diff --git a/shared/source/command_container/command_encoder_xe_hpc_core_and_later.inl b/shared/source/command_container/command_encoder_xe_hpc_core_and_later.inl
index 1a62c3334d..dd3da591cf 100644
--- a/shared/source/command_container/command_encoder_xe_hpc_core_and_later.inl
+++ b/shared/source/command_container/command_encoder_xe_hpc_core_and_later.inl
@@ -7,6 +7,7 @@
 
 #include "shared/source/command_container/command_encoder.h"
 #include "shared/source/command_stream/linear_stream.h"
+#include "shared/source/gmm_helper/cache_settings_helper.h"
 #include "shared/source/memory_manager/graphics_allocation.h"
 #include "shared/source/utilities/lookup_array.h"
 
@@ -56,4 +57,51 @@ void EncodeDispatchKernel<Family>::programBarrierEnable(InterfaceDescriptorType
     interfaceDescriptor.setNumberOfBarriers(numBarriers);
 }
 
+template <typename Family>
+void EncodeMemoryPrefetch<Family>::programMemoryPrefetch(LinearStream &commandStream, const GraphicsAllocation &graphicsAllocation, uint32_t size, size_t offset, const RootDeviceEnvironment &rootDeviceEnvironment) {
+    using STATE_PREFETCH = typename Family::STATE_PREFETCH;
+
+    auto &hwInfo = *rootDeviceEnvironment.getHardwareInfo();
+    auto &productHelper = rootDeviceEnvironment.getHelper<ProductHelper>();
+
+    bool prefetch = productHelper.allowMemoryPrefetch(hwInfo);
+
+    if (!prefetch) {
+        return;
+    }
+
+    auto usage = CacheSettingsHelper::getGmmUsageType(graphicsAllocation.getAllocationType(), false, productHelper, &hwInfo);
+    uint32_t mocs = rootDeviceEnvironment.getGmmHelper()->getMOCS(usage);
+
+    uint64_t gpuVa = graphicsAllocation.getGpuAddress() + offset;
+
+    while (size > 0) {
+        uint32_t sizeInBytesToPrefetch = std::min(alignUp(size, MemoryConstants::cacheLineSize),
+                                                  static_cast<uint32_t>(MemoryConstants::pageSize64k));
+
+        uint32_t prefetchSize = sizeInBytesToPrefetch / MemoryConstants::cacheLineSize;
+
+        auto statePrefetch = commandStream.getSpaceForCmd<STATE_PREFETCH>();
+        STATE_PREFETCH cmd = Family::cmdInitStatePrefetch;
+
+        cmd.setAddress(gpuVa);
+        cmd.setPrefetchSize(prefetchSize);
+        cmd.setMemoryObjectControlState(mocs);
+        cmd.setKernelInstructionPrefetch(GraphicsAllocation::isIsaAllocationType(graphicsAllocation.getAllocationType()));
+
+        if (debugManager.flags.ForceCsStallForStatePrefetch.get() == 1) {
+            cmd.setParserStall(true);
+        }
+
+        *statePrefetch = cmd;
+
+        if (sizeInBytesToPrefetch > size) {
+            break;
+        }
+
+        gpuVa += sizeInBytesToPrefetch;
+        size -= sizeInBytesToPrefetch;
+    }
+}
+
 } // namespace NEO
\ No newline at end of file
diff --git a/shared/source/generated/xe_hpc_core/hw_cmds_generated_xe_hpc_core.inl b/shared/source/generated/xe_hpc_core/hw_cmds_generated_xe_hpc_core.inl
index 1c055823e7..c54d3bc6c0 100644
--- a/shared/source/generated/xe_hpc_core/hw_cmds_generated_xe_hpc_core.inl
+++ b/shared/source/generated/xe_hpc_core/hw_cmds_generated_xe_hpc_core.inl
@@ -7178,10 +7178,10 @@ struct STATE_PREFETCH {
     }
 
     inline void setPrefetchSize(const uint32_t value) {
-        TheStructure.Common.PrefetchSize = value;
+        TheStructure.Common.PrefetchSize = value - 1; // patched
     }
     inline uint32_t getPrefetchSize() const {
-        return TheStructure.Common.PrefetchSize;
+        return TheStructure.Common.PrefetchSize + 1; // patched
     }
 
