[CI] Build macOS arm64 CPU wheel natively on the macmini queue (#48289)

Signed-off-by: mgoin <mgoin64@gmail.com>
This commit is contained in:
Michael Goin
2026-07-14 19:40:26 +00:00
committed by GitHub
parent 32e632dfeb
commit 0b54201a04
4 changed files with 90 additions and 6 deletions
+3 -2
View File
@@ -570,7 +570,8 @@ TORCH_LIBRARY_EXPAND(TORCH_EXTENSION_NAME, ops) {
#endif
// fused moe
#if defined(__AVX512F__) || (defined(ARM_BF16_SUPPORT))
#if defined(__AVX512F__) || \
(defined(__aarch64__) && !defined(__APPLE__) && defined(ARM_BF16_SUPPORT))
ops.def(
"prepack_moe_weight(Tensor weight, Tensor(a1!) packed_weight, str isa) "
"-> ()");
@@ -581,7 +582,7 @@ TORCH_LIBRARY_EXPAND(TORCH_EXTENSION_NAME, ops) {
"bool skip_weighted, "
"str act, str isa) -> ()");
ops.impl("cpu_fused_moe", torch::kCPU, &cpu_fused_moe);
#endif // #if defined(__AVX512F__) || (defined(ARM_BF16_SUPPORT))
#endif
ops.def(
"mla_decode_kvcache("
" Tensor! out, Tensor query, Tensor kv_cache,"