.07 (+8.98%)0.06 (+1.46%)0.06 (-8.66%)0.05 (+15.48%)0.01 (-8.57%)169.20 (-7.03%)138.74 (+4.63%)136.40 (+17.59%)109.40 (-1.44%)20.69 (-22.16%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.07 (n/a)0.04 (n/a)0.01 (n/a)182.00 (n/a)132.60 (n/a)116.00 (n/a)111.00 (n/a)26.58 (n/a)

mem_copy_8_cols_1_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
003b49d — 2025-11-14 19:30:540.05 (-13.48%)0.05 (-17.79%)0.05 (-15.33%)0.03 (-25.09%)0.01 (+6.67%)267.00 (+33.50%)176.80 (+24.51%)150.00 (+18.11%)141.00 (+15.57%)47.41 (+61.41%)
0999a9a — 2025-11-13 19:24:080.06 (n/a)0.06 (n/a)0.06 (n/a)0.04 (n/a)0.01 (n/a)200.00 (n/a)142.00 (n/a)127.00 (n/a)122.00 (n/a)29.37 (n/a)

mem_copy_8_cols_2_channels_2048_tile_128

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
003b49d — 2025-11-14 19:30:540.05 (+13.38%)0.05 (+23.85%)0.05 (+37.91%)0.05 (+30.72%)0.00 (-56.85%)166.00 (-23.50%)153.40 (-20.60%)153.00 (-27.49%)142.00 (-11.80%)7.63 (-70.60%)
0999a9a — 2025-11-13 19:24:080.05 (n/a)0.04 (n/a)0.04 (n/a)0.04 (n/a)0.01 (n/a)217.00 (n/a)193.20 (n/a)211.00 (n/a)161.00 (n/a)25.96 (n/a)

mem_copy_8_cores_1_chans_2048_tile_256_False

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.08 (+25.96%)0.07 (+34.89%)0.07 (+27.52%)0.06 (+101.29%)0.01 (-40.01%)145.60 (-46.67%)125.44 (-25.33%)123.80 (-15.78%)107.40 (-14.76%)12.82 (-76.01%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.05 (n/a)0.05 (n/a)0.03 (n/a)0.01 (n/a)273.00 (n/a)168.00 (n/a)147.00 (n/a)126.00 (n/a)53.43 (n/a)

mem_copy_8_cores_2_chans_2048_tile_256_False

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.10 (+65.50%)0.08 (+49.54%)0.08 (+51.00%)0.05 (+14.52%)0.02 (+179.79%)177.20 (-6.24%)115.20 (-24.80%)106.00 (-28.86%)84.30 (-35.15%)32.06 (+64.89%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.05 (n/a)0.05 (n/a)0.04 (n/a)0.01 (n/a)189.00 (n/a)153.20 (n/a)149.00 (n/a)130.00 (n/a)19.45 (n/a)

mha

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.18 (-88.01%)0.18 (-88.02%)0.18 (-88.02%)0.18 (-88.04%)0.00 (-79.84%)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)47465.80 (n/a)47283.32 (n/a)47252.90 (n/a)47177.50 (n/a)98.35 (n/a)
3a053d3 — 2025-11-26 01:12:161.48 (n/a)1.48 (n/a)1.48 (n/a)1.48 (n/a)0.00 (n/a)759.22 (n/a)758.39 (n/a)758.89 (n/a)756.71 (n/a)0.94 (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)

relu_1_cols_1_channels_2048_tile_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.10 (+58.08%)0.08 (+45.37%)0.08 (+47.40%)0.06 (+21.87%)0.01 (+279.44%)133.90 (-11.91%)106.64 (-23.83%)102.70 (-27.16%)83.50 (-32.11%)20.25 (+114.91%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.05 (n/a)0.05 (n/a)0.05 (n/a)0.00 (n/a)152.00 (n/a)140.00 (n/a)141.00 (n/a)123.00 (n/a)9.42 (n/a)

relu_2_cols_1_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.08 (+1.61%)0.07 (+23.29%)0.07 (+22.61%)0.06 (+88.86%)0.01 (-61.99%)137.00 (-43.15%)120.44 (-19.17%)118.20 (-12.44%)108.80 (+5.63%)9.64 (-79.86%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.06 (n/a)0.03 (n/a)0.01 (n/a)241.00 (n/a)149.00 (n/a)135.00 (n/a)103.00 (n/a)47.85 (n/a)

relu_4_cols_1_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.08 (+31.27%)0.07 (+33.99%)0.07 (+19.41%)0.06 (+37.66%)0.01 (+43.88%)138.80 (-22.02%)118.06 (-19.69%)125.00 (-10.07%)96.50 (-18.22%)17.28 (-17.08%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.05 (n/a)0.05 (n/a)0.04 (n/a)0.01 (n/a)178.00 (n/a)147.00 (n/a)139.00 (n/a)118.00 (n/a)20.84 (n/a)

relu_8_cols_1_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.06 (+12.63%)0.05 (+9.94%)0.06 (+19.69%)0.03 (-23.57%)0.01 (+120.60%)278.10 (+40.45%)166.86 (+4.29%)139.90 (-10.32%)131.60 (-4.64%)56.17 (+174.88%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.05 (n/a)0.05 (n/a)0.04 (n/a)0.01 (n/a)198.00 (n/a)160.00 (n/a)156.00 (n/a)138.00 (n/a)20.44 (n/a)

rms_norm_1_cols_1_channels_2048_tile_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.08 (-30.76%)0.06 (-38.78%)0.05 (-41.60%)0.05 (-39.90%)0.01 (-5.75%)175.10 (+19.12%)148.70 (+19.15%)158.10 (+22.56%)101.30 (+3.37%)25.23 (+58.91%)
3a053d3 — 2025-11-26 01:12:160.12 (n/a)0.09 (n/a)0.09 (n/a)0.08 (n/a)0.01 (n/a)147.00 (n/a)124.80 (n/a)129.00 (n/a)98.00 (n/a)15.88 (n/a)

rms_norm_1_cols_2_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.08 (-12.85%)0.07 (-18.66%)0.06 (-14.90%)0.05 (-30.63%)0.01 (+49.02%)160.00 (+3.23%)128.96 (-9.82%)127.00 (-15.89%)100.20 (-17.87%)23.58 (+74.41%)
3a053d3 — 2025-11-26 01:12:160.09 (n/a)0.08 (n/a)0.08 (n/a)0.07 (n/a)0.01 (n/a)155.00 (n/a)143.00 (n/a)151.00 (n/a)122.00 (n/a)13.52 (n/a)

rms_norm_2_cols_1_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.10 (+14.04%)0.07 (+3.19%)0.06 (-23.29%)0.06 (+49.64%)0.02 (-24.27%)144.50 (-52.15%)125.20 (-37.15%)140.00 (-6.67%)79.10 (-37.22%)24.55 (-69.15%)
3a053d3 — 2025-11-26 01:12:160.09 (n/a)0.07 (n/a)0.08 (n/a)0.04 (n/a)0.02 (n/a)302.00 (n/a)199.20 (n/a)150.00 (n/a)126.00 (n/a)79.57 (n/a)

rms_norm_2_cols_2_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.10 (+7.99%)0.08 (+6.22%)0.08 (+10.88%)0.07 (+2.50%)0.01 (+37.70%)121.50 (-30.17%)104.20 (-31.98%)103.90 (-35.47%)86.20 (-33.69%)15.20 (-9.51%)
3a053d3 — 2025-11-26 01:12:160.09 (n/a)0.08 (n/a)0.07 (n/a)0.07 (n/a)0.01 (n/a)174.00 (n/a)153.20 (n/a)161.00 (n/a)130.00 (n/a)16.80 (n/a)

rms_norm_4_cols_1_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.08 (-9.53%)0.06 (-18.76%)0.06 (-17.77%)0.05 (-22.15%)0.01 (+27.26%)164.60 (-8.04%)140.84 (-10.86%)141.90 (-12.94%)106.80 (-20.89%)19.12 (+26.43%)
3a053d3 — 2025-11-26 01:12:160.08 (n/a)0.07 (n/a)0.07 (n/a)0.06 (n/a)0.01 (n/a)179.00 (n/a)158.00 (n/a)163.00 (n/a)135.00 (n/a)15.13 (n/a)

rms_norm_4_cols_2_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.08 (+0.28%)0.06 (-3.85%)0.06 (-10.54%)0.04 (-0.27%)0.02 (+16.76%)182.30 (-28.23%)133.58 (-24.62%)127.20 (-20.00%)99.20 (-28.63%)32.41 (-20.97%)
3a053d3 — 2025-11-26 01:12:160.08 (n/a)0.07 (n/a)0.07 (n/a)0.05 (n/a)0.01 (n/a)254.00 (n/a)177.20 (n/a)159.00 (n/a)139.00 (n/a)41.01 (n/a)

rms_norm_8_cols_1_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.06 (-15.39%)0.05 (-15.32%)0.05 (-20.83%)0.05 (+12.78%)0.01 (-54.72%)172.60 (-36.54%)153.92 (-18.21%)151.00 (-9.58%)131.10 (-15.42%)14.28 (-66.91%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.07 (n/a)0.04 (n/a)0.01 (n/a)272.00 (n/a)188.20 (n/a)167.00 (n/a)155.00 (n/a)43.17 (n/a)

rms_norm_8_cols_2_channels_2048_tile_128

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.09 (+30.70%)0.07 (+1.59%)0.07 (-3.01%)0.05 (-14.05%)0.01 (+214.99%)164.90 (-16.72%)129.10 (-27.23%)124.70 (-26.21%)92.00 (-45.24%)23.85 (+99.39%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.07 (n/a)0.06 (n/a)0.00 (n/a)198.00 (n/a)177.40 (n/a)169.00 (n/a)168.00 (n/a)11.96 (n/a)

rope_1_cols_2_channels_4096_tile_4096_0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.23 (+55.00%)0.19 (+41.67%)0.17 (+18.90%)0.15 (+58.65%)0.03 (+70.43%)161.80 (-32.30%)133.82 (-23.97%)144.50 (-9.69%)106.70 (-30.71%)22.45 (-29.76%)
3a053d3 — 2025-11-26 01:12:160.15 (n/a)0.13 (n/a)0.14 (n/a)0.10 (n/a)0.02 (n/a)239.00 (n/a)176.00 (n/a)160.00 (n/a)154.00 (n/a)31.96 (n/a)

rope_2_cols_2_channels_4096_tile_2048_0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.24 (+30.64%)0.19 (+40.37%)0.19 (+32.25%)0.13 (+60.02%)0.03 (+7.72%)182.50 (-32.90%)132.70 (-25.70%)130.70 (-18.82%)102.70 (-17.84%)27.15 (-45.73%)
3a053d3 — 2025-11-26 01:12:160.18 (n/a)0.14 (n/a)0.14 (n/a)0.08 (n/a)0.03 (n/a)272.00 (n/a)178.60 (n/a)161.00 (n/a)125.00 (n/a)50.03 (n/a)

rope_4_cols_2_channels_4096_tile_1024_0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.22 (+12.46%)0.20 (+24.60%)0.21 (+37.42%)0.18 (+22.38%)0.02 (+0.11%)138.60 (-12.28%)121.76 (-14.13%)116.40 (-21.88%)109.80 (-4.52%)12.19 (-21.90%)
3a053d3 — 2025-11-26 01:12:160.20 (n/a)0.16 (n/a)0.15 (n/a)0.14 (n/a)0.02 (n/a)158.00 (n/a)141.80 (n/a)149.00 (n/a)115.00 (n/a)15.61 (n/a)

rope_8_cols_2_channels_4096_tile_512_0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.25 (+49.82%)0.20 (+35.46%)0.21 (+43.04%)0.13 (-5.25%)0.04 (+322.51%)187.00 (+13.33%)127.22 (-16.96%)114.80 (-24.97%)98.20 (-28.32%)31.60 (+232.01%)
3a053d3 — 2025-11-26 01:12:160.17 (n/a)0.15 (n/a)0.15 (n/a)0.14 (n/a)0.01 (n/a)165.00 (n/a)153.20 (n/a)153.00 (n/a)137.00 (n/a)9.52 (n/a)

sigmoid_1_cols_1_channels_2048_tile_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.09 (+55.44%)0.07 (+36.05%)0.06 (+16.23%)0.05 (+10.53%)0.02 (+336.19%)157.40 (-2.84%)124.68 (-17.54%)138.60 (-7.60%)92.60 (-30.90%)26.37 (+159.85%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.05 (n/a)0.05 (n/a)0.05 (n/a)0.00 (n/a)162.00 (n/a)151.20 (n/a)150.00 (n/a)134.00 (n/a)10.15 (n/a)

