diff --git a/SYCL/Matrix/element_wise_all_ops_bf16.cpp b/SYCL/Matrix/element_wise_all_ops_bf16.cpp
index 993194d6b5..eb817fd851 100644
--- a/SYCL/Matrix/element_wise_all_ops_bf16.cpp
+++ b/SYCL/Matrix/element_wise_all_ops_bf16.cpp
@@ -7,7 +7,7 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: matrix
 
-// RUN: %clangxx -fsycl %s -o %t.out -DSYCL_EXT_ONEAPI_MATRIX=1
+// RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 
diff --git a/SYCL/Matrix/element_wise_all_ops_cuda.cpp b/SYCL/Matrix/element_wise_all_ops_cuda.cpp
index 739fcc365c..832194e3d9 100644
--- a/SYCL/Matrix/element_wise_all_ops_cuda.cpp
+++ b/SYCL/Matrix/element_wise_all_ops_cuda.cpp
@@ -7,7 +7,7 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: cuda
 
-// RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple -Xsycl-target-backend --cuda-gpu-arch=sm_80 -DSYCL_EXT_ONEAPI_MATRIX=3 %s -o %t.out
+// RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple -Xsycl-target-backend --cuda-gpu-arch=sm_80 -DSYCL_EXT_ONEAPI_MATRIX_VERSION=3 %s -o %t.out
 // RUN: %t.out
 
 #include <sycl/sycl.hpp>
diff --git a/SYCL/Matrix/element_wise_all_ops_half.cpp b/SYCL/Matrix/element_wise_all_ops_half.cpp
index 91891d6e6a..16972bb980 100644
--- a/SYCL/Matrix/element_wise_all_ops_half.cpp
+++ b/SYCL/Matrix/element_wise_all_ops_half.cpp
@@ -9,7 +9,7 @@
 
 // Only runs on DPAS because AMX implementation does not support half data type
 // yet
-// RUN: %clangxx -fsycl %s -o %t.out -DSYCL_EXT_ONEAPI_MATRIX=1
+// RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 
 #include <iostream>
diff --git a/SYCL/Matrix/element_wise_all_ops_int8.cpp b/SYCL/Matrix/element_wise_all_ops_int8.cpp
index cce3f227c8..91cf80a600 100644
--- a/SYCL/Matrix/element_wise_all_ops_int8.cpp
+++ b/SYCL/Matrix/element_wise_all_ops_int8.cpp
@@ -7,7 +7,7 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: matrix
 
-// RUN: %clangxx -fsycl %s -o %t.out -DSYCL_EXT_ONEAPI_MATRIX=1
+// RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 
diff --git a/SYCL/Matrix/element_wise_all_ops_int8_packed.cpp b/SYCL/Matrix/element_wise_all_ops_int8_packed.cpp
index a11d92f431..2d5d934d5a 100644
--- a/SYCL/Matrix/element_wise_all_ops_int8_packed.cpp
+++ b/SYCL/Matrix/element_wise_all_ops_int8_packed.cpp
@@ -7,7 +7,7 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: matrix
 
-// RUN: %clangxx -fsycl %s -o %t.out -DSYCL_EXT_ONEAPI_MATRIX=1
+// RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 
diff --git a/SYCL/Matrix/element_wise_irreg_sum_rows.cpp b/SYCL/Matrix/element_wise_irreg_sum_rows.cpp
index bfd23a1716..73256abfd3 100644
--- a/SYCL/Matrix/element_wise_irreg_sum_rows.cpp
+++ b/SYCL/Matrix/element_wise_irreg_sum_rows.cpp
@@ -7,7 +7,7 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: matrix
 
-// RUN: %clangxx -fsycl %s -o %t.out -DSYCL_EXT_ONEAPI_MATRIX=1
+// RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 
diff --git a/SYCL/Matrix/element_wise_ops.cpp b/SYCL/Matrix/element_wise_ops.cpp
index 3b82fdff37..5f5d801d58 100644
--- a/SYCL/Matrix/element_wise_ops.cpp
+++ b/SYCL/Matrix/element_wise_ops.cpp
@@ -7,7 +7,7 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: matrix
 