     inline void setKernelInstructionPrefetch(const bool value) {
diff --git a/shared/source/xe2_hpg_core/command_encoder_xe2_hpg_core.cpp b/shared/source/xe2_hpg_core/command_encoder_xe2_hpg_core.cpp
index 1bc761e516..15698028ce 100644
--- a/shared/source/xe2_hpg_core/command_encoder_xe2_hpg_core.cpp
+++ b/shared/source/xe2_hpg_core/command_encoder_xe2_hpg_core.cpp
@@ -16,7 +16,6 @@
 #include "shared/source/command_container/command_encoder_xe_hpc_core_and_later.inl"
 #include "shared/source/command_container/command_encoder_xehp_and_later.inl"
 #include "shared/source/command_stream/stream_properties.h"
-#include "shared/source/gmm_helper/cache_settings_helper.h"
 #include "shared/source/helpers/constants.h"
 #include "shared/source/kernel/grf_config.h"
 #include "shared/source/release_helper/release_helper.h"
@@ -86,52 +85,6 @@ void EncodeComputeMode<Family>::programComputeModeCommand(LinearStream &csr, Sta
     *buffer = stateComputeMode;
 }
 
-template <>
-void EncodeMemoryPrefetch<Family>::programMemoryPrefetch(LinearStream &commandStream, const GraphicsAllocation &graphicsAllocation, uint32_t size, size_t offset, const RootDeviceEnvironment &rootDeviceEnvironment) {
-    using STATE_PREFETCH = typename Family::STATE_PREFETCH;
-
-    auto &hwInfo = *rootDeviceEnvironment.getHardwareInfo();
-    auto &productHelper = rootDeviceEnvironment.getHelper<ProductHelper>();
-    bool prefetch = productHelper.allowMemoryPrefetch(hwInfo);
-
-    if (!prefetch) {
-        return;
-    }
-
-    auto usage = CacheSettingsHelper::getGmmUsageType(graphicsAllocation.getAllocationType(), false, productHelper, &hwInfo);
-    uint32_t mocs = rootDeviceEnvironment.getGmmHelper()->getMOCS(usage);
-
-    uint64_t gpuVa = graphicsAllocation.getGpuAddress() + offset;
-
-    while (size > 0) {
-        uint32_t sizeInBytesToPrefetch = std::min(alignUp(size, MemoryConstants::cacheLineSize),
-                                                  static_cast<uint32_t>(MemoryConstants::pageSize64k));
-
-        uint32_t prefetchSize = sizeInBytesToPrefetch / MemoryConstants::cacheLineSize;
-
-        auto statePrefetch = commandStream.getSpaceForCmd<STATE_PREFETCH>();
-        STATE_PREFETCH cmd = Family::cmdInitStatePrefetch;
-
-        cmd.setAddress(gpuVa);
-        cmd.setPrefetchSize(prefetchSize);
-        cmd.setMemoryObjectControlState(mocs);
-        cmd.setKernelInstructionPrefetch(GraphicsAllocation::isIsaAllocationType(graphicsAllocation.getAllocationType()));
-
-        if (debugManager.flags.ForceCsStallForStatePrefetch.get() == 1) {
-            cmd.setParserStall(true);
-        }
-
-        *statePrefetch = cmd;
-
-        if (sizeInBytesToPrefetch > size) {
-            break;
-        }
-
-        gpuVa += sizeInBytesToPrefetch;
-        size -= sizeInBytesToPrefetch;
-    }
-}
-
 template <>
 void EncodeSurfaceState<Family>::setAuxParamsForMCSCCS(R_SURFACE_STATE *surfaceState, const ReleaseHelper *releaseHelper) {
     if (releaseHelper && releaseHelper->isAuxSurfaceModeOverrideRequired())
diff --git a/shared/source/xe3_core/command_encoder_xe3_core.cpp b/shared/source/xe3_core/command_encoder_xe3_core.cpp
index d8d1f3a1a5..3ede5261fe 100644
--- a/shared/source/xe3_core/command_encoder_xe3_core.cpp
+++ b/shared/source/xe3_core/command_encoder_xe3_core.cpp
@@ -15,7 +15,6 @@
 #include "shared/source/command_container/command_encoder_xe_hpc_core_and_later.inl"
 #include "shared/source/command_container/command_encoder_xehp_and_later.inl"
 #include "shared/source/command_stream/stream_properties.h"
-#include "shared/source/gmm_helper/cache_settings_helper.h"
 #include "shared/source/helpers/cache_policy.h"
 #include "shared/source/helpers/constants.h"
 #include "shared/source/kernel/grf_config.h"
@@ -72,53 +71,6 @@ void EncodeComputeMode<Family>::programComputeModeCommand(LinearStream &csr, Sta
     *buffer = stateComputeMode;
 }
 