sigmoid_2_cols_1_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.06 (+9.17%)0.06 (+17.99%)0.06 (+11.28%)0.05 (+70.81%)0.00 (-68.96%)158.40 (-37.14%)144.76 (-13.21%)142.80 (-3.51%)136.70 (-1.65%)7.55 (-82.43%)
3a053d3 — 2025-11-26 01:12:160.05 (n/a)0.05 (n/a)0.05 (n/a)0.03 (n/a)0.01 (n/a)252.00 (n/a)166.80 (n/a)148.00 (n/a)139.00 (n/a)42.99 (n/a)

sigmoid_4_cols_1_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.07 (+24.32%)0.06 (+39.33%)0.06 (+35.90%)0.05 (+79.80%)0.00 (-45.87%)155.30 (-40.27%)139.34 (-26.20%)139.80 (-21.02%)121.80 (-13.62%)11.34 (-73.97%)
3a053d3 — 2025-11-26 01:12:160.05 (n/a)0.04 (n/a)0.04 (n/a)0.03 (n/a)0.01 (n/a)260.00 (n/a)188.80 (n/a)177.00 (n/a)141.00 (n/a)43.55 (n/a)

sigmoid_8_cols_1_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.10 (+78.72%)0.07 (+32.68%)0.06 (+19.76%)0.04 (-18.59%)0.02 (+766.56%)209.70 (+31.89%)135.42 (-10.67%)138.10 (-10.32%)84.70 (-39.93%)43.87 (+526.45%)
3a053d3 — 2025-11-26 01:12:160.05 (n/a)0.05 (n/a)0.05 (n/a)0.05 (n/a)0.00 (n/a)159.00 (n/a)151.60 (n/a)154.00 (n/a)141.00 (n/a)7.00 (n/a)

silu_1_cols_1_channels_2048_tile_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.07 (+2.68%)0.06 (+4.12%)0.06 (-4.91%)0.05 (+21.81%)0.01 (-28.20%)149.00 (-11.83%)134.02 (+1.68%)138.90 (+12.93%)111.90 (+4.58%)12.64 (-40.32%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.06 (n/a)0.05 (n/a)0.01 (n/a)169.00 (n/a)131.80 (n/a)123.00 (n/a)107.00 (n/a)21.19 (n/a)

silu_2_cols_1_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.07 (+5.21%)0.06 (-1.90%)0.06 (-7.88%)0.05 (-3.88%)0.01 (+22.14%)165.30 (+11.69%)141.98 (+10.06%)141.00 (+16.53%)115.30 (+2.04%)18.77 (+29.06%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.06 (n/a)0.05 (n/a)0.01 (n/a)148.00 (n/a)129.00 (n/a)121.00 (n/a)113.00 (n/a)14.55 (n/a)

silu_4_cols_1_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.06 (-14.39%)0.06 (-2.62%)0.06 (+5.44%)0.05 (+6.10%)0.00 (-71.04%)150.80 (+1.21%)142.38 (+9.02%)139.50 (+1.82%)138.00 (+25.45%)4.94 (-65.61%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.06 (n/a)0.05 (n/a)0.01 (n/a)149.00 (n/a)130.60 (n/a)137.00 (n/a)110.00 (n/a)14.37 (n/a)

silu_8_cols_1_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.08 (+26.46%)0.06 (+16.93%)0.06 (+12.62%)0.05 (+33.44%)0.01 (+11.32%)160.10 (-19.55%)136.00 (-8.85%)136.30 (-4.69%)106.10 (-15.12%)19.59 (-28.10%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.05 (n/a)0.05 (n/a)0.04 (n/a)0.01 (n/a)199.00 (n/a)149.20 (n/a)143.00 (n/a)125.00 (n/a)27.25 (n/a)

softmax_1_cols_2_channels_4096_tile_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.18 (-15.32%)0.14 (-11.68%)0.15 (+0.69%)0.10 (-26.88%)0.03 (+7.89%)162.50 (-2.11%)118.06 (-17.44%)107.40 (-28.87%)93.00 (-15.45%)24.69 (+29.68%)
3a053d3 — 2025-11-26 01:12:160.21 (n/a)0.16 (n/a)0.15 (n/a)0.14 (n/a)0.02 (n/a)166.00 (n/a)143.00 (n/a)151.00 (n/a)110.00 (n/a)19.04 (n/a)

softmax_2_cols_2_channels_4096_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.18 (-2.65%)0.15 (+5.33%)0.16 (+17.89%)0.12 (+6.24%)0.02 (-24.55%)141.50 (-32.62%)109.42 (-33.12%)105.70 (-39.25%)89.00 (-26.45%)17.35 (-45.24%)
3a053d3 — 2025-11-26 01:12:160.19 (n/a)0.15 (n/a)0.13 (n/a)0.11 (n/a)0.03 (n/a)210.00 (n/a)163.60 (n/a)174.00 (n/a)121.00 (n/a)31.68 (n/a)

softmax_2_cols_2_channels_4096_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.18 (+8.85%)0.14 (-2.67%)0.13 (-2.77%)0.10 (-20.26%)0.03 (+100.76%)163.40 (-10.22%)127.18 (-23.29%)125.90 (-26.37%)88.80 (-34.22%)28.68 (+69.79%)
3a053d3 — 2025-11-26 01:12:160.17 (n/a)0.14 (n/a)0.13 (n/a)0.13 (n/a)0.02 (n/a)182.00 (n/a)165.80 (n/a)171.00 (n/a)135.00 (n/a)16.89 (n/a)

swiglu

No metrics available.

swiglu_decode_1x2048x2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.00 (n/a)0.00 (n/a)0.00 (n/a)0.00 (n/a)0.00 (n/a)5177.70 (n/a)3893.34 (n/a)3408.80 (n/a)3397.50 (n/a)694.48 (n/a)

tanh_1_cols_1_channels_2048_tile_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.08 (-2.00%)0.06 (+11.08%)0.06 (-2.80%)0.06 (+86.41%)0.01 (-58.92%)139.40 (-42.40%)130.60 (-11.28%)138.10 (+10.48%)108.50 (+9.60%)11.70 (-76.89%)
3a053d3 — 2025-11-26 01:12:160.08 (n/a)0.06 (n/a)0.06 (n/a)0.03 (n/a)0.02 (n/a)242.00 (n/a)147.20 (n/a)125.00 (n/a)99.00 (n/a)50.64 (n/a)

tanh_2_cols_1_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.08 (+10.99%)0.07 (+11.83%)0.06 (+1.59%)0.06 (+27.96%)0.01 (-11.87%)140.10 (-16.11%)127.02 (-5.35%)132.10 (+5.68%)98.70 (-3.24%)15.35 (-34.88%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.06 (n/a)0.05 (n/a)0.01 (n/a)167.00 (n/a)134.20 (n/a)125.00 (n/a)102.00 (n/a)23.57 (n/a)

tanh_4_cols_1_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.10 (+65.47%)0.09 (+58.70%)0.08 (+42.31%)0.08 (+89.56%)0.01 (+23.26%)102.50 (-43.37%)94.90 (-32.98%)100.40 (-24.51%)80.50 (-35.08%)8.71 (-58.00%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.05 (n/a)0.06 (n/a)0.04 (n/a)0.01 (n/a)181.00 (n/a)141.60 (n/a)133.00 (n/a)124.00 (n/a)20.74 (n/a)

tanh_8_cols_1_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.08 (+37.86%)0.06 (+32.98%)0.06 (+33.22%)0.05 (+25.26%)0.01 (+79.47%)164.60 (-14.27%)137.18 (-18.15%)137.80 (-19.42%)103.60 (-22.11%)22.32 (+15.74%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.05 (n/a)0.04 (n/a)0.04 (n/a)0.01 (n/a)192.00 (n/a)167.60 (n/a)171.00 (n/a)133.00 (n/a)19.28 (n/a)

transpose_2048_M_64_N_1_cols_1_channels_64_m_64_n_8_s

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:574.63 (+23.29%)3.91 (+21.80%)3.77 (+16.71%)3.45 (+36.55%)0.43 (+8.94%)151.80 (-21.35%)135.60 (-12.29%)138.90 (-8.01%)113.20 (-12.92%)14.13 (-32.18%)
3a053d3 — 2025-11-26 01:12:163.76 (n/a)3.21 (n/a)3.23 (n/a)2.53 (n/a)0.39 (n/a)193.00 (n/a)154.60 (n/a)151.00 (n/a)130.00 (n/a)20.83 (n/a)

transpose_2048_M_64_N_1_cols_2_channels_64_m_64_n_8_s

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:574.79 (+33.52%)3.92 (+43.85%)3.97 (+45.43%)3.29 (+76.68%)0.51 (-5.94%)159.20 (-39.24%)136.14 (-27.35%)132.20 (-26.15%)109.40 (-19.56%)17.04 (-58.48%)
3a053d3 — 2025-11-26 01:12:163.59 (n/a)2.72 (n/a)2.73 (n/a)1.86 (n/a)0.55 (n/a)262.00 (n/a)187.40 (n/a)179.00 (n/a)136.00 (n/a)41.03 (n/a)

weighted_rms_norm_1_cols_2_channels_2048_weights_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.08 (-12.00%)0.07 (-6.04%)0.07 (-9.67%)0.07 (+6.25%)0.00 (-54.59%)123.30 (-32.62%)116.22 (-24.63%)119.70 (-20.73%)108.20 (-18.65%)6.37 (-65.35%)
3a053d3 — 2025-11-26 01:12:160.09 (n/a)0.08 (n/a)0.08 (n/a)0.06 (n/a)0.01 (n/a)183.00 (n/a)154.20 (n/a)151.00 (n/a)133.00 (n/a)18.39 (n/a)

weighted_rms_norm_2_cols_2_channels_2048_weights_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.07 (-27.94%)0.07 (-15.55%)0.06 (-15.67%)0.06 (-15.42%)0.01 (-49.09%)142.20 (-15.36%)126.94 (-16.38%)134.10 (-15.13%)110.30 (-0.63%)12.62 (-39.94%)
3a053d3 — 2025-11-26 01:12:160.10 (n/a)0.08 (n/a)0.07 (n/a)0.07 (n/a)0.01 (n/a)168.00 (n/a)151.80 (n/a)158.00 (n/a)111.00 (n/a)21.01 (n/a)

weighted_rms_norm_4_cols_2_channels_2048_weights_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.08 (-7.92%)0.06 (-21.12%)0.06 (-33.50%)0.04 (-34.80%)0.02 (+67.91%)183.40 (+9.82%)136.56 (-5.17%)148.50 (+7.61%)96.40 (-22.26%)31.98 (+91.79%)
3a053d3 — 2025-11-26 01:12:160.09 (n/a)0.08 (n/a)0.08 (n/a)0.07 (n/a)0.01 (n/a)167.00 (n/a)144.00 (n/a)138.00 (n/a)124.00 (n/a)16.67 (n/a)

weighted_rms_norm_8_cols_2_channels_2048_weights_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
6b841b9 — 2025-12-03 21:29:570.09 (+8.73%)0.06 (-7.30%)0.06 (-14.91%)0.05 (+9.28%)0.01 (+1.76%)165.70 (-34.51%)132.80 (-23.41%)136.30 (-15.86%)93.50 (-34.15%)23.34 (-42.68%)
3a053d3 — 2025-11-26 01:12:160.08 (n/a)0.07 (n/a)0.07 (n/a)0.05 (n/a)0.01 (n/a)253.00 (n/a)173.40 (n/a)162.00 (n/a)142.00 (n/a)40.72 (n/a)

@github-actions
Copy link
Contributor

github-actions bot commented Dec 3, 2025

📊 Test Results for Test Example Applications

98350dc (2025_12_03_22_19_28)

IRONCLAD

Tested on 2025_12_03_22_19_28 at commit 98350dc.

Test Checks TTFT (mean)TPS (mean)
llama_3.2_1b ❌ 0/5 n/a n/a
📈 Trends (vs main branch) for Test Example Applications

98350dc (2025_12_03_22_19_28)

IRONCLAD Trends

llama_3.2_1b

Commit/Date Num Tokens (max)Num Tokens (mean)Num Tokens (median)Num Tokens (min)Num Tokens (stddev)TPS (max)TPS (mean)TPS (median)TPS (min)TPS (stddev)TTFT (max)TTFT (mean)TTFT (median)TTFT (min)TTFT (stddev)Total (max)Total (mean)Total (median)Total (min)Total (stddev)
98350dc — 2025-12-03 22:19:24n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)
3a053d3 — 2025-11-26 01:24:4440.00 (n/a)40.00 (n/a)40.00 (n/a)40.00 (n/a)0.00 (n/a)4.73 (n/a)4.67 (n/a)4.68 (n/a)4.61 (n/a)0.04 (n/a)4.20 (n/a)3.92 (n/a)4.04 (n/a)3.57 (n/a)0.28 (n/a)12.54 (n/a)12.27 (n/a)12.50 (n/a)11.83 (n/a)0.32 (n/a)

@github-actions
Copy link
Contributor

github-actions bot commented Dec 3, 2025

📊 Test Results for Test Example Applications

a02e460 (2025_12_03_22_40_30)

IRONCLAD

Tested on 2025_12_03_22_40_30 at commit a02e460.