-// RUN: %clangxx -fsycl %s -o %t.out -DSYCL_EXT_ONEAPI_MATRIX=1
+// RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 
diff --git a/SYCL/Matrix/element_wise_wi_marray.cpp b/SYCL/Matrix/element_wise_wi_marray.cpp
index ddab6e4414..8bce3fe880 100644
--- a/SYCL/Matrix/element_wise_wi_marray.cpp
+++ b/SYCL/Matrix/element_wise_wi_marray.cpp
@@ -7,7 +7,7 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: cuda
 
-// RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple -Xsycl-target-backend --cuda-gpu-arch=sm_80 -DSYCL_EXT_ONEAPI_MATRIX=3 %s -o %t.out
+// RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple -Xsycl-target-backend --cuda-gpu-arch=sm_80 -DSYCL_EXT_ONEAPI_MATRIX_VERSION=3 %s -o %t.out
 // RUN: %t.out
 
 #include <sycl/sycl.hpp>
diff --git a/SYCL/Matrix/elemwise_irreg_size_ops_bf16.cpp b/SYCL/Matrix/elemwise_irreg_size_ops_bf16.cpp
index 3d1fe393d3..ba0851f969 100644
--- a/SYCL/Matrix/elemwise_irreg_size_ops_bf16.cpp
+++ b/SYCL/Matrix/elemwise_irreg_size_ops_bf16.cpp
@@ -7,7 +7,7 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: matrix
 
-// RUN: %clangxx -fsycl %s -o %t.out -DSYCL_EXT_ONEAPI_MATRIX=1
+// RUN: %clangxx -fsycl %s -o %t.out
 // This test is for element wise operations when matrix size does not multiply
 // SG size. This corner case only applies to AMX. Also, it tests bf16 type.
 // only run this on AMX
diff --git a/SYCL/Matrix/joint_matrix_bf16.cpp b/SYCL/Matrix/joint_matrix_bf16.cpp
index c82903d16d..ef0c4ffa30 100644
--- a/SYCL/Matrix/joint_matrix_bf16.cpp
+++ b/SYCL/Matrix/joint_matrix_bf16.cpp
@@ -7,7 +7,7 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: matrix
 
-// RUN: %clangxx -fsycl %s -o %t.out -DSYCL_EXT_ONEAPI_MATRIX=1
+// RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 
diff --git a/SYCL/Matrix/joint_matrix_bfloat16.cpp b/SYCL/Matrix/joint_matrix_bfloat16.cpp
index e9d628e874..53f9063650 100644
--- a/SYCL/Matrix/joint_matrix_bfloat16.cpp
+++ b/SYCL/Matrix/joint_matrix_bfloat16.cpp
@@ -7,7 +7,7 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: matrix
 
-// RUN: %clangxx -fsycl %s -o %t.out -DSYCL_EXT_ONEAPI_MATRIX=1
+// RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 
diff --git a/SYCL/Matrix/joint_matrix_bfloat16_use.cpp b/SYCL/Matrix/joint_matrix_bfloat16_use.cpp
index da3bd8b1d8..893c00c963 100644
--- a/SYCL/Matrix/joint_matrix_bfloat16_use.cpp
+++ b/SYCL/Matrix/joint_matrix_bfloat16_use.cpp
@@ -7,7 +7,7 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: matrix
 
-// RUN: %clangxx -fsycl %s -o %t.out
+// RUN: %clangxx -fsycl %s -o %t.out -D-DSYCL_EXT_ONEAPI_MATRIX_VERSION=2
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 
diff --git a/SYCL/Matrix/joint_matrix_half.cpp b/SYCL/Matrix/joint_matrix_half.cpp
index d109b6f5d8..8f81ccef71 100644
--- a/SYCL/Matrix/joint_matrix_half.cpp
+++ b/SYCL/Matrix/joint_matrix_half.cpp
@@ -7,7 +7,7 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: matrix
 
-// RUN: %clangxx -fsycl %s -o %t.out -DSYCL_EXT_ONEAPI_MATRIX=1
+// RUN: %clangxx -fsycl %s -o %t.out
 // Only run on the GPU because half is not supported on AMX hardware
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 
diff --git a/SYCL/Matrix/joint_matrix_int8_vnni.cpp b/SYCL/Matrix/joint_matrix_int8_vnni.cpp
index 05f420184b..d7d617f5d6 100644
--- a/SYCL/Matrix/joint_matrix_int8_vnni.cpp
+++ b/SYCL/Matrix/joint_matrix_int8_vnni.cpp
@@ -7,7 +7,7 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: matrix
 
