Skip to content

bench: run the full compress suite under --gpu-decompress - #9399

Open
joseph-isaacs wants to merge 12 commits into
ji/cuda-delta-kernelfrom
claude/gpu-compress-benchmarks-x788xc
Open

bench: run the full compress suite under --gpu-decompress#9399
joseph-isaacs wants to merge 12 commits into
ji/cuda-delta-kernelfrom
claude/gpu-compress-benchmarks-x788xc

fix(cuda): stage the OnPair dictionary from device-resident buffers

3daf6d6
Select commit
Loading
Failed to load commit list.
CodSpeed HQ / CodSpeed Performance Analysis failed Aug 21, 2026 in 0s

Performance Regression: -14.64%

⚠️ Unknown Walltime execution environment detected

Using the Walltime instrument on standard Hosted Runners will lead to inconsistent data.

For the most accurate results, we recommend using CodSpeed Macro Runners: bare-metal machines fine-tuned for performance measurement consistency.

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

❌ 1 regressed benchmark
✅ 2010 untouched benchmarks
🆕 12 new benchmarks
⏩ 12 skipped benchmarks1
🗄️ 1 archived benchmark run2

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

Mode Benchmark BASE HEAD Efficiency
❌ Simulation take[duplicates/repeated/primitive/nonnull/chunks=16/indices=1000] 228.9 µs 268.1 µs -14.64%
🆕 WallTime words_gather_dispatch_avx512[1024] N/A 9 ns N/A
🆕 WallTime words_gather_dispatch_avx512[65536] N/A 1.4 µs N/A
🆕 WallTime words_gather_scalar_avx512[1024] N/A 100 ns N/A
🆕 WallTime words_gather_scalar_avx512[65536] N/A 6.4 µs N/A
🆕 WallTime words_gather_dispatch_avx2[1024] N/A 17 ns N/A
🆕 WallTime words_gather_dispatch_avx2[65536] N/A 1.3 µs N/A
🆕 WallTime words_gather_scalar_avx2[1024] N/A 147 ns N/A
🆕 WallTime words_gather_scalar_avx2[65536] N/A 9.4 µs N/A
🆕 WallTime words_gather_dispatch_neon[1024] N/A 30 ns N/A
🆕 WallTime words_gather_dispatch_neon[65536] N/A 2 µs N/A
🆕 WallTime words_gather_scalar_neon[1024] N/A 358 ns N/A
🆕 WallTime words_gather_scalar_neon[65536] N/A 22.5 µs N/A

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

Comparing claude/gpu-compress-benchmarks-x788xc (3daf6d6) with ji/cuda-delta-kernel (a91e43d)3

Open in CodSpeed

Footnotes

  1. 12 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

  2. 1 benchmark was run, but is now archived. If it was deleted in another branch, consider rebasing to remove it from the report. Instead if it was added back, click here to restore it. ↩

  3. No successful run was found on ji/cuda-delta-kernel (811aba2) during the generation of this report, so 815e901 was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩

View more details on CodSpeed HQ