Test Checks TTFT (mean)TPS (mean)
llama_3.2_1b ❌ 0/5 n/a n/a
📈 Trends (vs main branch) for Test Example Applications

a02e460 (2025_12_03_22_40_30)

IRONCLAD Trends

llama_3.2_1b

Commit/Date Num Tokens (max)Num Tokens (mean)Num Tokens (median)Num Tokens (min)Num Tokens (stddev)TPS (max)TPS (mean)TPS (median)TPS (min)TPS (stddev)TTFT (max)TTFT (mean)TTFT (median)TTFT (min)TTFT (stddev)Total (max)Total (mean)Total (median)Total (min)Total (stddev)
a02e460 — 2025-12-03 22:38:05n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)
3a053d3 — 2025-11-26 01:24:4440.00 (n/a)40.00 (n/a)40.00 (n/a)40.00 (n/a)0.00 (n/a)4.73 (n/a)4.67 (n/a)4.68 (n/a)4.61 (n/a)0.04 (n/a)4.20 (n/a)3.92 (n/a)4.04 (n/a)3.57 (n/a)0.28 (n/a)12.54 (n/a)12.27 (n/a)12.50 (n/a)11.83 (n/a)0.32 (n/a)

@github-actions
Copy link
Contributor

github-actions bot commented Dec 3, 2025

📊 Test Results for Test Example Applications

79c1558 (2025_12_03_22_52_18)

IRONCLAD

Tested on 2025_12_03_22_52_18 at commit 79c1558.

Test Checks TTFT (mean)TPS (mean)
llama_3.2_1b 🟠 4/5 4.40 4.60
📈 Trends (vs main branch) for Test Example Applications

79c1558 (2025_12_03_22_52_18)

IRONCLAD Trends

llama_3.2_1b

Commit/Date Num Tokens (max)Num Tokens (mean)Num Tokens (median)Num Tokens (min)Num Tokens (stddev)TPS (max)TPS (mean)TPS (median)TPS (min)TPS (stddev)TTFT (max)TTFT (mean)TTFT (median)TTFT (min)TTFT (stddev)Total (max)Total (mean)Total (median)Total (min)Total (stddev)
79c1558 — 2025-12-03 22:47:4240.00 (+0.00%)40.00 (+0.00%)40.00 (+0.00%)40.00 (+0.00%)0.00 (n/a)4.71 (-0.42%)4.60 (-1.58%)4.62 (-1.28%)4.45 (-3.47%)0.09 (+146.56%)4.45 (+5.78%)4.40 (+12.07%)4.38 (+8.38%)4.38 (+22.45%)0.03 (-89.53%)13.14 (+4.77%)12.88 (+4.97%)12.86 (+2.87%)12.66 (+6.94%)0.18 (-44.00%)
3a053d3 — 2025-11-26 01:24:4440.00 (n/a)40.00 (n/a)40.00 (n/a)40.00 (n/a)0.00 (n/a)4.73 (n/a)4.67 (n/a)4.68 (n/a)4.61 (n/a)0.04 (n/a)4.20 (n/a)3.92 (n/a)4.04 (n/a)3.57 (n/a)0.28 (n/a)12.54 (n/a)12.27 (n/a)12.50 (n/a)11.83 (n/a)0.32 (n/a)

@github-actions
Copy link
Contributor

github-actions bot commented Dec 3, 2025

📊 Test Results for Small Benchmark/Test Suite

79c1558 (2025_12_03_23_09_12)

IRONCLAD

Tested on 2025_12_03_23_09_12 at commit 79c1558.

Test Checks Latency (mean)Bandwidth (mean)GFLOP/s (mean)
axpy_1_cols_2_channels_2048_tile_2048_3.0 ✅ 5/5 119.12 0.11 n/a
axpy_2_cols_2_channels_2048_tile_1024_3.0 ✅ 5/5 134.56 0.09 n/a
axpy_4_cols_2_channels_2048_tile_512_3.0 ✅ 5/5 121.76 0.11 n/a
axpy_8_cols_2_channels_2048_tile_256_3.0 ✅ 5/5 158.98 0.08 n/a
dequant_1_cols_1_channels_2048_tile_2048 ✅ 5/5 108.74 0.05 n/a
dequant_1_cols_2_channels_2048_tile_1024 ✅ 5/5 109.00 0.05 n/a
dequant_2_cols_1_channels_2048_tile_1024 ✅ 5/5 122.20 0.04 n/a
dequant_2_cols_2_channels_2048_tile_512 ✅ 5/5 126.52 0.04 n/a
dequant_4_cols_1_channels_2048_tile_512 ✅ 5/5 111.08 0.05 n/a
dequant_4_cols_2_channels_2048_tile_256 ✅ 5/5 134.52 0.04 n/a
dequant_8_cols_1_channels_2048_tile_256 ✅ 5/5 146.84 0.04 n/a
dequant_8_cols_2_channels_2048_tile_128 ✅ 5/5 150.28 0.04 n/a
eltwise_add_1_cols_2_channels_2048_tile_2048 ✅ 5/5 120.90 0.10 n/a
eltwise_add_2_cols_2_channels_2048_tile_1024 ✅ 5/5 126.72 0.10 n/a
eltwise_add_4_cols_2_channels_2048_tile_512 ✅ 5/5 136.20 0.09 n/a
eltwise_add_8_cols_2_channels_2048_tile_256 ✅ 5/5 148.30 0.09 n/a
eltwise_mul_1_cols_2_channels_2048_tile_2048 ✅ 5/5 122.88 0.10 n/a
eltwise_mul_2_cols_2_channels_2048_tile_1024 ✅ 5/5 137.26 0.09 n/a
eltwise_mul_4_cols_2_channels_2048_tile_512 ✅ 5/5 134.74 0.09 n/a
eltwise_mul_8_cols_2_channels_2048_tile_256 ✅ 5/5 140.56 0.09 n/a
gelu_1_cols_1_channels_2048_tile_2048 ✅ 5/5 123.04 0.07 n/a
gelu_1_cols_2_channels_2048_tile_1024 ✅ 5/5 127.42 0.06 n/a
gelu_2_cols_1_channels_2048_tile_1024 ✅ 5/5 121.30 0.07 n/a
gelu_2_cols_2_channels_2048_tile_512 ✅ 5/5 134.44 0.06 n/a
gelu_4_cols_1_channels_2048_tile_512 ✅ 5/5 119.58 0.07 n/a
gelu_4_cols_2_channels_2048_tile_256 ✅ 5/5 148.82 0.06 n/a
gelu_8_cols_1_channels_2048_tile_256 ✅ 5/5 157.36 0.05 n/a
gelu_8_cols_2_channels_2048_tile_128 ✅ 5/5 151.26 0.06 n/a
gemm_2048x2048x2048_64x64x64_2_cols_0_bcolmaj_0_ccolmaj_0 ✅ 5/5 24525.50 1.03 n/a
gemm_2048x2048x2048_64x64x64_2_cols_0_bcolmaj_1_ccolmaj_0 ✅ 5/5 25901.84 0.97 n/a
gemm_2048x2048x2048_64x64x64_2_cols_1_bcolmaj_0_ccolmaj_0 ✅ 5/5 27429.50 0.92 n/a
layer_norm_1_cols_1_channels_2048_tile_2048 ✅ 5/5 142.96 0.06 n/a
layer_norm_1_cols_2_channels_2048_tile_1024 ✅ 5/5 140.34 0.06 n/a
layer_norm_2_cols_1_channels_2048_tile_1024 ✅ 5/5 179.52 0.05 n/a
layer_norm_2_cols_2_channels_2048_tile_512 ✅ 5/5 124.44 0.07 n/a
layer_norm_4_cols_1_channels_2048_tile_512 ✅ 5/5 135.04 0.06 n/a
layer_norm_4_cols_2_channels_2048_tile_256 ✅ 5/5 119.06 0.07 n/a
layer_norm_8_cols_1_channels_2048_tile_256 ✅ 5/5 129.34 0.06 n/a
layer_norm_8_cols_2_channels_2048_tile_128 ✅ 5/5 137.14 0.06 n/a
matrix_vector_mul_128x128_32_1col ✅ 5/5 n/a 0.24 n/a
matrix_vector_mul_2048x8192_1_1col ✅ 5/5 n/a 13.16 n/a
matrix_vector_mul_2048x8192_1_2col ✅ 5/5 n/a 24.39 n/a
matrix_vector_mul_2048x8192_1_4col ✅ 5/5 n/a 39.97 n/a
matrix_vector_mul_2048x8192_1_8col ✅ 5/5 n/a 44.99 n/a
matrix_vector_mul_8192x2048_4_1col ✅ 5/5 n/a 13.12 n/a
matrix_vector_mul_8192x2048_4_2col ✅ 5/5 n/a 22.74 n/a
matrix_vector_mul_8192x2048_4_4col ✅ 5/5 n/a 40.56 n/a
matrix_vector_mul_8192x2048_4_8col ✅ 5/5 n/a 39.79 n/a
mem_copy_16_cores_2_chans_2048_tile_128_False ✅ 5/5 143.14 0.06 n/a
mem_copy_1_cores_1_chans_2048_tile_2048_False ✅ 5/5 116.78 0.07 n/a
mem_copy_2_cores_1_chans_2048_tile_1024_False ✅ 5/5 115.72 0.07 n/a
mem_copy_2_cores_2_chans_2048_tile_1024_False ✅ 5/5 131.44 0.06 n/a
mem_copy_4_cores_1_chans_2048_tile_512_False ✅ 5/5 134.90 0.06 n/a
mem_copy_4_cores_2_chans_2048_tile_512_False ✅ 5/5 102.72 0.08 n/a
mem_copy_8_cores_1_chans_2048_tile_256_False ✅ 5/5 130.98 0.06 n/a
mem_copy_8_cores_2_chans_2048_tile_256_False ✅ 5/5 139.10 0.06 n/a
mha ✅ 5/5 47185.94 0.18 n/a
relu_1_cols_1_channels_2048_tile_2048 ✅ 5/5 118.22 0.07 n/a
relu_2_cols_1_channels_2048_tile_1024 ✅ 5/5 119.48 0.07 n/a
relu_4_cols_1_channels_2048_tile_512 ✅ 5/5 110.92 0.08 n/a
relu_8_cols_1_channels_2048_tile_256 ✅ 5/5 137.08 0.06 n/a
rms_norm_1_cols_1_channels_2048_tile_2048 ✅ 5/5 133.00 0.06 n/a
rms_norm_1_cols_2_channels_2048_tile_1024 ✅ 5/5 151.08 0.05 n/a
rms_norm_2_cols_1_channels_2048_tile_1024 ✅ 5/5 125.80 0.07 n/a
rms_norm_2_cols_2_channels_2048_tile_512 ✅ 5/5 126.46 0.07 n/a
rms_norm_4_cols_1_channels_2048_tile_512 ✅ 5/5 128.26 0.07 n/a
rms_norm_4_cols_2_channels_2048_tile_256 ✅ 5/5 152.64 0.06 n/a
rms_norm_8_cols_1_channels_2048_tile_256 ✅ 5/5 138.18 0.06 n/a
rms_norm_8_cols_2_channels_2048_tile_128 ✅ 5/5 134.28 0.06 n/a
rope_1_cols_2_channels_4096_tile_4096_0 ✅ 5/5 123.80 0.20 n/a
rope_2_cols_2_channels_4096_tile_2048_0 ✅ 5/5 133.26 0.19 n/a
rope_4_cols_2_channels_4096_tile_1024_0 ✅ 5/5 157.22 0.16 n/a
rope_8_cols_2_channels_4096_tile_512_0 ✅ 5/5 151.58 0.16 n/a
sigmoid_1_cols_1_channels_2048_tile_2048 ✅ 5/5 137.08 0.06 n/a
sigmoid_2_cols_1_channels_2048_tile_1024 ✅ 5/5 129.56 0.06 n/a
sigmoid_4_cols_1_channels_2048_tile_512 ✅ 5/5 126.28 0.07 n/a
sigmoid_8_cols_1_channels_2048_tile_256 ✅ 5/5 109.88 0.08 n/a
silu_1_cols_1_channels_2048_tile_2048 ✅ 5/5 114.88 0.07 n/a
silu_2_cols_1_channels_2048_tile_1024 ✅ 5/5 128.74 0.06 n/a
silu_4_cols_1_channels_2048_tile_512 ✅ 5/5 123.02 0.07 n/a
silu_8_cols_1_channels_2048_tile_256 ✅ 5/5 125.04 0.07 n/a
softmax_1_cols_2_channels_4096_tile_2048 ✅ 5/5 128.74 0.13 n/a
softmax_2_cols_2_channels_4096_tile_1024 ✅ 5/5 125.22 0.14 n/a
softmax_2_cols_2_channels_4096_tile_512 ✅ 5/5 109.52 0.15 n/a
swiglu_decode_1x2048x2048 ✅ 5/5 4880.18 0.00 n/a
tanh_1_cols_1_channels_2048_tile_2048 ✅ 5/5 112.74 0.08 n/a
tanh_2_cols_1_channels_2048_tile_1024 ✅ 5/5 118.38 0.07 n/a
tanh_4_cols_1_channels_2048_tile_512 ✅ 5/5 120.48 0.07 n/a
tanh_8_cols_1_channels_2048_tile_256 ✅ 5/5 154.48 0.05 n/a
transpose_2048_M_64_N_1_cols_1_channels_64_m_64_n_8_s ✅ 5/5 131.66 3.99 n/a
transpose_2048_M_64_N_1_cols_2_channels_64_m_64_n_8_s ✅ 5/5 137.78 3.82 n/a
weighted_rms_norm_1_cols_2_channels_2048_weights_2048 ✅ 5/5 129.56 0.07 n/a
weighted_rms_norm_2_cols_2_channels_2048_weights_1024 ✅ 5/5 124.48 0.07 n/a
weighted_rms_norm_4_cols_2_channels_2048_weights_512 ✅ 5/5 119.46 0.07 n/a
weighted_rms_norm_8_cols_2_channels_2048_weights_256 ✅ 5/5 133.36 0.06 n/a
📈 Trends (vs main branch) for Small Benchmark/Test Suite

