[SYCL][Joint Matirx] Add test for multiple elem-wise ops

Add new test for multiple elem-wise ops for different
matrix types in the same kernel

Author: YuriPlyakhin

sycl/test-e2e/Matrix/XMX8/element_wise_abc.cpp

@@ -0,0 +1,15 @@
+//==----------- element_wise_abc.cpp  - DPC++ joint_matrix------------- ----==//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+// REQUIRES: matrix-xmx8
+
+// RUN: %{build} -o %t.out -DSYCL_EXT_ONEAPI_MATRIX_VERSION=4
+// RUN: %{run} %t.out
+
+#define SG_SZ 8
+
+#include "../element_wise_abc_impl.hpp"

sycl/test-e2e/Matrix/element_wise_abc.cpp

@@ -0,0 +1,15 @@
+//==----------- element_wise_abc.cpp  - DPC++ joint_matrix------------- ----==//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+// REQUIRES: matrix
+
+// RUN: %{build} -o %t.out -DSYCL_EXT_ONEAPI_MATRIX_VERSION=4
+// RUN: %{run} %t.out
+
+#define SG_SZ 16
+
+#include "element_wise_abc_impl.hpp"

sycl/test-e2e/Matrix/element_wise_abc_impl.hpp

@@ -0,0 +1,127 @@
+//==----------- element_wise_abc_impl.hpp  - DPC++ joint_matrix-------------
+//----==//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include <iostream>
+#include <sycl/sycl.hpp>
+
+using namespace sycl;
+using namespace sycl::ext::oneapi::experimental::matrix;
+
+#define TM 8
+#define TN SG_SZ
+#define TK 32
+
+template <typename T, size_t NUM_ROWS, size_t NUM_COLS> struct big_matrix {
+public:
+  T *mat;
+
+public:
+  T *get_data() { return mat; }
+  void set_data(T *data) { mat = data; }
+  big_matrix(T *data) : mat(data) {}
+};
+
+template <typename T1, typename T2, size_t NUM_ROWS_A, size_t NUM_COLS_A,
+          size_t NUM_ROWS_B, size_t NUM_COLS_B, size_t NUM_ROWS_C,
+          size_t NUM_COLS_C>
+void matrix_elem_wise_ops(big_matrix<T1, NUM_ROWS_C, NUM_COLS_C> &C,
+                          big_matrix<T2, NUM_ROWS_A, NUM_COLS_A> &A,
+                          big_matrix<T2, NUM_ROWS_B, NUM_COLS_B> &B) {
+  size_t M = NUM_ROWS_C;
+  size_t N = NUM_COLS_C;
+  size_t K = NUM_COLS_A;
+
+  // B => K/4 x N*4, A => M x K, C => M, N
+  // stride should be X's cols, e.g., B's stirde = N*4
+  assert(NUM_ROWS_C == NUM_ROWS_A && NUM_COLS_A == NUM_ROWS_B * 4);
+
+  size_t NDRangeM = M / TM;
+  size_t NDRangeN = N / TN;
+  buffer<int8_t, 2> bufA(A.get_data(), range<2>(M, K));
+  buffer<int8_t, 2> bufB(B.get_data(), range<2>(K, N));
+  buffer<int32_t, 2> bufC(C.get_data(), range<2>(M, N));
+
+  queue q;
+  q.submit([&](handler &cgh) {
+     auto accC = bufC.get_access<access::mode::read_write>(cgh);
+     auto accA = bufA.get_access<access::mode::read_write>(cgh);
+     auto accB = bufB.get_access<access::mode::read_write>(cgh);
+
+     cgh.parallel_for(
+         nd_range<2>({NDRangeM, NDRangeN * SG_SZ}, {1, 1 * SG_SZ}),
+         [accA, accB, accC, M, N,
+          K](nd_item<2> spmd_item) [[intel::reqd_sub_group_size(SG_SZ)]] {
+           // The submatrix API has to be accessed by all the workitems in a
+           // subgroup these functions will be called once by the subgroup no
+           // code divergence between the workitems
+           const auto global_idx = spmd_item.get_global_id(0);
+           const auto global_idy = spmd_item.get_global_id(1);
+           const auto sg_startx = global_idx - spmd_item.get_local_id(0);
+           const auto sg_starty = global_idy - spmd_item.get_local_id(1);
+
+           ext::oneapi::sub_group sg = spmd_item.get_sub_group();
+           joint_matrix<sub_group, int8_t, use::a, TM, TK, layout::row_major>
+               sub_a;
+
+           // For B, we assume B has been already VNNIed.
+           joint_matrix<sub_group, int8_t, use::b, TK, TN,
+                        ext::intel::experimental::matrix::layout::packed>
+               sub_b;
+
+           joint_matrix<sub_group, int32_t, use::accumulator, TM, TN> sub_c;
+
+           joint_matrix_load(
+               sg, sub_c,
+               accC.template get_multi_ptr<access::decorated::no>() +
+                   (sg_startx * TM) * N + sg_starty / SG_SZ * TN,
+               N, layout::row_major);
+
+           joint_matrix_load(
+               sg, sub_a,
+               accA.template get_multi_ptr<access::decorated::no>() +
+                   (sg_startx * TM) * K,
+               K);
+           auto wi_slice_a =
+               sycl::ext::intel::experimental::matrix::get_wi_data(sg, sub_a);
+           for (int i = 0; i < wi_slice_a.length(); i++) {
+             wi_slice_a[i] += 1;
+           }
+
+           joint_matrix_load(
+               sg, sub_b,
+               accB.template get_multi_ptr<access::decorated::no>() +
+                   +sg_starty / SG_SZ * TN * 4,
+               N * 4);
+           auto wi_slice_b =
+               sycl::ext::intel::experimental::matrix::get_wi_data(sg, sub_b);
+           for (int i = 0; i < wi_slice_b.length(); i++) {
+             wi_slice_b[i] += 1;
+           }
+
+           auto wi_slice_c =
+               sycl::ext::intel::experimental::matrix::get_wi_data(sg, sub_c);
+           for (int i = 0; i < wi_slice_c.length(); i++) {
+             wi_slice_c[i] += 1;
+           }
+         }); // parallel for
+   }).wait();
+}
+
+int8_t A[TM][TK];
+int8_t B[TK / 4][TN * 4];
+int32_t C[TM][TN];
+
+int main() {
+  big_matrix<int32_t, TM, TN> MC((int32_t *)&C);
+  big_matrix<int8_t, TM, TK> MA((int8_t *)&A);
+  big_matrix<int8_t, TK / 4, TN * 4> MB((int8_t *)&B);
+  matrix_elem_wise_ops(MC, MA, MB);
+
+  return 0;
+}