-template <>
-void EncodeMemoryPrefetch<Family>::programMemoryPrefetch(LinearStream &commandStream, const GraphicsAllocation &graphicsAllocation, uint32_t size, size_t offset, const RootDeviceEnvironment &rootDeviceEnvironment) {
-    using STATE_PREFETCH = typename Family::STATE_PREFETCH;
-
-    auto &hwInfo = *rootDeviceEnvironment.getHardwareInfo();
-    auto &productHelper = rootDeviceEnvironment.getHelper<ProductHelper>();
-
-    bool prefetch = productHelper.allowMemoryPrefetch(hwInfo);
-
-    if (!prefetch) {
-        return;
-    }
-
-    auto usage = CacheSettingsHelper::getGmmUsageType(graphicsAllocation.getAllocationType(), false, productHelper, &hwInfo);
-    uint32_t mocs = rootDeviceEnvironment.getGmmHelper()->getMOCS(usage);
-
-    uint64_t gpuVa = graphicsAllocation.getGpuAddress() + offset;
-
-    while (size > 0) {
-        uint32_t sizeInBytesToPrefetch = std::min(alignUp(size, MemoryConstants::cacheLineSize),
-                                                  static_cast<uint32_t>(MemoryConstants::pageSize64k));
-
-        uint32_t prefetchSize = sizeInBytesToPrefetch / MemoryConstants::cacheLineSize;
-
-        auto statePrefetch = commandStream.getSpaceForCmd<STATE_PREFETCH>();
-        STATE_PREFETCH cmd = Family::cmdInitStatePrefetch;
-
-        cmd.setAddress(gpuVa);
-        cmd.setPrefetchSize(prefetchSize);
-        cmd.setMemoryObjectControlState(mocs);
-        cmd.setKernelInstructionPrefetch(GraphicsAllocation::isIsaAllocationType(graphicsAllocation.getAllocationType()));
-
-        if (debugManager.flags.ForceCsStallForStatePrefetch.get() == 1) {
-            cmd.setParserStall(true);
-        }
-
-        *statePrefetch = cmd;
-
-        if (sizeInBytesToPrefetch > size) {
-            break;
-        }
-
-        gpuVa += sizeInBytesToPrefetch;
-        size -= sizeInBytesToPrefetch;
-    }
-}
-
 template <>
 void EncodeSurfaceState<Family>::setAuxParamsForMCSCCS(R_SURFACE_STATE *surfaceState, const ReleaseHelper *releaseHelper) {
     surfaceState->setAuxiliarySurfaceMode(AUXILIARY_SURFACE_MODE::AUXILIARY_SURFACE_MODE_AUX_MCS);
diff --git a/shared/source/xe_hpc_core/command_encoder_xe_hpc_core.cpp b/shared/source/xe_hpc_core/command_encoder_xe_hpc_core.cpp
index dc6ee7ab1f..84c4001f6a 100644
--- a/shared/source/xe_hpc_core/command_encoder_xe_hpc_core.cpp
+++ b/shared/source/xe_hpc_core/command_encoder_xe_hpc_core.cpp
@@ -16,7 +16,6 @@
 #include "shared/source/command_container/command_encoder_xe_hpg_core_and_xe_hpc.inl"
 #include "shared/source/command_container/command_encoder_xehp_and_later.inl"
 #include "shared/source/command_stream/stream_properties.h"
-#include "shared/source/gmm_helper/cache_settings_helper.h"
 #include "shared/source/helpers/constants.h"
 #include "shared/source/kernel/grf_config.h"
 #include "shared/source/release_helper/release_helper.h"
@@ -74,54 +73,6 @@ void EncodeComputeMode<Family>::programComputeModeCommand(LinearStream &csr, Sta
     *buffer = stateComputeMode;
 }
 