79c1558 (2025_12_03_23_09_12)

IRONCLAD Trends

axpy_1_cols_2_channels_2048_tile_2048_3.0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.13 (+60.27%)0.11 (+56.95%)0.11 (+61.09%)0.07 (+37.15%)0.02 (+101.03%)174.60 (-21.70%)119.12 (-29.93%)112.60 (-33.37%)92.50 (-32.97%)29.93 (-1.45%)
3a053d3 — 2025-11-26 01:12:160.08 (n/a)0.07 (n/a)0.07 (n/a)0.05 (n/a)0.01 (n/a)223.00 (n/a)170.00 (n/a)169.00 (n/a)138.00 (n/a)30.37 (n/a)

axpy_2_cols_2_channels_2048_tile_1024_3.0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.11 (-6.36%)0.09 (+7.56%)0.09 (+11.96%)0.08 (+18.88%)0.01 (-34.12%)155.40 (-9.65%)134.56 (-2.35%)135.20 (-4.11%)108.90 (+14.63%)16.46 (-33.53%)
3a053d3 — 2025-11-26 01:12:160.12 (n/a)0.09 (n/a)0.08 (n/a)0.07 (n/a)0.02 (n/a)172.00 (n/a)137.80 (n/a)141.00 (n/a)95.00 (n/a)24.77 (n/a)

axpy_4_cols_2_channels_2048_tile_512_3.0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.15 (+56.92%)0.11 (+31.95%)0.09 (+16.34%)0.07 (-0.92%)0.03 (+204.88%)174.50 (+8.39%)121.76 (-13.40%)131.10 (-7.68%)82.80 (-31.57%)33.45 (+103.65%)
3a053d3 — 2025-11-26 01:12:160.09 (n/a)0.08 (n/a)0.08 (n/a)0.07 (n/a)0.01 (n/a)161.00 (n/a)140.60 (n/a)142.00 (n/a)121.00 (n/a)16.43 (n/a)

axpy_8_cols_2_channels_2048_tile_256_3.0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.09 (+28.20%)0.08 (+31.95%)0.07 (+22.15%)0.07 (+47.09%)0.01 (+14.21%)178.10 (-27.01%)158.98 (-19.05%)167.90 (-12.09%)139.90 (-16.23%)15.86 (-38.17%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.06 (n/a)0.05 (n/a)0.01 (n/a)244.00 (n/a)196.40 (n/a)191.00 (n/a)167.00 (n/a)25.66 (n/a)

dequant_1_cols_1_channels_2048_tile_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.06 (+33.36%)0.05 (+41.08%)0.05 (+27.00%)0.04 (+86.53%)0.01 (-23.55%)131.20 (-42.46%)108.74 (-27.60%)104.80 (-15.48%)91.80 (-19.47%)14.00 (-67.03%)
3a053d3 — 2025-11-26 01:12:160.04 (n/a)0.03 (n/a)0.04 (n/a)0.02 (n/a)0.01 (n/a)228.00 (n/a)150.20 (n/a)124.00 (n/a)114.00 (n/a)42.45 (n/a)

dequant_1_cols_2_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.07 (+69.46%)0.05 (+40.07%)0.06 (+40.70%)0.03 (+1.64%)0.01 (+267.76%)164.80 (+5.64%)109.00 (-17.67%)93.90 (-23.66%)76.00 (-36.67%)33.10 (+131.59%)
3a053d3 — 2025-11-26 01:12:160.04 (n/a)0.04 (n/a)0.04 (n/a)0.03 (n/a)0.00 (n/a)156.00 (n/a)132.40 (n/a)123.00 (n/a)120.00 (n/a)14.29 (n/a)

dequant_2_cols_1_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.06 (+80.16%)0.04 (+47.39%)0.04 (+32.78%)0.04 (+44.27%)0.01 (+225.17%)136.20 (-25.57%)122.20 (-25.40%)131.80 (-19.14%)84.60 (-40.42%)19.22 (+31.37%)
3a053d3 — 2025-11-26 01:12:160.03 (n/a)0.03 (n/a)0.03 (n/a)0.03 (n/a)0.00 (n/a)183.00 (n/a)163.80 (n/a)163.00 (n/a)142.00 (n/a)14.63 (n/a)

dequant_2_cols_2_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.05 (+36.20%)0.04 (+24.58%)0.04 (+20.19%)0.04 (+21.66%)0.01 (+74.29%)148.30 (-11.73%)126.52 (-12.98%)126.90 (-10.63%)97.80 (-21.13%)17.63 (+11.63%)
3a053d3 — 2025-11-26 01:12:160.04 (n/a)0.03 (n/a)0.03 (n/a)0.03 (n/a)0.00 (n/a)168.00 (n/a)145.40 (n/a)142.00 (n/a)124.00 (n/a)15.79 (n/a)

dequant_4_cols_1_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.06 (+34.35%)0.05 (+49.56%)0.05 (+66.23%)0.04 (+37.10%)0.01 (+25.24%)146.50 (-21.66%)111.08 (-28.52%)103.30 (-35.44%)83.10 (-20.10%)21.71 (-23.04%)
3a053d3 — 2025-11-26 01:12:160.05 (n/a)0.03 (n/a)0.03 (n/a)0.03 (n/a)0.01 (n/a)187.00 (n/a)155.40 (n/a)160.00 (n/a)104.00 (n/a)28.20 (n/a)

dequant_4_cols_2_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.05 (+32.47%)0.04 (+19.44%)0.04 (+21.59%)0.03 (-4.62%)0.01 (+260.23%)173.40 (+12.60%)134.52 (-7.36%)129.80 (-11.70%)105.40 (-18.92%)24.91 (+209.97%)
3a053d3 — 2025-11-26 01:12:160.04 (n/a)0.03 (n/a)0.03 (n/a)0.03 (n/a)0.00 (n/a)154.00 (n/a)145.20 (n/a)147.00 (n/a)130.00 (n/a)8.03 (n/a)

dequant_8_cols_1_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.04 (-13.61%)0.04 (+5.56%)0.04 (-2.72%)0.03 (+20.28%)0.00 (-52.79%)183.00 (-10.73%)146.84 (-3.90%)144.60 (+10.38%)125.50 (+24.26%)19.40 (-52.55%)
3a053d3 — 2025-11-26 01:12:160.05 (n/a)0.03 (n/a)0.04 (n/a)0.02 (n/a)0.01 (n/a)205.00 (n/a)152.80 (n/a)131.00 (n/a)101.00 (n/a)40.88 (n/a)

dequant_8_cols_2_channels_2048_tile_128

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.05 (+68.12%)0.04 (+24.82%)0.04 (+15.61%)0.03 (+5.60%)0.01 (+377.82%)188.10 (+1.68%)150.28 (-10.23%)149.50 (-7.14%)100.90 (-36.14%)30.09 (+186.06%)
3a053d3 — 2025-11-26 01:12:160.03 (n/a)0.03 (n/a)0.03 (n/a)0.03 (n/a)0.00 (n/a)185.00 (n/a)167.40 (n/a)161.00 (n/a)158.00 (n/a)10.52 (n/a)

eltwise_add_1_cols_2_channels_2048_tile_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.12 (+8.41%)0.10 (+30.00%)0.10 (+28.85%)0.09 (+67.74%)0.01 (-38.77%)138.90 (-35.99%)120.90 (-20.98%)126.70 (-16.64%)104.00 (-0.95%)13.09 (-64.70%)
3a053d3 — 2025-11-26 01:12:160.11 (n/a)0.08 (n/a)0.08 (n/a)0.05 (n/a)0.02 (n/a)217.00 (n/a)153.00 (n/a)152.00 (n/a)105.00 (n/a)37.09 (n/a)

eltwise_add_2_cols_2_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.12 (+19.05%)0.10 (+13.44%)0.10 (+13.83%)0.08 (+20.99%)0.01 (+16.63%)152.60 (-11.28%)126.72 (-5.43%)118.90 (-5.63%)102.80 (-9.82%)18.74 (-11.59%)
3a053d3 — 2025-11-26 01:12:160.10 (n/a)0.09 (n/a)0.09 (n/a)0.07 (n/a)0.01 (n/a)172.00 (n/a)134.00 (n/a)126.00 (n/a)114.00 (n/a)21.19 (n/a)

eltwise_add_4_cols_2_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.13 (+37.25%)0.09 (+18.09%)0.08 (+13.77%)0.07 (+13.85%)0.02 (+74.49%)164.10 (-5.69%)136.20 (-7.60%)146.30 (-5.61%)96.20 (-21.79%)23.21 (+19.43%)
3a053d3 — 2025-11-26 01:12:160.09 (n/a)0.08 (n/a)0.07 (n/a)0.07 (n/a)0.01 (n/a)174.00 (n/a)147.40 (n/a)155.00 (n/a)123.00 (n/a)19.44 (n/a)

eltwise_add_8_cols_2_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.12 (+45.72%)0.09 (+17.41%)0.08 (+12.95%)0.07 (+1.97%)0.02 (+209.44%)186.40 (+5.31%)148.30 (-5.54%)149.30 (-4.90%)105.40 (-26.29%)27.80 (+123.17%)
3a053d3 — 2025-11-26 01:12:160.08 (n/a)0.07 (n/a)0.07 (n/a)0.06 (n/a)0.01 (n/a)177.00 (n/a)157.00 (n/a)157.00 (n/a)143.00 (n/a)12.46 (n/a)

eltwise_mul_1_cols_2_channels_2048_tile_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.13 (+23.89%)0.10 (+32.40%)0.10 (+23.99%)0.09 (+63.10%)0.02 (-9.57%)140.90 (-34.16%)122.88 (-21.13%)128.20 (-13.38%)94.50 (-13.30%)16.12 (-53.02%)
3a053d3 — 2025-11-26 01:12:160.10 (n/a)0.08 (n/a)0.08 (n/a)0.05 (n/a)0.02 (n/a)214.00 (n/a)155.80 (n/a)148.00 (n/a)109.00 (n/a)34.31 (n/a)

eltwise_mul_2_cols_2_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.13 (+49.08%)0.09 (+22.70%)0.09 (+16.22%)0.07 (+9.72%)0.02 (+149.12%)167.30 (-2.16%)137.26 (-10.17%)140.40 (-7.63%)94.40 (-27.94%)23.89 (+53.67%)
3a053d3 — 2025-11-26 01:12:160.09 (n/a)0.08 (n/a)0.08 (n/a)0.07 (n/a)0.01 (n/a)171.00 (n/a)152.80 (n/a)152.00 (n/a)131.00 (n/a)15.55 (n/a)

eltwise_mul_4_cols_2_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.12 (+48.31%)0.09 (+27.57%)0.09 (+20.26%)0.07 (+14.25%)0.02 (+168.00%)166.30 (-6.05%)134.74 (-14.18%)136.60 (-10.72%)102.10 (-27.59%)21.06 (+66.52%)
3a053d3 — 2025-11-26 01:12:160.08 (n/a)0.07 (n/a)0.07 (n/a)0.06 (n/a)0.01 (n/a)177.00 (n/a)157.00 (n/a)153.00 (n/a)141.00 (n/a)12.65 (n/a)

