feat[fuzz]: cuda kernel fuzzer #6131

joseph-isaacs · 2026-01-23T19:36:00Z

No description provided.

codspeed-hq · 2026-01-23T19:43:51Z

Merging this PR will degrade performance by 84.24%

⚠️

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.

⚡ 11 improved benchmarks
❌ 8 regressed benchmarks
✅ 1246 untouched benchmarks
⏩ 1294 skipped benchmarks¹

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	`BASE`	`HEAD`	Efficiency
⚡	WallTime	`u64_values_u8_codes[10M]`	260.3 µs	222.8 µs	+16.84%
❌	WallTime	`u8_FoR[10M]`	8 µs	50.9 µs	-84.24%
⚡	Simulation	`canonical_into_non_nullable[(10000, 1, 0.0)]`	36.2 µs	25.8 µs	+40.31%
⚡	Simulation	`canonical_into_non_nullable[(10000, 1, 0.01)]`	41.1 µs	32.3 µs	+27.31%
⚡	Simulation	`canonical_into_non_nullable[(10000, 10, 0.1)]`	471.6 µs	382.2 µs	+23.4%
⚡	Simulation	`canonical_into_non_nullable[(10000, 1, 0.1)]`	57 µs	48.1 µs	+18.63%
⚡	Simulation	`canonical_into_non_nullable[(10000, 10, 0.0)]`	278.9 µs	195.6 µs	+42.57%
⚡	Simulation	`canonical_into_non_nullable[(10000, 10, 0.01)]`	306 µs	222.7 µs	+37.45%
❌	Simulation	`into_canonical_non_nullable[(10000, 1, 0.1)]`	55.3 µs	63.7 µs	-13.17%
❌	Simulation	`into_canonical_non_nullable[(10000, 1, 0.0)]`	33 µs	41.1 µs	-19.65%
⚡	Simulation	`into_canonical_non_nullable[(10000, 10, 0.01)]`	310.4 µs	227.8 µs	+36.24%
❌	Simulation	`canonical_into_nullable[(10000, 100, 0.0)]`	4.3 ms	4.9 ms	-12.51%
❌	Simulation	`into_canonical_non_nullable[(10000, 1, 0.01)]`	39.2 µs	47.1 µs	-16.92%
⚡	Simulation	`into_canonical_non_nullable[(10000, 10, 0.0)]`	283.5 µs	200.2 µs	+41.58%
⚡	Simulation	`into_canonical_non_nullable[(10000, 10, 0.1)]`	472.8 µs	383.7 µs	+23.24%
❌	Simulation	`into_canonical_nullable[(10000, 10, 0.0)]`	456 µs	536.8 µs	-15.06%
⚡	Simulation	`into_canonical_nullable[(10000, 10, 0.1)]`	716.9 µs	627.6 µs	+14.23%
❌	Simulation	`into_canonical_nullable[(10000, 100, 0.0)]`	4.3 ms	5 ms	-14.39%
❌	Simulation	`into_canonical_nullable[(10000, 100, 0.1)]`	6.1 ms	6.9 ms	-11.97%

_{Comparing ji/fuzz-arb-gpu (6c8c9e8) with develop (a66a58d)}

1294 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. ↩

0ax1 · 2026-01-26T13:25:33Z

@claude review

claude · 2026-01-26T13:26:05Z

[gone]

.github/workflows/fuzz.yml

vortex-array/src/arrays/primitive/compute/take/avx2.rs

fuzz/build.rs

vortex/Cargo.toml

fuzz/Cargo.toml

Signed-off-by: Alexander Droste <alexander.droste@protonmail.com>

joseph-isaacs force-pushed the ji/fuzz-arb-gpu branch from 7c01a71 to f8efeb8 Compare January 23, 2026 19:37

joseph-isaacs changed the title ~~feat[cuda]: to_host canonical~~ feat[fuzz]: cuda kernel fuzzer Jan 23, 2026

joseph-isaacs requested a review from 0ax1 January 26, 2026 13:08

joseph-isaacs marked this pull request as ready for review January 26, 2026 13:08

joseph-isaacs added the changelog/chore A trivial change label Jan 26, 2026

joseph-isaacs enabled auto-merge (squash) January 26, 2026 13:09