KernelBench hard · RTX PRO 6000

W4A16 GEMM GPT-5.6 Sol

manually audited: clean

Genuine fused W4A16 kernel pair, built via the PROMPT's stated intended path (CUTLASS mixed-input GEMM). Prefill path (M>1): SM90a TmaWarpSpecializedCooperative mixed-input mainloop instantiated from a cloned CUTLASS checkout, narrow operand uint4b_t with per-group (g=128) bf16 scale + zero fed through the mainloop's register transform; the agent patched cutlass mixed_input_utils.hpp (5-line diff) to change the affine order from stock q*scale+bias to (q+(-zero))*scale so dequant matches the reference's bf16 rounding order — a numerics fix in its own dependency, not grader tampering. Decode path (M=1): hand-written CUDA GEMV, 32 warps per CTA each owning one 128-K quant group, magic-number bf16x2 nibble dequant (0x43004300 bias trick) with fully coalesced packed-byte loads and a shared-memory cross-group reduction. Dequant + matmul happen fused every forward call on live inputs; no forbidden ops (no bitsandbytes, no marlin, no F.linear). Two Triton kernels remain in solution.py as dead code (earlier iteration; Model.forward never calls _launch). peak_fraction 0.217 (decode shapes 409-569 GB/s vs 2.0 TB/s HBM2e) is physically plausible, nowhere near the >>1.0 signature of an output-cache lookup. template_mutated=false.

20260721_150450_codex_gpt-5.6-sol_07_w4a16_gemm