eltwise_mul_8_cols_2_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.12 (+14.64%)0.09 (+24.82%)0.09 (+42.82%)0.07 (+14.32%)0.02 (-2.13%)186.00 (-6.06%)140.56 (-15.02%)139.80 (-24.84%)105.80 (-6.37%)26.07 (-19.18%)
3a053d3 — 2025-11-26 01:12:160.10 (n/a)0.07 (n/a)0.06 (n/a)0.06 (n/a)0.02 (n/a)198.00 (n/a)165.40 (n/a)186.00 (n/a)113.00 (n/a)32.25 (n/a)

gelu_1_cols_1_channels_2048_tile_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.08 (-0.77%)0.07 (+19.12%)0.06 (+7.91%)0.06 (+89.12%)0.01 (-51.41%)141.20 (-47.12%)123.04 (-23.58%)127.90 (-7.32%)102.80 (+0.78%)14.41 (-75.13%)
3a053d3 — 2025-11-26 01:12:160.08 (n/a)0.06 (n/a)0.06 (n/a)0.03 (n/a)0.02 (n/a)267.00 (n/a)161.00 (n/a)138.00 (n/a)102.00 (n/a)57.94 (n/a)

gelu_1_cols_2_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.08 (-19.40%)0.06 (-2.63%)0.07 (+8.34%)0.06 (+11.52%)0.01 (-58.09%)148.00 (-10.30%)127.42 (-1.07%)125.50 (-7.72%)109.20 (+24.09%)12.62 (-52.38%)
3a053d3 — 2025-11-26 01:12:160.09 (n/a)0.07 (n/a)0.06 (n/a)0.05 (n/a)0.02 (n/a)165.00 (n/a)128.80 (n/a)136.00 (n/a)88.00 (n/a)26.50 (n/a)

gelu_2_cols_1_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.10 (+19.08%)0.07 (+7.54%)0.06 (-16.93%)0.05 (+52.61%)0.02 (+11.02%)150.70 (-34.48%)121.30 (-9.88%)133.60 (+20.36%)83.10 (-16.06%)27.93 (-42.17%)
3a053d3 — 2025-11-26 01:12:160.08 (n/a)0.07 (n/a)0.07 (n/a)0.04 (n/a)0.02 (n/a)230.00 (n/a)134.60 (n/a)111.00 (n/a)99.00 (n/a)48.30 (n/a)

gelu_2_cols_2_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.07 (+3.35%)0.06 (+17.35%)0.06 (+11.57%)0.05 (+47.12%)0.01 (-62.74%)153.60 (-32.04%)134.44 (-20.36%)130.90 (-10.34%)119.00 (-3.25%)11.36 (-75.85%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.05 (n/a)0.06 (n/a)0.04 (n/a)0.01 (n/a)226.00 (n/a)168.80 (n/a)146.00 (n/a)123.00 (n/a)47.02 (n/a)

gelu_4_cols_1_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.08 (+16.30%)0.07 (+15.08%)0.07 (+8.51%)0.06 (+33.04%)0.01 (+9.85%)148.80 (-24.85%)119.58 (-13.97%)116.10 (-7.86%)98.00 (-14.04%)20.12 (-33.34%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.07 (n/a)0.04 (n/a)0.01 (n/a)198.00 (n/a)139.00 (n/a)126.00 (n/a)114.00 (n/a)30.19 (n/a)

gelu_4_cols_2_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.06 (+6.80%)0.06 (+5.51%)0.05 (+4.04%)0.05 (+2.97%)0.01 (+6.99%)167.00 (-2.91%)148.82 (-5.21%)151.90 (-3.86%)127.30 (-6.40%)13.07 (-4.88%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.05 (n/a)0.05 (n/a)0.05 (n/a)0.00 (n/a)172.00 (n/a)157.00 (n/a)158.00 (n/a)136.00 (n/a)13.74 (n/a)

gelu_8_cols_1_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.06 (-14.62%)0.05 (-10.13%)0.05 (-18.57%)0.04 (+29.02%)0.01 (-50.35%)186.00 (-22.50%)157.36 (+5.33%)154.70 (+22.78%)134.70 (+17.13%)19.69 (-57.32%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.07 (n/a)0.03 (n/a)0.01 (n/a)240.00 (n/a)149.40 (n/a)126.00 (n/a)115.00 (n/a)46.13 (n/a)

gelu_8_cols_2_channels_2048_tile_128

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.07 (+30.26%)0.06 (+12.96%)0.06 (+12.96%)0.04 (+7.25%)0.01 (+120.31%)185.60 (-6.73%)151.26 (-9.21%)141.60 (-11.50%)113.60 (-23.24%)28.25 (+61.12%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.05 (n/a)0.05 (n/a)0.04 (n/a)0.00 (n/a)199.00 (n/a)166.60 (n/a)160.00 (n/a)148.00 (n/a)17.53 (n/a)

gemm_2048x2048x2048_64x64x32_8_cols_0_bcolmaj_0_ccolmaj_0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
3a053d3 — 2025-11-26 01:12:163.72 (-0.11%)3.55 (-2.75%)3.56 (-2.75%)3.36 (-5.33%)0.12 (+76.10%)2725.23 (-0.11%)2602.52 (-2.75%)2612.11 (-2.75%)2459.89 (-5.33%)86.27 (+76.10%)6984.00 (+5.63%)6608.60 (+2.91%)6577.00 (+2.83%)6304.00 (+0.11%)221.87 (+87.15%)
9e98049 — 2025-11-25 20:50:553.72 (n/a)3.65 (n/a)3.66 (n/a)3.54 (n/a)0.07 (n/a)2728.26 (n/a)2676.06 (n/a)2686.03 (n/a)2598.29 (n/a)48.99 (n/a)6612.00 (n/a)6422.00 (n/a)6396.00 (n/a)6297.00 (n/a)118.55 (n/a)

gemm_2048x2048x2048_64x64x32_8_cols_0_bcolmaj_1_ccolmaj_0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
3a053d3 — 2025-11-26 01:12:163.40 (-3.14%)3.38 (-0.82%)3.38 (-0.82%)3.35 (+4.93%)0.02 (-84.51%)2493.81 (-3.14%)2475.27 (-0.82%)2479.06 (-0.82%)2453.92 (+4.93%)13.33 (-84.51%)7001.00 (-4.70%)6940.80 (+0.71%)6930.00 (+0.83%)6889.00 (+3.24%)37.43 (-84.76%)
9e98049 — 2025-11-25 20:50:553.51 (n/a)3.40 (n/a)3.41 (n/a)3.19 (n/a)0.12 (n/a)2574.53 (n/a)2495.72 (n/a)2499.62 (n/a)2338.67 (n/a)86.02 (n/a)7346.00 (n/a)6892.20 (n/a)6873.00 (n/a)6673.00 (n/a)245.57 (n/a)

gemm_2048x2048x2048_64x64x32_8_cols_1_bcolmaj_0_ccolmaj_0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
3a053d3 — 2025-11-26 01:12:163.02 (+0.52%)2.87 (-2.05%)2.86 (-2.24%)2.70 (-6.16%)0.11 (+131.66%)2216.19 (+0.52%)2104.44 (-2.05%)2094.08 (-2.24%)1980.84 (-6.16%)80.25 (+131.67%)8673.00 (+6.56%)8175.60 (+2.21%)8204.00 (+2.29%)7752.00 (-0.51%)314.03 (+145.36%)
9e98049 — 2025-11-25 20:50:553.01 (n/a)2.93 (n/a)2.92 (n/a)2.88 (n/a)0.05 (n/a)2204.81 (n/a)2148.41 (n/a)2142.13 (n/a)2110.81 (n/a)34.64 (n/a)8139.00 (n/a)7998.60 (n/a)8020.00 (n/a)7792.00 (n/a)127.99 (n/a)

gemm_2048x2048x2048_64x64x64_2_cols_0_bcolmaj_0_ccolmaj_0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:031.04 (n/a)1.03 (n/a)1.03 (n/a)1.02 (n/a)0.01 (n/a)24737.00 (n/a)24525.50 (n/a)24522.20 (n/a)24179.00 (n/a)194.34 (n/a)

gemm_2048x2048x2048_64x64x64_2_cols_0_bcolmaj_1_ccolmaj_0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.97 (n/a)0.97 (n/a)0.97 (n/a)0.97 (n/a)0.00 (n/a)25922.40 (n/a)25901.84 (n/a)25899.90 (n/a)25885.50 (n/a)14.56 (n/a)

gemm_2048x2048x2048_64x64x64_2_cols_1_bcolmaj_0_ccolmaj_0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.93 (n/a)0.92 (n/a)0.92 (n/a)0.91 (n/a)0.00 (n/a)27580.20 (n/a)27429.50 (n/a)27425.80 (n/a)27198.20 (n/a)129.73 (n/a)

gemm_2048x2048x2048_64x64x64_8_cols_0_bcolmaj_0_ccolmaj_0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
31ab24e — 2025-11-22 02:01:563.65 (-1.06%)3.59 (+0.83%)3.63 (-0.26%)3.45 (+2.66%)0.07 (-44.71%)2678.08 (-1.06%)2632.13 (+0.83%)2658.60 (-0.26%)2528.31 (+2.66%)53.86 (-44.71%)6795.00 (-2.59%)6529.80 (-0.92%)6462.00 (+0.26%)6415.00 (+1.07%)137.14 (-45.27%)
d9a64c5 — 2025-11-21 20:49:353.69 (n/a)3.56 (n/a)3.64 (n/a)3.36 (n/a)0.13 (n/a)2706.77 (n/a)2610.50 (n/a)2665.61 (n/a)2462.71 (n/a)97.41 (n/a)6976.00 (n/a)6590.40 (n/a)6445.00 (n/a)6347.00 (n/a)250.58 (n/a)

layer_norm_1_cols_1_channels_2048_tile_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.06 (-32.97%)0.06 (-33.81%)0.06 (-34.67%)0.05 (-32.05%)0.00 (-43.33%)154.00 (-1.91%)142.96 (+0.53%)146.90 (+2.01%)129.30 (-0.54%)9.05 (-16.05%)
3a053d3 — 2025-11-26 01:12:160.09 (n/a)0.09 (n/a)0.09 (n/a)0.08 (n/a)0.01 (n/a)157.00 (n/a)142.20 (n/a)144.00 (n/a)130.00 (n/a)10.78 (n/a)

layer_norm_1_cols_2_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.07 (-28.93%)0.06 (-29.36%)0.06 (-28.56%)0.05 (-33.32%)0.01 (-18.73%)170.00 (+0.00%)140.34 (-5.18%)137.20 (-6.67%)116.30 (-6.21%)17.97 (+15.53%)
3a053d3 — 2025-11-26 01:12:160.10 (n/a)0.08 (n/a)0.08 (n/a)0.07 (n/a)0.01 (n/a)170.00 (n/a)148.00 (n/a)147.00 (n/a)124.00 (n/a)15.56 (n/a)

layer_norm_2_cols_1_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.05 (-31.11%)0.05 (-31.25%)0.05 (-38.25%)0.04 (-15.42%)0.01 (-51.70%)205.70 (-21.19%)179.52 (-5.32%)181.40 (+7.98%)149.00 (-3.25%)23.18 (-43.24%)
3a053d3 — 2025-11-26 01:12:160.08 (n/a)0.07 (n/a)0.07 (n/a)0.05 (n/a)0.01 (n/a)261.00 (n/a)189.60 (n/a)168.00 (n/a)154.00 (n/a)40.83 (n/a)

layer_norm_2_cols_2_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.08 (-7.34%)0.07 (-6.34%)0.07 (-5.88%)0.05 (+13.04%)0.01 (-26.82%)163.40 (-41.01%)124.44 (-30.79%)113.30 (-29.19%)102.20 (-28.03%)21.87 (-55.46%)
3a053d3 — 2025-11-26 01:12:160.09 (n/a)0.07 (n/a)0.08 (n/a)0.04 (n/a)0.01 (n/a)277.00 (n/a)179.80 (n/a)160.00 (n/a)142.00 (n/a)49.10 (n/a)

layer_norm_4_cols_1_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.08 (-34.45%)0.06 (-24.67%)0.06 (-27.81%)0.05 (-14.82%)0.01 (-44.60%)160.40 (-21.76%)135.04 (-13.32%)147.80 (-7.62%)105.80 (+1.73%)21.98 (-32.91%)
3a053d3 — 2025-11-26 01:12:160.12 (n/a)0.08 (n/a)0.08 (n/a)0.06 (n/a)0.02 (n/a)205.00 (n/a)155.80 (n/a)160.00 (n/a)104.00 (n/a)32.76 (n/a)

