diff --git a/tools/util/include/cutlass/util/reference/device/tensor_foreach.h b/tools/util/include/cutlass/util/reference/device/tensor_foreach.h
index 728c0a02f0..37e238e86e 100644
--- a/tools/util/include/cutlass/util/reference/device/tensor_foreach.h
+++ b/tools/util/include/cutlass/util/reference/device/tensor_foreach.h
@@ -54,7 +54,9 @@ struct TensorForEach {
 #if defined (CUTLASS_ENABLE_SYCL)
       // TODO: query the queue for block size
       block_size = 128;
-      grid_size = (size(size) + block_size - 1) / block_size;
+      grid_size = (size.product() + block_size - 1) / block_size;
+      int sm_count = KernelHardwareInfo::query_device_multiprocessor_count();
+      grid_size = grid_size > sm_count / 2 ? sm_count / 2 : grid_size;
 #else
       // if grid_size or block_size are zero, query occupancy using the CUDA Occupancy API
       cudaError_t result = cudaOccupancyMaxPotentialBlockSize(
@@ -75,7 +77,7 @@ struct TensorForEach {
 #if defined(CUTLASS_ENABLE_SYCL)
     const auto sycl_block = syclcompat::dim3(block_size, 1, 1);
     const auto sycl_grid = syclcompat::dim3(grid_size, 1, 1);
-    syclcompat::launch<kernel::TensorForEach<Func, Rank, Params>>(sycl_grid, sycl_block, 0, size, params);
+    syclcompat::launch<kernel::TensorForEach<Func, Rank, Params>>(sycl_grid, sycl_block, size, params);
 #else
     dim3 grid(grid_size, 1, 1);
     dim3 block(block_size, 1, 1);
@@ -103,7 +105,7 @@ struct TensorDiagonalForEach {
 #if defined(CUTLASS_ENABLE_SYCL)
     const auto sycl_block = syclcompat::dim3(block_size, 1, 1);
     const auto sycl_grid = syclcompat::dim3((end - start + block_size - 1) / block_size, 1, 1);
-    syclcompat::launch<kernel::TensorDiagonalForEach<Func, Rank, Params>>(sycl_grid, sycl_block, 0, size, params, start, end);
+    syclcompat::launch<kernel::TensorDiagonalForEach<Func, Rank, Params>>(sycl_grid, sycl_block, size, params, start, end);
 #else
     dim3 block(block_size, 1, 1);
     dim3 grid((end - start + block_size - 1) / block_size, 1, 1);
@@ -153,7 +155,7 @@ struct BlockForEach {
 #if defined(CUTLASS_ENABLE_SYCL)
     const auto sycl_block = syclcompat::dim3(block_size, 1, 1);
     const auto sycl_grid = syclcompat::dim3(grid_size, 1, 1);
-    syclcompat::launch<kernel::BlockForEach<Element, Func>>(sycl_grid, sycl_block, 0, ptr, capacity, params);
+    syclcompat::launch<kernel::BlockForEach<Element, Func>>(sycl_grid, sycl_block, ptr, capacity, params);
 #else
     dim3 grid(grid_size, 1, 1);
     dim3 block(block_size, 1, 1);