gpuemu logo gpuemu

The evidence (P1–P4)

In short: every gpuemu default comes from a measured study. The headline results: 9/9 LLM-style bugs caught across 5 GPU classes with 0 false positives on 15 controls (P1); recall lifted from 65% to 82% by calibrated tolerances (P2); 99% recall from adversarial input generation (P3).

9/9
LLM-style kernel bugs caught
across 5 GPU classes (P1)
0
false positives
on 15/15 control kernels (P1)
99%
bug recall
under adversarial inputs (P3)
65→82%
recall from calibrated tolerances
+23 pp over fixed atol/rtol (P2)
P1

The correctness illusion in LLM-generated GPU kernels

A hardware-free fuzz oracle catches 9/9 LLM-style bugs across 5 GPU classes (RTX 3060, A10, L40S, A100 SXM4, H100 NVL) with 0 false positives on 15/15 controls.

P2

Operator-aware mixed-precision tolerance calibration

A p95-of-controls × 1.5 envelope raises kernel-bug recall from 65% to 82% over the field-standard fixed atol/rtol — at zero precision cost.

P3

Test-input generation for tensor programs

A seven-strategy ablation: adversarial value sampling wins at 99% recall; "regular shape only" misses 100% of tail-mask bugs.

P4

Static PTX metrics track structural regressions but miss semantic ones

Structural Δregs / Δinstrs predicts Δperf% consistently across 5 GPU classes; semantic bugs compile to identical PTX and need the correctness oracle.

P3: the recall swing from input strategy

Holding the kernel and oracle fixed, only the input-generation strategy changed. Adversarial sampling beats the field-standard default by +28 percentage points, and beats “regular shapes only” by 35.

Input strategy Bug recall
Adversarial values (gpuemu default) 99%
Field-standard default oracle 71%
Regular shapes only 64%

Replayable, not just reported

Each study ships as a LaTeX manuscript plus run-id records and a kernel corpus you can replay. Because gpuemu’s RNG is bit-identical across Rust and Python, every result reproduces byte-for-byte from its seed. The full manuscripts and run records live in the gpuemu research program.

Stop shipping silently-wrong kernels

Open source, dual-licensed MIT / Apache-2.0. Validate your first kernel in five minutes — or talk to us about an enterprise pilot.