layer_norm_4_cols_2_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.08 (+18.03%)0.07 (+21.47%)0.07 (+17.55%)0.06 (+21.82%)0.01 (-3.21%)146.10 (-45.28%)119.06 (-45.49%)116.30 (-43.27%)106.20 (-43.51%)14.53 (-54.27%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.06 (n/a)0.05 (n/a)0.01 (n/a)267.00 (n/a)218.40 (n/a)205.00 (n/a)188.00 (n/a)31.77 (n/a)

layer_norm_8_cols_1_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.07 (-31.72%)0.06 (-19.45%)0.06 (-15.43%)0.05 (-11.49%)0.01 (-60.87%)150.60 (-24.70%)129.34 (-19.46%)129.30 (-21.16%)110.30 (-2.39%)12.88 (-55.75%)
3a053d3 — 2025-11-26 01:12:160.11 (n/a)0.08 (n/a)0.07 (n/a)0.06 (n/a)0.02 (n/a)200.00 (n/a)160.60 (n/a)164.00 (n/a)113.00 (n/a)29.10 (n/a)

layer_norm_8_cols_2_channels_2048_tile_128

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.07 (+1.20%)0.06 (-6.73%)0.06 (-8.69%)0.05 (-20.89%)0.01 (+105.76%)170.20 (-15.74%)137.14 (-27.59%)134.30 (-27.01%)116.60 (-34.12%)18.15 (+72.22%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.07 (n/a)0.07 (n/a)0.06 (n/a)0.00 (n/a)202.00 (n/a)189.40 (n/a)184.00 (n/a)177.00 (n/a)10.54 (n/a)

matrix_vector_mul_128x128_32_1col

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.28 (+453.14%)0.24 (+431.27%)0.24 (+436.70%)0.21 (+403.19%)0.03 (+694.14%)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)
3a053d3 — 2025-11-26 01:12:160.05 (n/a)0.05 (n/a)0.04 (n/a)0.04 (n/a)0.00 (n/a)0.05 (n/a)0.04 (n/a)0.04 (n/a)0.04 (n/a)0.00 (n/a)790.00 (n/a)738.40 (n/a)749.00 (n/a)659.00 (n/a)50.18 (n/a)

matrix_vector_mul_2048x8192_1_1col

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:0313.31 (+17.77%)13.16 (+21.45%)13.13 (+21.60%)13.08 (+24.27%)0.08 (-69.78%)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)
3a053d3 — 2025-11-26 01:12:1611.30 (n/a)10.83 (n/a)10.80 (n/a)10.52 (n/a)0.27 (n/a)11.29 (n/a)10.83 (n/a)10.79 (n/a)10.52 (n/a)0.27 (n/a)3191.00 (n/a)3101.20 (n/a)3110.00 (n/a)2972.00 (n/a)74.99 (n/a)

matrix_vector_mul_2048x8192_1_2col

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:0325.37 (+44.01%)24.39 (+47.49%)24.19 (+40.26%)24.04 (+76.08%)0.50 (-66.38%)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)
3a053d3 — 2025-11-26 01:12:1617.62 (n/a)16.54 (n/a)17.24 (n/a)13.65 (n/a)1.47 (n/a)17.60 (n/a)16.53 (n/a)17.23 (n/a)13.65 (n/a)1.47 (n/a)2459.00 (n/a)2048.80 (n/a)1947.00 (n/a)1906.00 (n/a)207.97 (n/a)

matrix_vector_mul_2048x8192_1_4col

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:0341.43 (+102.61%)39.97 (+103.55%)40.50 (+109.88%)37.76 (+98.07%)1.42 (+144.85%)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)
3a053d3 — 2025-11-26 01:12:1620.45 (n/a)19.64 (n/a)19.30 (n/a)19.07 (n/a)0.58 (n/a)20.44 (n/a)19.63 (n/a)19.28 (n/a)19.05 (n/a)0.58 (n/a)1761.00 (n/a)1711.20 (n/a)1740.00 (n/a)1642.00 (n/a)49.93 (n/a)

matrix_vector_mul_2048x8192_1_8col

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:0346.83 (+144.21%)44.99 (+140.89%)44.73 (+140.49%)43.23 (+137.79%)1.22 (+228.25%)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)
3a053d3 — 2025-11-26 01:12:1619.17 (n/a)18.68 (n/a)18.60 (n/a)18.18 (n/a)0.37 (n/a)19.16 (n/a)18.67 (n/a)18.59 (n/a)18.17 (n/a)0.37 (n/a)1847.00 (n/a)1798.20 (n/a)1805.00 (n/a)1751.00 (n/a)35.70 (n/a)

matrix_vector_mul_8192x2048_4_1col

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:0313.17 (+20.35%)13.12 (+23.78%)13.11 (+22.02%)13.07 (+34.10%)0.03 (-92.31%)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)
3a053d3 — 2025-11-26 01:12:1610.94 (n/a)10.60 (n/a)10.75 (n/a)9.75 (n/a)0.43 (n/a)10.94 (n/a)10.59 (n/a)10.74 (n/a)9.74 (n/a)0.43 (n/a)3445.00 (n/a)3174.00 (n/a)3124.00 (n/a)3068.00 (n/a)137.12 (n/a)

matrix_vector_mul_8192x2048_4_2col

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:0325.19 (+50.67%)22.74 (+38.18%)24.31 (+48.66%)15.68 (-3.39%)3.55 (+1629.27%)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)
3a053d3 — 2025-11-26 01:12:1616.72 (n/a)16.46 (n/a)16.35 (n/a)16.23 (n/a)0.21 (n/a)16.71 (n/a)16.45 (n/a)16.34 (n/a)16.22 (n/a)0.21 (n/a)2069.00 (n/a)2040.40 (n/a)2053.00 (n/a)2008.00 (n/a)25.37 (n/a)

matrix_vector_mul_8192x2048_4_4col

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:0342.10 (+98.00%)40.56 (+104.93%)41.13 (+99.69%)38.31 (+112.44%)1.39 (+6.28%)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)
3a053d3 — 2025-11-26 01:12:1621.26 (n/a)19.79 (n/a)20.60 (n/a)18.03 (n/a)1.31 (n/a)21.25 (n/a)19.78 (n/a)20.59 (n/a)18.02 (n/a)1.31 (n/a)1862.00 (n/a)1704.00 (n/a)1630.00 (n/a)1579.00 (n/a)115.16 (n/a)

matrix_vector_mul_8192x2048_4_8col

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:0345.99 (+138.86%)39.79 (+138.34%)44.30 (+173.36%)19.45 (+40.45%)10.20 (+410.38%)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)
3a053d3 — 2025-11-26 01:12:1619.25 (n/a)16.70 (n/a)16.20 (n/a)13.85 (n/a)2.00 (n/a)19.24 (n/a)16.68 (n/a)16.19 (n/a)13.84 (n/a)2.00 (n/a)2424.00 (n/a)2040.60 (n/a)2072.00 (n/a)1744.00 (n/a)247.94 (n/a)

mem_copy_16_cores_2_chans_2048_tile_128_False

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.08 (+75.10%)0.06 (+38.10%)0.05 (+24.94%)0.05 (+11.86%)0.01 (+457.77%)180.50 (-3.99%)143.14 (-18.85%)150.40 (-14.06%)98.70 (-38.70%)28.91 (+201.71%)
3a053d3 — 2025-11-26 01:12:160.05 (n/a)0.04 (n/a)0.04 (n/a)0.04 (n/a)0.00 (n/a)188.00 (n/a)176.40 (n/a)175.00 (n/a)161.00 (n/a)9.58 (n/a)

mem_copy_1_cols_1_channels_2048_tile_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
003b49d — 2025-11-14 19:30:540.05 (-16.99%)0.05 (-12.47%)0.05 (-7.32%)0.04 (-11.17%)0.00 (-29.61%)197.00 (+12.57%)171.00 (+13.70%)164.00 (+7.89%)153.00 (+20.47%)17.49 (-4.32%)
0999a9a — 2025-11-13 19:24:080.06 (n/a)0.05 (n/a)0.05 (n/a)0.04 (n/a)0.01 (n/a)175.00 (n/a)150.40 (n/a)152.00 (n/a)127.00 (n/a)18.28 (n/a)

mem_copy_1_cols_2_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
003b49d — 2025-11-14 19:30:540.06 (+5.56%)0.05 (-1.59%)0.05 (-1.39%)0.03 (-14.07%)0.01 (+25.03%)263.00 (+16.37%)166.00 (+4.53%)144.00 (+1.41%)126.00 (-5.26%)49.38 (+43.17%)
0999a9a — 2025-11-13 19:24:080.06 (n/a)0.05 (n/a)0.05 (n/a)0.03 (n/a)0.01 (n/a)226.00 (n/a)158.80 (n/a)142.00 (n/a)133.00 (n/a)34.49 (n/a)

mem_copy_1_cores_1_chans_2048_tile_2048_False

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.08 (+15.43%)0.07 (+24.10%)0.07 (+10.25%)0.06 (+41.63%)0.01 (-2.40%)136.50 (-24.17%)116.78 (-14.76%)122.70 (-2.62%)96.70 (-7.02%)16.23 (-37.98%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.06 (n/a)0.04 (n/a)0.01 (n/a)180.00 (n/a)137.00 (n/a)126.00 (n/a)104.00 (n/a)26.17 (n/a)

mem_copy_2_cols_1_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
003b49d — 2025-11-14 19:30:540.06 (-15.32%)0.05 (-17.09%)0.05 (-23.81%)0.04 (-15.54%)0.01 (-18.18%)193.00 (+18.40%)160.80 (+20.36%)168.00 (+31.25%)124.00 (+18.10%)22.92 (+11.39%)
0999a9a — 2025-11-13 19:24:080.07 (n/a)0.06 (n/a)0.06 (n/a)0.05 (n/a)0.01 (n/a)163.00 (n/a)133.60 (n/a)128.00 (n/a)105.00 (n/a)20.58 (n/a)

mem_copy_2_cols_2_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
003b49d — 2025-11-14 19:30:540.06 (-23.02%)0.05 (-6.98%)0.05 (-2.13%)0.04 (-15.31%)0.01 (-33.29%)196.00 (+18.07%)147.00 (+6.52%)141.00 (+2.17%)126.00 (+29.90%)25.27 (+8.13%)
0999a9a — 2025-11-13 19:24:080.08 (n/a)0.06 (n/a)0.06 (n/a)0.05 (n/a)0.01 (n/a)166.00 (n/a)138.00 (n/a)138.00 (n/a)97.00 (n/a)23.37 (n/a)

mem_copy_2_cores_1_chans_2048_tile_1024_False

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.08 (+29.73%)0.07 (+32.03%)0.07 (+21.84%)0.06 (+51.87%)0.01 (-1.36%)142.10 (-29.30%)115.72 (-20.19%)112.80 (-11.88%)98.50 (-17.23%)16.06 (-47.09%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.05 (n/a)0.06 (n/a)0.04 (n/a)0.01 (n/a)201.00 (n/a)145.00 (n/a)128.00 (n/a)119.00 (n/a)30.35 (n/a)

mem_copy_2_cores_2_chans_2048_tile_1024_False

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.08 (+28.40%)0.06 (+15.29%)0.06 (+12.13%)0.05 (+2.41%)0.01 (+134.91%)159.40 (+4.87%)131.44 (-5.44%)131.20 (-4.23%)105.40 (-16.35%)18.48 (+90.21%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.06 (n/a)0.06 (n/a)0.05 (n/a)0.00 (n/a)152.00 (n/a)139.00 (n/a)137.00 (n/a)126.00 (n/a)9.72 (n/a)

mem_copy_4_cols_1_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
003b49d — 2025-11-14 19:30:540.07 (+11.97%)0.05 (+3.09%)0.05 (+2.10%)0.03 (-11.35%)0.01 (+41.65%)229.00 (+12.81%)156.00 (-0.51%)143.00 (-2.05%)117.00 (-10.69%)38.21 (+48.71%)
0999a9a — 2025-11-13 19:24:080.06 (n/a)0.05 (n/a)0.05 (n/a)0.04 (n/a)0.01 (n/a)203.00 (n/a)156.80 (n/a)146.00 (n/a)131.00 (n/a)25.69 (n/a)

mem_copy_4_cols_2_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
003b49d — 2025-11-14 19:30:540.06 (+9.02%)0.05 (+5.59%)0.05 (+0.00%)0.04 (+3.63%)0.01 (+25.48%)193.00 (-3.50%)147.40 (-4.66%)147.00 (+0.00%)122.00 (-8.27%)25.24 (+7.48%)
0999a9a — 2025-11-13 19:24:080.06 (n/a)0.05 (n/a)0.05 (n/a)0.04 (n/a)0.01 (n/a)200.00 (n/a)154.60 (n/a)147.00 (n/a)133.00 (n/a)23.48 (n/a)

