diff --git a/sycl/test/matrix/matrix-int8-test-SG-16.cpp b/sycl/test/matrix/matrix-int8-test-SG-16.cpp
index 0aace5f46c036..2905a2ea66311 100644
--- a/sycl/test/matrix/matrix-int8-test-SG-16.cpp
+++ b/sycl/test/matrix/matrix-int8-test-SG-16.cpp
@@ -75,7 +75,7 @@ void matrix_multiply(big_matrix<T1, NUM_ROWS_C, NUM_COLS_C> &C, big_matrix<T2, N
            for (int k = 0; k < K / TK; k += 1) {
              joint_matrix_load(
                  sg, sub_a, accA.get_pointer() + (sg_startx * TM) * K + k * TK,
-                 K, matrix_layout::packed_a);
+                 K, matrix_layout::row_major);
              // Assuming B data is already in VNNI format.
              joint_matrix_load(sg, sub_b,
                                accB.get_pointer() + (k * TK / 4) * (N * 4) +
diff --git a/sycl/test/matrix/matrix-int8-test.cpp b/sycl/test/matrix/matrix-int8-test.cpp
index 263cc75c1097f..4bd4a0aa16742 100644
--- a/sycl/test/matrix/matrix-int8-test.cpp
+++ b/sycl/test/matrix/matrix-int8-test.cpp
@@ -75,7 +75,7 @@ void matrix_multiply(big_matrix<T1, NUM_ROWS_C, NUM_COLS_C> &C, big_matrix<T2, N
            for (int k = 0; k < K / TK; k += 1) {
              joint_matrix_load(
                  sg, sub_a, accA.get_pointer() + (sg_startx * TM) * K + k * TK,
-                 K, matrix_layout::packed_a);
+                 K, matrix_layout::row_major);
              // Assuming B data is already in VNNI format.
              joint_matrix_load(sg, sub_b,
                                accB.get_pointer() + (k * TK / 4) * (N * 4) +