-// RUN: %clangxx -fsycl %s -o %t.out -DSYCL_EXT_ONEAPI_MATRIX=1
+// RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 
diff --git a/SYCL/Matrix/joint_matrix_query_default.cpp b/SYCL/Matrix/joint_matrix_query_default.cpp
index 1e2a990e55..4482ee3309 100644
--- a/SYCL/Matrix/joint_matrix_query_default.cpp
+++ b/SYCL/Matrix/joint_matrix_query_default.cpp
@@ -7,7 +7,7 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: matrix
 
-// RUN: %clangxx -fsycl %s -o %t.out -DSYCL_EXT_ONEAPI_MATRIX=1
+// RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 
 #include <iostream>
diff --git a/SYCL/Matrix/joint_matrix_ss_int8.cpp b/SYCL/Matrix/joint_matrix_ss_int8.cpp
index 9936184ad0..c3a6605be0 100644
--- a/SYCL/Matrix/joint_matrix_ss_int8.cpp
+++ b/SYCL/Matrix/joint_matrix_ss_int8.cpp
@@ -7,7 +7,7 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: matrix
 
-// RUN: %clangxx -fsycl %s -o %t.out -DSYCL_EXT_ONEAPI_MATRIX=1
+// RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 
diff --git a/SYCL/Matrix/joint_matrix_ss_int8_use.cpp b/SYCL/Matrix/joint_matrix_ss_int8_use.cpp
index a8b649db71..dc9da94807 100644
--- a/SYCL/Matrix/joint_matrix_ss_int8_use.cpp
+++ b/SYCL/Matrix/joint_matrix_ss_int8_use.cpp
@@ -7,7 +7,7 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: matrix
 
-// RUN: %clangxx -fsycl %s -o %t.out
+// RUN: %clangxx -fsycl %s -o %t.out -DSYCL_EXT_ONEAPI_MATRIX=2
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 
diff --git a/SYCL/Matrix/joint_matrix_su_int8.cpp b/SYCL/Matrix/joint_matrix_su_int8.cpp
index 332a083078..bfae6682a9 100644
--- a/SYCL/Matrix/joint_matrix_su_int8.cpp
+++ b/SYCL/Matrix/joint_matrix_su_int8.cpp
@@ -7,7 +7,7 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: matrix
 
-// RUN: %clangxx -fsycl %s -o %t.out -DSYCL_EXT_ONEAPI_MATRIX=1
+// RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 
diff --git a/SYCL/Matrix/joint_matrix_tensorcore.cpp b/SYCL/Matrix/joint_matrix_tensorcore.cpp
index 8582ef955f..72483bf282 100644
--- a/SYCL/Matrix/joint_matrix_tensorcore.cpp
+++ b/SYCL/Matrix/joint_matrix_tensorcore.cpp
@@ -1,5 +1,5 @@
 // REQUIRES: cuda
-// RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple -Xsycl-target-backend --cuda-gpu-arch=sm_80 -DSYCL_EXT_ONEAPI_MATRIX=3 %s -o %t.out
+// RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple -Xsycl-target-backend --cuda-gpu-arch=sm_80 -DSYCL_EXT_ONEAPI_MATRIX_VERSION=3 %s -o %t.out
 // RUN: %t.out
 //
 // Specifying the sm version via the --cuda-gpu-arch flag is necessary
diff --git a/SYCL/Matrix/joint_matrix_us_int8.cpp b/SYCL/Matrix/joint_matrix_us_int8.cpp
index 4b9336a6b5..6780df16ca 100644
--- a/SYCL/Matrix/joint_matrix_us_int8.cpp
+++ b/SYCL/Matrix/joint_matrix_us_int8.cpp
@@ -7,7 +7,7 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: matrix
 
-// RUN: %clangxx -fsycl %s -o %t.out -DSYCL_EXT_ONEAPI_MATRIX=1
+// RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 
diff --git a/SYCL/Matrix/joint_matrix_uu_int8.cpp b/SYCL/Matrix/joint_matrix_uu_int8.cpp
index 5ff02f37e8..cadfad28ff 100644
--- a/SYCL/Matrix/joint_matrix_uu_int8.cpp
+++ b/SYCL/Matrix/joint_matrix_uu_int8.cpp
@@ -7,7 +7,7 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: matrix
 
-// RUN: %clangxx -fsycl %s -o %t.out -DSYCL_EXT_ONEAPI_MATRIX=1
+// RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out