mem_copy_4_cores_1_chans_2048_tile_512_False

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.07 (+5.93%)0.06 (+6.21%)0.06 (+3.85%)0.06 (+13.92%)0.01 (-0.95%)148.00 (-5.73%)134.90 (+0.82%)143.70 (+3.38%)109.50 (+1.39%)15.16 (-9.88%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.05 (n/a)0.05 (n/a)0.01 (n/a)157.00 (n/a)133.80 (n/a)139.00 (n/a)108.00 (n/a)16.82 (n/a)

mem_copy_4_cores_2_chans_2048_tile_512_False

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.10 (+52.33%)0.08 (+37.94%)0.08 (+21.04%)0.07 (+58.38%)0.01 (+43.03%)123.40 (-32.20%)102.72 (-22.53%)102.90 (-11.29%)78.20 (-29.55%)17.15 (-35.48%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.07 (n/a)0.04 (n/a)0.01 (n/a)182.00 (n/a)132.60 (n/a)116.00 (n/a)111.00 (n/a)26.58 (n/a)

mem_copy_8_cols_1_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
003b49d — 2025-11-14 19:30:540.05 (-13.48%)0.05 (-17.79%)0.05 (-15.33%)0.03 (-25.09%)0.01 (+6.67%)267.00 (+33.50%)176.80 (+24.51%)150.00 (+18.11%)141.00 (+15.57%)47.41 (+61.41%)
0999a9a — 2025-11-13 19:24:080.06 (n/a)0.06 (n/a)0.06 (n/a)0.04 (n/a)0.01 (n/a)200.00 (n/a)142.00 (n/a)127.00 (n/a)122.00 (n/a)29.37 (n/a)

mem_copy_8_cols_2_channels_2048_tile_128

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
003b49d — 2025-11-14 19:30:540.05 (+13.38%)0.05 (+23.85%)0.05 (+37.91%)0.05 (+30.72%)0.00 (-56.85%)166.00 (-23.50%)153.40 (-20.60%)153.00 (-27.49%)142.00 (-11.80%)7.63 (-70.60%)
0999a9a — 2025-11-13 19:24:080.05 (n/a)0.04 (n/a)0.04 (n/a)0.04 (n/a)0.01 (n/a)217.00 (n/a)193.20 (n/a)211.00 (n/a)161.00 (n/a)25.96 (n/a)

mem_copy_8_cores_1_chans_2048_tile_256_False

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.07 (+16.83%)0.06 (+29.39%)0.06 (+21.04%)0.05 (+85.22%)0.01 (-40.63%)158.30 (-42.01%)130.98 (-22.04%)130.40 (-11.29%)115.80 (-8.10%)14.97 (-71.98%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.05 (n/a)0.05 (n/a)0.03 (n/a)0.01 (n/a)273.00 (n/a)168.00 (n/a)147.00 (n/a)126.00 (n/a)53.43 (n/a)

mem_copy_8_cores_2_chans_2048_tile_256_False

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.07 (+18.35%)0.06 (+19.12%)0.06 (+25.36%)0.05 (+14.20%)0.01 (+43.57%)177.70 (-5.98%)139.10 (-9.20%)127.60 (-14.36%)117.90 (-9.31%)21.79 (+12.07%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.05 (n/a)0.05 (n/a)0.04 (n/a)0.01 (n/a)189.00 (n/a)153.20 (n/a)149.00 (n/a)130.00 (n/a)19.45 (n/a)

mha

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)GFLOP/s (max)GFLOP/s (mean)GFLOP/s (median)GFLOP/s (min)GFLOP/s (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.18 (-87.94%)0.18 (-88.00%)0.18 (-88.02%)0.18 (-88.01%)0.00 (-69.92%)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)47327.40 (n/a)47185.94 (n/a)47236.50 (n/a)46905.90 (n/a)145.28 (n/a)
3a053d3 — 2025-11-26 01:12:161.48 (n/a)1.48 (n/a)1.48 (n/a)1.48 (n/a)0.00 (n/a)759.22 (n/a)758.39 (n/a)758.89 (n/a)756.71 (n/a)0.94 (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)n/a (n/a)

relu_1_cols_1_channels_2048_tile_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.09 (+37.15%)0.07 (+30.19%)0.07 (+24.16%)0.05 (+7.33%)0.01 (+198.75%)152.10 (+0.07%)118.22 (-15.56%)121.90 (-13.55%)96.30 (-21.71%)20.36 (+116.01%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.05 (n/a)0.05 (n/a)0.05 (n/a)0.00 (n/a)152.00 (n/a)140.00 (n/a)141.00 (n/a)123.00 (n/a)9.42 (n/a)

relu_2_cols_1_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.08 (+14.47%)0.07 (+28.71%)0.07 (+28.49%)0.05 (+52.38%)0.01 (-5.40%)169.80 (-29.54%)119.48 (-19.81%)112.80 (-16.44%)96.60 (-6.21%)26.53 (-44.56%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.06 (n/a)0.03 (n/a)0.01 (n/a)241.00 (n/a)149.00 (n/a)135.00 (n/a)103.00 (n/a)47.85 (n/a)

relu_4_cols_1_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.11 (+64.02%)0.08 (+44.76%)0.07 (+28.69%)0.06 (+44.31%)0.02 (+112.66%)132.40 (-25.62%)110.92 (-24.54%)116.00 (-16.55%)77.20 (-34.58%)19.76 (-5.19%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.05 (n/a)0.05 (n/a)0.04 (n/a)0.01 (n/a)178.00 (n/a)147.00 (n/a)139.00 (n/a)118.00 (n/a)20.84 (n/a)

relu_8_cols_1_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.08 (+40.03%)0.06 (+26.40%)0.06 (+21.10%)0.05 (+26.16%)0.01 (+71.13%)168.50 (-14.90%)137.08 (-14.33%)138.30 (-11.35%)105.80 (-23.33%)20.56 (+0.59%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.05 (n/a)0.05 (n/a)0.04 (n/a)0.01 (n/a)198.00 (n/a)160.00 (n/a)156.00 (n/a)138.00 (n/a)20.44 (n/a)

rms_norm_1_cols_1_channels_2048_tile_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.07 (-36.43%)0.06 (-33.04%)0.06 (-36.02%)0.06 (-28.14%)0.01 (-39.75%)146.40 (-0.41%)133.00 (+6.57%)144.30 (+11.86%)110.30 (+12.55%)15.60 (-1.77%)
3a053d3 — 2025-11-26 01:12:160.12 (n/a)0.09 (n/a)0.09 (n/a)0.08 (n/a)0.01 (n/a)147.00 (n/a)124.80 (n/a)129.00 (n/a)98.00 (n/a)15.88 (n/a)

rms_norm_1_cols_2_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.06 (-34.56%)0.05 (-32.20%)0.06 (-25.92%)0.05 (-35.15%)0.01 (-32.49%)171.10 (+10.39%)151.08 (+5.65%)145.90 (-3.38%)133.40 (+9.34%)15.34 (+13.45%)
3a053d3 — 2025-11-26 01:12:160.09 (n/a)0.08 (n/a)0.08 (n/a)0.07 (n/a)0.01 (n/a)155.00 (n/a)143.00 (n/a)151.00 (n/a)122.00 (n/a)13.52 (n/a)

rms_norm_2_cols_1_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.08 (-14.83%)0.07 (-0.59%)0.07 (-8.39%)0.05 (+36.12%)0.01 (-62.74%)158.80 (-47.42%)125.80 (-36.85%)117.20 (-21.87%)105.90 (-15.95%)18.31 (-76.98%)
3a053d3 — 2025-11-26 01:12:160.09 (n/a)0.07 (n/a)0.08 (n/a)0.04 (n/a)0.02 (n/a)302.00 (n/a)199.20 (n/a)150.00 (n/a)126.00 (n/a)79.57 (n/a)

rms_norm_2_cols_2_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.07 (-16.59%)0.07 (-12.94%)0.07 (-0.08%)0.05 (-18.42%)0.01 (-5.48%)152.70 (-12.24%)126.46 (-17.45%)115.30 (-28.39%)111.60 (-14.15%)16.68 (-0.73%)
3a053d3 — 2025-11-26 01:12:160.09 (n/a)0.08 (n/a)0.07 (n/a)0.07 (n/a)0.01 (n/a)174.00 (n/a)153.20 (n/a)161.00 (n/a)130.00 (n/a)16.80 (n/a)

rms_norm_4_cols_1_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.08 (-8.44%)0.07 (-10.87%)0.07 (-7.37%)0.05 (-18.19%)0.01 (+26.71%)156.60 (-12.51%)128.26 (-18.82%)126.00 (-22.70%)105.50 (-21.85%)18.42 (+21.77%)
3a053d3 — 2025-11-26 01:12:160.08 (n/a)0.07 (n/a)0.07 (n/a)0.06 (n/a)0.01 (n/a)179.00 (n/a)158.00 (n/a)163.00 (n/a)135.00 (n/a)15.13 (n/a)

rms_norm_4_cols_2_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.07 (-11.64%)0.06 (-18.14%)0.05 (-23.81%)0.04 (-5.09%)0.01 (-23.94%)191.60 (-24.57%)152.64 (-13.86%)149.40 (-6.04%)112.60 (-18.99%)25.47 (-37.88%)
3a053d3 — 2025-11-26 01:12:160.08 (n/a)0.07 (n/a)0.07 (n/a)0.05 (n/a)0.01 (n/a)254.00 (n/a)177.20 (n/a)159.00 (n/a)139.00 (n/a)41.01 (n/a)

rms_norm_8_cols_1_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.08 (+2.13%)0.06 (-5.08%)0.06 (-15.89%)0.05 (+24.15%)0.01 (-30.62%)156.80 (-42.35%)138.18 (-26.58%)142.10 (-14.91%)108.60 (-29.94%)15.95 (-63.06%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.07 (n/a)0.04 (n/a)0.01 (n/a)272.00 (n/a)188.20 (n/a)167.00 (n/a)155.00 (n/a)43.17 (n/a)

rms_norm_8_cols_2_channels_2048_tile_128

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.07 (+0.27%)0.06 (-4.50%)0.06 (-6.59%)0.05 (-15.44%)0.01 (+64.47%)167.60 (-15.35%)134.28 (-24.31%)129.50 (-23.37%)119.90 (-28.63%)17.15 (+43.39%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.07 (n/a)0.06 (n/a)0.00 (n/a)198.00 (n/a)177.40 (n/a)169.00 (n/a)168.00 (n/a)11.96 (n/a)

rope_1_cols_2_channels_4096_tile_4096_0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.24 (+62.35%)0.20 (+52.02%)0.21 (+44.36%)0.15 (+59.34%)0.03 (+46.93%)161.10 (-32.59%)123.80 (-29.66%)119.00 (-25.62%)101.90 (-33.83%)19.81 (-38.02%)
3a053d3 — 2025-11-26 01:12:160.15 (n/a)0.13 (n/a)0.14 (n/a)0.10 (n/a)0.02 (n/a)239.00 (n/a)176.00 (n/a)160.00 (n/a)154.00 (n/a)31.96 (n/a)

rope_2_cols_2_channels_4096_tile_2048_0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.26 (+43.43%)0.19 (+39.62%)0.18 (+26.88%)0.15 (+74.92%)0.04 (+19.61%)167.00 (-38.60%)133.26 (-25.39%)136.20 (-15.40%)93.60 (-25.12%)23.64 (-52.74%)
3a053d3 — 2025-11-26 01:12:160.18 (n/a)0.14 (n/a)0.14 (n/a)0.08 (n/a)0.03 (n/a)272.00 (n/a)178.60 (n/a)161.00 (n/a)125.00 (n/a)50.03 (n/a)

rope_4_cols_2_channels_4096_tile_1024_0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.19 (-2.77%)0.16 (-1.82%)0.16 (+7.27%)0.12 (-13.79%)0.03 (+30.42%)196.80 (+24.56%)157.22 (+10.87%)149.10 (+0.07%)127.00 (+10.43%)26.28 (+68.34%)
3a053d3 — 2025-11-26 01:12:160.20 (n/a)0.16 (n/a)0.15 (n/a)0.14 (n/a)0.02 (n/a)158.00 (n/a)141.80 (n/a)149.00 (n/a)115.00 (n/a)15.61 (n/a)

rope_8_cols_2_channels_4096_tile_512_0

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.19 (+13.57%)0.16 (+9.04%)0.16 (+6.16%)0.14 (+4.42%)0.02 (+61.28%)169.70 (+2.85%)151.58 (-1.06%)154.70 (+1.11%)129.50 (-5.47%)13.91 (+46.14%)
3a053d3 — 2025-11-26 01:12:160.17 (n/a)0.15 (n/a)0.15 (n/a)0.14 (n/a)0.01 (n/a)165.00 (n/a)153.20 (n/a)153.00 (n/a)137.00 (n/a)9.52 (n/a)

