From a7db97ca5aaf7f5437f0476829e664da1de9212a Mon Sep 17 00:00:00 2001
From: "Plyakhin, Yury" <yury.plyakhin@intel.com>
Date: Wed, 20 Nov 2024 17:49:55 -0800
Subject: [PATCH] new tests

---
 sycl/test-e2e/Matrix/element_wise_all_ops_impl.hpp            | 1 +
 sycl/test-e2e/Matrix/element_wise_ops_impl.hpp                | 1 +
 sycl/test-e2e/Matrix/joint_matrix_bf16_fill_k_cache_impl.hpp  | 4 ++++
 .../test-e2e/Matrix/joint_matrix_rowmajorA_rowmajorB_impl.hpp | 2 ++
 sycl/test-e2e/Matrix/joint_matrix_transposeC_impl.hpp         | 2 ++
 5 files changed, 10 insertions(+)

diff --git a/sycl/test-e2e/Matrix/element_wise_all_ops_impl.hpp b/sycl/test-e2e/Matrix/element_wise_all_ops_impl.hpp
index 22539e1441102..9fc3acbf9ffcd 100644
--- a/sycl/test-e2e/Matrix/element_wise_all_ops_impl.hpp
+++ b/sycl/test-e2e/Matrix/element_wise_all_ops_impl.hpp
@@ -229,6 +229,7 @@ int main() {
         test_ewops_c<float, 16, 16>();
 // This combination is not currently supported for sub group size = 32 in IGC
 #if (!defined(SG_SZ) || SG_SZ != 32)
+        test_ewops_ab<bfloat16,  1, 32, use::a, layout::row_major, 1>();
         test_ewops_ab<bfloat16, 32, 16, use::a, layout::row_major, 1>();
         test_ewops_ab<bfloat16, 32, 32, use::a, layout::row_major, 1>();
         test_ewops_ab<bfloat16, 16, 64, use::b, layout::ext_intel_packed, 2>();
diff --git a/sycl/test-e2e/Matrix/element_wise_ops_impl.hpp b/sycl/test-e2e/Matrix/element_wise_ops_impl.hpp
index f1577e732a739..44929295e6363 100644
--- a/sycl/test-e2e/Matrix/element_wise_ops_impl.hpp
+++ b/sycl/test-e2e/Matrix/element_wise_ops_impl.hpp
@@ -133,6 +133,7 @@ int main() {
       // IGC
       passed &= test<bfloat16, float, 16, 16, 16, 2, class pvc_bf16_16x16x16>();
       passed &= test<bfloat16, float, 1, 64, 16, 2, class pvc_bf16_1x64x16>();
+      passed &= test<bfloat16, float, 1, 64, 32, 2, class pvc_bf16_1x64x32>();
       passed &= test<bfloat16, float, 32, 64, 16, 2, class pvc_bf16_32x64x16>();
       passed &= test<bfloat16, float, 32, 64, 32, 2, class pvc_bf16_32x64x32>();
 #endif
diff --git a/sycl/test-e2e/Matrix/joint_matrix_bf16_fill_k_cache_impl.hpp b/sycl/test-e2e/Matrix/joint_matrix_bf16_fill_k_cache_impl.hpp
index 22ce2b3f0e16a..e0faccafe8628 100644
--- a/sycl/test-e2e/Matrix/joint_matrix_bf16_fill_k_cache_impl.hpp
+++ b/sycl/test-e2e/Matrix/joint_matrix_bf16_fill_k_cache_impl.hpp
@@ -483,6 +483,10 @@ size_t matrix_size = -1;
            MCache1, NCache1, KCache1, MCache2, NCache2, KCache2>(matrix_size);
       test<bfloat16, float, VnniFactor, /*TM*/ 32, /*TN*/ 64, /*TK*/ 16,
            MCache1, NCache1, KCache1, MCache2, NCache2, KCache2>(matrix_size);
+      test<bfloat16, float, VnniFactor, /*TM*/ 1, /*TN*/ 64, /*TK*/ 32,
+           MCache1, NCache1, KCache1, MCache2, NCache2, KCache2>(matrix_size);
+      test<bfloat16, float, VnniFactor, /*TM*/ 32, /*TN*/ 64, /*TK*/ 32,
+           MCache1, NCache1, KCache1, MCache2, NCache2, KCache2>(matrix_size);
 #endif
       break;
     }
diff --git a/sycl/test-e2e/Matrix/joint_matrix_rowmajorA_rowmajorB_impl.hpp b/sycl/test-e2e/Matrix/joint_matrix_rowmajorA_rowmajorB_impl.hpp
index e83ab94b41512..c0ba6bf16d1b5 100644
--- a/sycl/test-e2e/Matrix/joint_matrix_rowmajorA_rowmajorB_impl.hpp
+++ b/sycl/test-e2e/Matrix/joint_matrix_rowmajorA_rowmajorB_impl.hpp
@@ -127,6 +127,8 @@ int main() {
                               bfloat16, float>();
         res += gemm_row_major<1, 64, 16, class bf16_1x64x16, bfloat16, bfloat16,
                               float>();
+        res += gemm_row_major<1, 64, 32, class bf16_1x64x32, bfloat16, bfloat16,
+                              float>();
         res += gemm_row_major<32, 64, 16, class bf16_32x64x16, bfloat16,
                               bfloat16, float>();
         res += gemm_row_major<32, 64, 32, class bf16_32x64x32, bfloat16,
diff --git a/sycl/test-e2e/Matrix/joint_matrix_transposeC_impl.hpp b/sycl/test-e2e/Matrix/joint_matrix_transposeC_impl.hpp
index 58505b6fd4fb6..7c040db041f8d 100644
--- a/sycl/test-e2e/Matrix/joint_matrix_transposeC_impl.hpp
+++ b/sycl/test-e2e/Matrix/joint_matrix_transposeC_impl.hpp
@@ -124,6 +124,7 @@ int main() {
     }
 
     if (combinations[i].nsize == 16) { // architecture::intel_gpu_pvc
+      run_matrix_test<float, /*TM*/32, /*TN*/ 64>();
       run_matrix_test<float, /*TM*/ 8, /*TN*/ 16>();
       run_matrix_test<float, /*TM*/ 7, /*TN*/ 16>();
       run_matrix_test<float, /*TM*/ 6, /*TN*/ 16>();
@@ -131,6 +132,7 @@ int main() {
       run_matrix_test<float, /*TM*/ 4, /*TN*/ 16>();
       run_matrix_test<float, /*TM*/ 3, /*TN*/ 16>();
       run_matrix_test<float, /*TM*/ 2, /*TN*/ 16>();
+      run_matrix_test<float, /*TM*/ 1, /*TN*/ 64>();
       run_matrix_test<float, /*TM*/ 1, /*TN*/ 16>();
       break;
     }