add reqd_sub_group_size (#1248)

intel · Jan 8, 2025 · 28e7c8c · 28e7c8c
1 parent ba04a4a
commit 28e7c8c
Showing 1 changed file with 2 additions and 1 deletion.
diff --git a/src/ATen/native/xpu/sycl/LinearInt4.cpp b/src/ATen/native/xpu/sycl/LinearInt4.cpp
@@ -39,7 +39,8 @@ struct LinearInt4KernelFunctor : public __SYCL_KER_CONFIG_CONVENTION__ {
         ldc(ldc) {}
   void sycl_ker_config_convention(sycl::handler& cgh) {}
 
-  void operator()(sycl::nd_item<1> it) const {
+  [[intel::reqd_sub_group_size(16)]] void operator()(
+      sycl::nd_item<1> it) const {
     int constexpr Unroll = 2;
     int constexpr SgSize = 16;
     int constexpr blocksize = block_size;