diff --git a/level_zero/core/source/kernel/kernel_imp.cpp b/level_zero/core/source/kernel/kernel_imp.cpp
index b749a7531c..0c68046f22 100644
--- a/level_zero/core/source/kernel/kernel_imp.cpp
+++ b/level_zero/core/source/kernel/kernel_imp.cpp
@@ -398,13 +398,22 @@ ze_result_t KernelImp::setGroupSize(uint32_t groupSizeX, uint32_t groupSizeY,
 
         if (numChannels > 0) {
             UNRECOVERABLE_IF(3 != numChannels);
+
+            std::array<uint8_t, 3> walkOrder{0, 1, 2};
+            if (kernelDescriptor.kernelAttributes.flags.requiresWorkgroupWalkOrder) {
+                walkOrder = {
+                    kernelDescriptor.kernelAttributes.workgroupWalkOrder[0],
+                    kernelDescriptor.kernelAttributes.workgroupWalkOrder[1],
+                    kernelDescriptor.kernelAttributes.workgroupWalkOrder[2]};
+            }
+
             NEO::generateLocalIDs(
                 perThreadDataForWholeThreadGroup,
                 static_cast<uint16_t>(simdSize),
                 std::array<uint16_t, 3>{{static_cast<uint16_t>(groupSizeX),
                                          static_cast<uint16_t>(groupSizeY),
                                          static_cast<uint16_t>(groupSizeZ)}},
-                std::array<uint8_t, 3>{{0, 1, 2}},
+                walkOrder,
                 false, grfSize, grfCount, rootDeviceEnvironment);
         }
 
diff --git a/level_zero/core/test/unit_tests/sources/kernel/test_kernel_2.cpp b/level_zero/core/test/unit_tests/sources/kernel/test_kernel_2.cpp
index 4a2b1209e6..c12a9a7f98 100644
--- a/level_zero/core/test/unit_tests/sources/kernel/test_kernel_2.cpp
+++ b/level_zero/core/test/unit_tests/sources/kernel/test_kernel_2.cpp
@@ -8,6 +8,7 @@
 #include "shared/source/helpers/aligned_memory.h"
 #include "shared/source/helpers/basic_math.h"
 #include "shared/source/helpers/gfx_core_helper.h"
+#include "shared/source/helpers/local_id_gen.h"
 #include "shared/source/helpers/simd_helper.h"
 #include "shared/test/common/helpers/raii_gfx_core_helper.h"
 #include "shared/test/common/mocks/mock_bindless_heaps_helper.h"
@@ -831,5 +832,49 @@ TEST_F(KernelImpTest, GivenGroupSizeRequiresSwLocalIdsGenerationWhenNextGroupSiz
     EXPECT_EQ(0u, kernel.getPerThreadDataSize());
 }
 
+TEST_F(KernelImpTest, GivenGroupSizeRequiresSwLocalIdsGenerationWhenKernelSpecifiesRequiredWalkOrderThenUseCorrectOrderToGenerateLocalIds) {
+    Mock<Module> module(device, nullptr);
+    Mock<::L0::KernelImp> kernel;
+    kernel.module = &module;
+
+    auto grfSize = device->getHwInfo().capabilityTable.grfSize;
+
+    WhiteBox<::L0::KernelImmutableData> kernelInfo = {};
+    NEO::KernelDescriptor descriptor;
+    kernelInfo.kernelDescriptor = &descriptor;
+    kernelInfo.kernelDescriptor->kernelAttributes.numLocalIdChannels = 3;
+    kernelInfo.kernelDescriptor->kernelAttributes.numGrfRequired = GrfConfig::defaultGrfNumber;
+    kernelInfo.kernelDescriptor->kernelAttributes.flags.requiresWorkgroupWalkOrder = true;
+    kernelInfo.kernelDescriptor->kernelAttributes.workgroupWalkOrder[0] = 2;
+    kernelInfo.kernelDescriptor->kernelAttributes.workgroupWalkOrder[1] = 1;
+    kernelInfo.kernelDescriptor->kernelAttributes.workgroupWalkOrder[2] = 0;
+    kernelInfo.kernelDescriptor->kernelAttributes.simdSize = 32;
+
+    kernel.kernelImmData = &kernelInfo;
+
+    kernel.enableForcingOfGenerateLocalIdByHw = true;
+    kernel.forceGenerateLocalIdByHw = false;
+
+    kernel.KernelImp::setGroupSize(12, 12, 1);
+
+    uint32_t perThreadSizeNeeded = kernel.getPerThreadDataSizeForWholeThreadGroup();
+    auto testPerThreadDataBuffer = static_cast<uint8_t *>(alignedMalloc(perThreadSizeNeeded, 32));
+
+    std::array<uint8_t, 3> walkOrder{2, 1, 0};
+
+    NEO::generateLocalIDs(
+        testPerThreadDataBuffer,
+        static_cast<uint16_t>(32),
+        std::array<uint16_t, 3>{{static_cast<uint16_t>(12),
+                                 static_cast<uint16_t>(12),
+                                 static_cast<uint16_t>(1)}},
+        walkOrder,
+        false, grfSize, GrfConfig::defaultGrfNumber, device->getNEODevice()->getRootDeviceEnvironment());
+
+    EXPECT_EQ(0, memcmp(testPerThreadDataBuffer, kernel.KernelImp::getPerThreadData(), perThreadSizeNeeded));
+
+    alignedFree(testPerThreadDataBuffer);
+}
+
 } // namespace ult
 } // namespace L0