perf: vectorize fp8 quantize and swizzle gemm smem

2026-08-23 20:31:44 +08:00
parent 2bc4d2b8a8
commit 4b10d3ca37
7 changed files with 387 additions and 182 deletions
@@ -78,7 +78,7 @@ class _CMakeBuildExt(_build_ext):
        if cmake is None:
            raise RuntimeError("cmake not found on PATH; install it to build kernels")

-        parallel = os.environ.get("BUILD_PARALLEL", "16")
+        parallel = os.environ.get("BUILD_PARALLEL", "4")
        cfg = [
            cmake,
            "-S",