-template <>
-void EncodeMemoryPrefetch<Family>::programMemoryPrefetch(LinearStream &commandStream, const GraphicsAllocation &graphicsAllocation, uint32_t size, size_t offset, const RootDeviceEnvironment &rootDeviceEnvironment) {
-    using STATE_PREFETCH = typename Family::STATE_PREFETCH;
-
-    auto &hwInfo = *rootDeviceEnvironment.getHardwareInfo();
-    auto &productHelper = rootDeviceEnvironment.getHelper<ProductHelper>();
-
-    bool prefetch = productHelper.allowMemoryPrefetch(hwInfo);
-
-    if (!prefetch) {
-        return;
-    }
-
-    auto usage = CacheSettingsHelper::getGmmUsageType(graphicsAllocation.getAllocationType(), false, productHelper, &hwInfo);
-    uint32_t mocs = rootDeviceEnvironment.getGmmHelper()->getMOCS(usage);
-
-    uint64_t gpuVa = graphicsAllocation.getGpuAddress() + offset;
-
-    while (size > 0) {
-        uint32_t sizeInBytesToPrefetch = std::min(alignUp(size, MemoryConstants::cacheLineSize),
-                                                  static_cast<uint32_t>(MemoryConstants::pageSize64k));
-
-        // zero based cacheline count (0 == 1 cacheline)
-        uint32_t prefetchSize = (sizeInBytesToPrefetch / MemoryConstants::cacheLineSize) - 1;
-
-        auto statePrefetch = commandStream.getSpaceForCmd<STATE_PREFETCH>();
-        STATE_PREFETCH cmd = Family::cmdInitStatePrefetch;
-
-        cmd.setAddress(gpuVa);
-        cmd.setPrefetchSize(prefetchSize);
-        cmd.setMemoryObjectControlState(mocs);
-        cmd.setKernelInstructionPrefetch(GraphicsAllocation::isIsaAllocationType(graphicsAllocation.getAllocationType()));
-
-        if (debugManager.flags.ForceCsStallForStatePrefetch.get() == 1) {
-            cmd.setParserStall(true);
-        }
-
-        *statePrefetch = cmd;
-
-        if (sizeInBytesToPrefetch > size) {
-            break;
-        }
-
-        gpuVa += sizeInBytesToPrefetch;
-        size -= sizeInBytesToPrefetch;
-    }
-}
-
 template <>
 size_t EncodeMemoryPrefetch<Family>::getSizeForMemoryPrefetch(size_t size, const RootDeviceEnvironment &rootDeviceEnvironment) {
     if (debugManager.flags.EnableMemoryPrefetch.get() == 0) {
diff --git a/shared/test/unit_test/xe_hpc_core/test_encode_xe_hpc_core.cpp b/shared/test/unit_test/xe_hpc_core/test_encode_xe_hpc_core.cpp
index 211537b889..6dce58f632 100644
--- a/shared/test/unit_test/xe_hpc_core/test_encode_xe_hpc_core.cpp
+++ b/shared/test/unit_test/xe_hpc_core/test_encode_xe_hpc_core.cpp
@@ -118,7 +118,7 @@ XE_HPC_CORETEST_F(CommandEncodeXeHpcCoreTest, givenDebugVariableSetwhenProgramin
         EXPECT_EQ(sizeof(STATE_PREFETCH) * expectedCmdsCount, linearStream.getUsed());
 
         for (uint32_t i = 0; i < expectedCmdsCount; i++) {
-            uint32_t programmedSize = (statePrefetchCmd[i].getPrefetchSize() + 1) * MemoryConstants::cacheLineSize;
+            uint32_t programmedSize = statePrefetchCmd[i].getPrefetchSize() * MemoryConstants::cacheLineSize;
 
             EXPECT_EQ(statePrefetchCmd[i].getAddress(), gpuVa + (i * MemoryConstants::pageSize64k));
             EXPECT_FALSE(statePrefetchCmd[i].getKernelInstructionPrefetch());