Causal Conv1D Benchmarks - Aggregated Results

This document combines benchmark results from multiple Causal Conv1D implementations.

Combined Summary and Visualization

▶ code ▼ output ▶ uv-logs | Cell: combine | 4.43s | Raw

# /// script
# requires-python = ">=3.10"
# dependencies = [
#     "numpy",
#     "torch==2.8.0",
#     "kernels-benchmark-tools",
#     "matplotlib",
# ]
#
# [tool.uv.sources]
# kernels-benchmark-tools = { path = "../../../../../tools", editable = true }
# ///
from kernels_benchmark_tools.core.visuals import generate_combined_results

# Map display names to uvnote environment variables
cache_env_map = {
    "HF Kernels Causal Conv1D": "UVNOTE_FILE_HF_KERNELS_CAUSAL_CONV1D_BENCHMARK",
    "PyTorch Causal Conv1D": "UVNOTE_FILE_TORCH_CAUSAL_CONV1D_BENCHMARK",
}

# Generate combined results with visualization
generate_combined_results(
    cache_env_map=cache_env_map,
    output_filename="causal_conv1d.jsonl",
    svg_filename="latency.svg"
)

======================================================================
LOADING BENCHMARK DATA
======================================================================
✓ HF Kernels Causal Conv1D      : /__w/kernels-benchmarks/kernels-benchmarks/benches/causal_conv1d/impls/.uvnote/cache/7a691bd653e23c412c5d29fbc92ea1454823ea437864cf9473fc561b116ef3d9
✓ PyTorch Causal Conv1D         : /__w/kernels-benchmarks/kernels-benchmarks/benches/causal_conv1d/impls/.uvnote/cache/70757e27f2df1dfde4905a24527bb4ca6f0f8df7dac2e2ecaa0ddc359c7d5e64

  ✓ Found HF Kernels Causal Conv1D
     Path: /__w/kernels-benchmarks/kernels-benchmarks/benches/causal_conv1d/impls/.uvnote/cache/7a691bd653e23c412c5d29fbc92ea1454823ea437864cf9473fc561b116ef3d9/causal_conv1d.jsonl
  ✓ Found PyTorch Causal Conv1D
     Path: /__w/kernels-benchmarks/kernels-benchmarks/benches/causal_conv1d/impls/.uvnote/cache/70757e27f2df1dfde4905a24527bb4ca6f0f8df7dac2e2ecaa0ddc359c7d5e64/causal_conv1d.jsonl

======================================================================
Summary: 2 found, 0 skipped, 0 missing
======================================================================

COMBINED BENCHMARK SUMMARY

impl                     wl                  p50(ms)  ok
hf_kernels_causal_conv1d cuda_B2_D2048_S128_W2     0.05  True
hf_kernels_causal_conv1d cuda_B2_D2048_S128_W4     0.05  True
hf_kernels_causal_conv1d cuda_B2_D2048_S2048_W2     0.05  True
hf_kernels_causal_conv1d cuda_B2_D2048_S2048_W4     0.05  True
hf_kernels_causal_conv1d cuda_B2_D2048_S512_W2     0.05  True
hf_kernels_causal_conv1d cuda_B2_D2048_S512_W4     0.05  True
hf_kernels_causal_conv1d cuda_B2_D64_S128_W2     0.05  True
hf_kernels_causal_conv1d cuda_B2_D64_S128_W4     0.05  True
hf_kernels_causal_conv1d cuda_B2_D64_S2048_W2     0.05  True
hf_kernels_causal_conv1d cuda_B2_D64_S2048_W4     0.05  True
hf_kernels_causal_conv1d cuda_B2_D64_S512_W2     0.05  True
hf_kernels_causal_conv1d cuda_B2_D64_S512_W4     0.05  True
hf_kernels_causal_conv1d cuda_B4_D2048_S128_W2     0.05  True
hf_kernels_causal_conv1d cuda_B4_D2048_S128_W4     0.05  True
hf_kernels_causal_conv1d cuda_B4_D2048_S2048_W2     0.05  True
hf_kernels_causal_conv1d cuda_B4_D2048_S2048_W4     0.05  True
hf_kernels_causal_conv1d cuda_B4_D2048_S512_W2     0.05  True
hf_kernels_causal_conv1d cuda_B4_D2048_S512_W4     0.05  True
hf_kernels_causal_conv1d cuda_B4_D64_S128_W2     0.05  True
hf_kernels_causal_conv1d cuda_B4_D64_S128_W4     0.05  True
hf_kernels_causal_conv1d cuda_B4_D64_S2048_W2     0.05  True
hf_kernels_causal_conv1d cuda_B4_D64_S2048_W4     0.05  True
hf_kernels_causal_conv1d cuda_B4_D64_S512_W2     0.05  True
hf_kernels_causal_conv1d cuda_B4_D64_S512_W4     0.05  True
torch_eager              cuda_B2_D2048_S128_W2     0.09  True
torch_eager              cuda_B2_D2048_S128_W4     0.08  True
torch_eager              cuda_B2_D2048_S2048_W2     0.15  True
torch_eager              cuda_B2_D2048_S2048_W4     0.16  True
torch_eager              cuda_B2_D2048_S512_W2     0.09  True
torch_eager              cuda_B2_D2048_S512_W4     0.09  True
torch_eager              cuda_B2_D64_S128_W2     0.07  True
torch_eager              cuda_B2_D64_S128_W4     0.09  True
torch_eager              cuda_B2_D64_S2048_W2     0.09  True
torch_eager              cuda_B2_D64_S2048_W4     0.09  True
torch_eager              cuda_B2_D64_S512_W2     0.09  True
torch_eager              cuda_B2_D64_S512_W4     0.09  True
torch_eager              cuda_B4_D2048_S128_W2     0.09  True
torch_eager              cuda_B4_D2048_S128_W4     0.09  True
torch_eager              cuda_B4_D2048_S2048_W2     0.49  True
torch_eager              cuda_B4_D2048_S2048_W4     0.50  True
torch_eager              cuda_B4_D2048_S512_W2     0.10  True
torch_eager              cuda_B4_D2048_S512_W4     0.10  True
torch_eager              cuda_B4_D64_S128_W2     0.09  True
torch_eager              cuda_B4_D64_S128_W4     0.08  True
torch_eager              cuda_B4_D64_S2048_W2     0.09  True
torch_eager              cuda_B4_D64_S2048_W4     0.09  True
torch_eager              cuda_B4_D64_S512_W2     0.09  True
torch_eager              cuda_B4_D64_S512_W4     0.09  True

GENERATING COMBINED VISUALIZATION

Loaded 48 records
✓ Visualization saved as latency.svg
Saved latency.png
✓ Visualization saved as latency.svg
✓ SVG visualization ready!

ANALYSIS COMPLETE
Total implementations analyzed: 2

Implementations included:
  ✓ HF Kernels Causal Conv1D
  ✓ PyTorch Causal Conv1D

▶ UV Install Logs

Artifacts:

latency.svg