- route dX/dW through the fused 128x64 fast kernel via contiguous transposes
- drop the legacy 64x64 kernel, cutting dX 1.55->0.38 ms and dW 1.28->0.26 ms
- sync all threads after cp.async.wait_group to fix sporadic NaN in large GEMMs
- add fp8_mm_prequant_fp8 custom op for FP8-in/FP8-out GEMM
- amax for delayed scale was the quantized max (always ~448), so scale collapsed to 1
- this made fp8 gradients diverge (cosine 0.05) and training stall
- stop w/x transpose-quantize amax from polluting the grad scale
- dX/dW run as fp8 cublasLt gemms via fused transpose-cast
- shared (m,k,n) algo cache for fwd/bwd, mutex-protected
- bias add in-place on bf16 output, drop output copy
- pass w as param A (op=T) and x as param B (op=N) so the col-major [N,M] output storage is row-major C[M,N] directly, zero copy
- transpose_bias_cast kernel becomes a plain bias+write kernel
- fp8 e2e now beats bf16: 1.09x at M=4096, 1.06x at M=8192 (was 0.88x)
ViperEkura