sigmoid_1_cols_1_channels_2048_tile_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.06 (+12.60%)0.06 (+18.51%)0.06 (+23.35%)0.05 (+11.64%)0.00 (+19.56%)155.80 (-3.83%)137.08 (-9.34%)130.60 (-12.93%)127.80 (-4.63%)10.34 (+1.94%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.05 (n/a)0.05 (n/a)0.05 (n/a)0.00 (n/a)162.00 (n/a)151.20 (n/a)150.00 (n/a)134.00 (n/a)10.15 (n/a)

sigmoid_2_cols_1_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.08 (+49.37%)0.06 (+33.81%)0.06 (+19.62%)0.06 (+88.52%)0.01 (+0.05%)143.50 (-43.06%)129.56 (-22.33%)132.90 (-10.20%)99.90 (-28.13%)15.81 (-63.22%)
3a053d3 — 2025-11-26 01:12:160.05 (n/a)0.05 (n/a)0.05 (n/a)0.03 (n/a)0.01 (n/a)252.00 (n/a)166.80 (n/a)148.00 (n/a)139.00 (n/a)42.99 (n/a)

sigmoid_4_cols_1_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.08 (+42.60%)0.07 (+55.23%)0.07 (+60.01%)0.05 (+84.64%)0.01 (-9.15%)151.20 (-41.85%)126.28 (-33.11%)118.80 (-32.88%)106.20 (-24.68%)16.36 (-62.42%)
3a053d3 — 2025-11-26 01:12:160.05 (n/a)0.04 (n/a)0.04 (n/a)0.03 (n/a)0.01 (n/a)260.00 (n/a)188.80 (n/a)177.00 (n/a)141.00 (n/a)43.55 (n/a)

sigmoid_8_cols_1_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.10 (+89.60%)0.08 (+54.55%)0.07 (+33.51%)0.06 (+29.87%)0.02 (+615.28%)131.50 (-17.30%)109.88 (-27.52%)123.90 (-19.55%)79.90 (-43.33%)21.90 (+212.79%)
3a053d3 — 2025-11-26 01:12:160.05 (n/a)0.05 (n/a)0.05 (n/a)0.05 (n/a)0.00 (n/a)159.00 (n/a)151.60 (n/a)154.00 (n/a)141.00 (n/a)7.00 (n/a)

silu_1_cols_1_channels_2048_tile_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.08 (+18.28%)0.07 (+22.23%)0.07 (+15.64%)0.06 (+28.22%)0.01 (+1.72%)141.50 (-16.27%)114.88 (-12.84%)114.20 (-7.15%)97.10 (-9.25%)15.14 (-28.56%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.06 (n/a)0.05 (n/a)0.01 (n/a)169.00 (n/a)131.80 (n/a)123.00 (n/a)107.00 (n/a)21.19 (n/a)

silu_2_cols_1_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.07 (+9.01%)0.06 (+8.55%)0.07 (+9.50%)0.05 (-2.38%)0.01 (+37.43%)162.80 (+10.00%)128.74 (-0.20%)118.60 (-1.98%)111.30 (-1.50%)19.65 (+35.11%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.06 (n/a)0.05 (n/a)0.01 (n/a)148.00 (n/a)129.00 (n/a)121.00 (n/a)113.00 (n/a)14.55 (n/a)

silu_4_cols_1_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.09 (+24.50%)0.07 (+15.02%)0.07 (+20.85%)0.06 (+10.12%)0.01 (+53.75%)145.30 (-2.48%)123.02 (-5.80%)121.70 (-11.17%)94.90 (-13.73%)17.34 (+20.60%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.06 (n/a)0.05 (n/a)0.01 (n/a)149.00 (n/a)130.60 (n/a)137.00 (n/a)110.00 (n/a)14.37 (n/a)

silu_8_cols_1_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.09 (+49.00%)0.07 (+33.20%)0.08 (+44.32%)0.05 (+28.20%)0.02 (+106.42%)166.70 (-16.23%)125.04 (-16.19%)106.40 (-25.59%)90.10 (-27.92%)33.40 (+22.59%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.05 (n/a)0.05 (n/a)0.04 (n/a)0.01 (n/a)199.00 (n/a)149.20 (n/a)143.00 (n/a)125.00 (n/a)27.25 (n/a)

softmax_1_cols_2_channels_4096_tile_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.17 (-18.34%)0.13 (-20.06%)0.12 (-17.78%)0.10 (-24.77%)0.02 (-11.22%)158.00 (-4.82%)128.74 (-9.97%)131.50 (-12.91%)96.40 (-12.36%)19.64 (+3.18%)
3a053d3 — 2025-11-26 01:12:160.21 (n/a)0.16 (n/a)0.15 (n/a)0.14 (n/a)0.02 (n/a)166.00 (n/a)143.00 (n/a)151.00 (n/a)110.00 (n/a)19.04 (n/a)

softmax_2_cols_2_channels_4096_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.17 (-9.83%)0.14 (-6.89%)0.13 (-1.18%)0.10 (-7.77%)0.02 (-14.65%)163.00 (-22.38%)125.22 (-23.46%)126.00 (-27.59%)96.10 (-20.58%)23.49 (-25.85%)
3a053d3 — 2025-11-26 01:12:160.19 (n/a)0.15 (n/a)0.13 (n/a)0.11 (n/a)0.03 (n/a)210.00 (n/a)163.60 (n/a)174.00 (n/a)121.00 (n/a)31.68 (n/a)

softmax_2_cols_2_channels_4096_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.18 (+8.11%)0.15 (+8.44%)0.15 (+12.70%)0.13 (+6.98%)0.02 (+10.35%)121.80 (-33.08%)109.52 (-33.94%)108.60 (-36.49%)89.40 (-33.78%)11.66 (-31.00%)
3a053d3 — 2025-11-26 01:12:160.17 (n/a)0.14 (n/a)0.13 (n/a)0.13 (n/a)0.02 (n/a)182.00 (n/a)165.80 (n/a)171.00 (n/a)135.00 (n/a)16.89 (n/a)

swiglu

No metrics available.

swiglu_decode_1x2048x2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.00 (n/a)0.00 (n/a)0.00 (n/a)0.00 (n/a)0.00 (n/a)5205.40 (n/a)4880.18 (n/a)4826.40 (n/a)4649.40 (n/a)186.72 (n/a)

tanh_1_cols_1_channels_2048_tile_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.09 (+11.02%)0.08 (+32.03%)0.08 (+32.57%)0.05 (+62.56%)0.01 (-20.55%)159.80 (-33.97%)112.74 (-23.41%)101.20 (-19.04%)95.80 (-3.23%)23.78 (-53.04%)
3a053d3 — 2025-11-26 01:12:160.08 (n/a)0.06 (n/a)0.06 (n/a)0.03 (n/a)0.02 (n/a)242.00 (n/a)147.20 (n/a)125.00 (n/a)99.00 (n/a)50.64 (n/a)

tanh_2_cols_1_channels_2048_tile_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.08 (+8.55%)0.07 (+18.89%)0.07 (+12.65%)0.06 (+38.50%)0.01 (-39.02%)129.50 (-22.46%)118.38 (-11.79%)119.10 (-4.72%)100.90 (-1.08%)10.04 (-57.41%)
3a053d3 — 2025-11-26 01:12:160.07 (n/a)0.06 (n/a)0.06 (n/a)0.05 (n/a)0.01 (n/a)167.00 (n/a)134.20 (n/a)125.00 (n/a)102.00 (n/a)23.57 (n/a)

tanh_4_cols_1_channels_2048_tile_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.09 (+51.03%)0.07 (+27.86%)0.07 (+18.17%)0.05 (+28.73%)0.01 (+86.09%)151.00 (-16.57%)120.48 (-14.92%)120.80 (-9.17%)88.20 (-28.87%)20.67 (-0.34%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.05 (n/a)0.06 (n/a)0.04 (n/a)0.01 (n/a)181.00 (n/a)141.60 (n/a)133.00 (n/a)124.00 (n/a)20.74 (n/a)

tanh_8_cols_1_channels_2048_tile_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.07 (+20.20%)0.05 (+17.33%)0.05 (+12.65%)0.04 (+13.17%)0.01 (+41.19%)182.20 (-5.10%)154.48 (-7.83%)163.00 (-4.68%)118.80 (-10.68%)21.79 (+13.00%)
3a053d3 — 2025-11-26 01:12:160.06 (n/a)0.05 (n/a)0.04 (n/a)0.04 (n/a)0.01 (n/a)192.00 (n/a)167.60 (n/a)171.00 (n/a)133.00 (n/a)19.28 (n/a)

transpose_2048_M_64_N_1_cols_1_channels_64_m_64_n_8_s

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:034.19 (+11.56%)3.99 (+24.30%)4.04 (+24.81%)3.72 (+47.22%)0.19 (-52.96%)140.80 (-27.05%)131.66 (-14.84%)129.90 (-13.97%)125.10 (-3.77%)6.23 (-70.11%)
3a053d3 — 2025-11-26 01:12:163.76 (n/a)3.21 (n/a)3.23 (n/a)2.53 (n/a)0.39 (n/a)193.00 (n/a)154.60 (n/a)151.00 (n/a)130.00 (n/a)20.83 (n/a)

transpose_2048_M_64_N_1_cols_2_channels_64_m_64_n_8_s

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:034.24 (+18.15%)3.82 (+40.37%)3.75 (+37.57%)3.50 (+87.84%)0.24 (-55.26%)149.80 (-42.82%)137.78 (-26.48%)139.70 (-21.96%)123.60 (-9.12%)8.58 (-79.08%)
3a053d3 — 2025-11-26 01:12:163.59 (n/a)2.72 (n/a)2.73 (n/a)1.86 (n/a)0.55 (n/a)262.00 (n/a)187.40 (n/a)179.00 (n/a)136.00 (n/a)41.03 (n/a)

weighted_rms_norm_1_cols_2_channels_2048_weights_2048

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.08 (-2.15%)0.07 (-12.51%)0.07 (-10.28%)0.05 (-24.37%)0.01 (+47.66%)173.20 (-5.36%)129.56 (-15.98%)120.50 (-20.20%)97.30 (-26.84%)26.69 (+45.13%)
3a053d3 — 2025-11-26 01:12:160.09 (n/a)0.08 (n/a)0.08 (n/a)0.06 (n/a)0.01 (n/a)183.00 (n/a)154.20 (n/a)151.00 (n/a)133.00 (n/a)18.39 (n/a)

weighted_rms_norm_2_cols_2_channels_2048_weights_1024

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.09 (-10.95%)0.07 (-9.84%)0.07 (-2.68%)0.05 (-25.51%)0.02 (+23.49%)161.40 (-3.93%)124.48 (-18.00%)116.20 (-26.46%)89.20 (-19.64%)29.44 (+40.14%)
3a053d3 — 2025-11-26 01:12:160.10 (n/a)0.08 (n/a)0.07 (n/a)0.07 (n/a)0.01 (n/a)168.00 (n/a)151.80 (n/a)158.00 (n/a)111.00 (n/a)21.01 (n/a)

weighted_rms_norm_4_cols_2_channels_2048_weights_512

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.08 (-10.05%)0.07 (-13.72%)0.07 (-15.26%)0.06 (-12.84%)0.01 (-11.92%)137.20 (-17.84%)119.46 (-17.04%)116.60 (-15.51%)98.70 (-20.40%)13.27 (-20.40%)
3a053d3 — 2025-11-26 01:12:160.09 (n/a)0.08 (n/a)0.08 (n/a)0.07 (n/a)0.01 (n/a)167.00 (n/a)144.00 (n/a)138.00 (n/a)124.00 (n/a)16.67 (n/a)

weighted_rms_norm_8_cols_2_channels_2048_weights_256

Commit/Date Bandwidth (max)Bandwidth (mean)Bandwidth (median)Bandwidth (min)Bandwidth (stddev)Latency (max)Latency (mean)Latency (median)Latency (min)Latency (stddev)
79c1558 — 2025-12-03 22:56:030.07 (-14.17%)0.06 (-9.80%)0.07 (-6.27%)0.05 (+19.26%)0.01 (-47.04%)151.90 (-39.96%)133.36 (-23.09%)123.70 (-23.64%)118.40 (-16.62%)14.87 (-63.48%)
3a053d3 — 2025-11-26 01:12:160.08 (n/a)0.07 (n/a)0.07 (n/a)0.05 (n/a)0.01 (n/a)253.00 (n/a)173.40 (n/a)162.00 (n/a)142.00 (n/a)40.72 (n/a)

@andrej andrej merged commit 0a6c11c into amd:devel Dec 3, 2025
3 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants