diff --git a/projects/rocr-runtime/runtime/hsa-runtime/core/driver/xdna/amd_xdna_driver.cpp b/projects/rocr-runtime/runtime/hsa-runtime/core/driver/xdna/amd_xdna_driver.cpp
index 40afc148664f..4f8e71e6cc9e 100644
--- a/projects/rocr-runtime/runtime/hsa-runtime/core/driver/xdna/amd_xdna_driver.cpp
+++ b/projects/rocr-runtime/runtime/hsa-runtime/core/driver/xdna/amd_xdna_driver.cpp
@@ -1273,13 +1273,6 @@ hsa_status_t XdnaDriver::AllocateMemory(const core::MemoryRegion& mem_region,
   if (use_bo_share) {
     create_bo_args.type = AMDXDNA_BO_SHARE;
   } else {
-    // While this is already checked in MemoryRegion::AllocateImpl, the max size is
-    // MemoryRegion::max_sysmem_alloc_size_ for HSA_HEAPTYPE_DEVICE_SVM which is incorrect
-    // for dev heap.
-    if (size > dev_heap_size) {
-      return HSA_STATUS_ERROR_INVALID_ALLOCATION;
-    }
-
     create_bo_args.type = AMDXDNA_BO_DEV;
   }
 
diff --git a/projects/rocr-runtime/runtime/hsa-runtime/core/runtime/amd_aie_agent.cpp b/projects/rocr-runtime/runtime/hsa-runtime/core/runtime/amd_aie_agent.cpp
index 09886e73ea9f..2ac809e62dfe 100644
--- a/projects/rocr-runtime/runtime/hsa-runtime/core/runtime/amd_aie_agent.cpp
+++ b/projects/rocr-runtime/runtime/hsa-runtime/core/runtime/amd_aie_agent.cpp
@@ -332,8 +332,8 @@ void AieAgent::InitRegionList() {
   // We use HSA_HEAPTYPE_DEVICE_SVM so that the recommended
   // HSA_AMD_MEMORY_POOL_INFO_RUNTIME_ALLOC_REC_GRANULE is 0. We can use that to determine if a pool
   // is regular pool or dev heap.
-  // The system will report the max allocatable size as MemoryRegion::max_sysmem_alloc_size_, but
-  // this is incorrect. The pool is exactly XdnaDriver::GetDevHeapByteSize() bytes.
+  // The pool reports its max allocatable size as MemoryRegion::max_single_alloc_size_, which is
+  // exactly XdnaDriver::GetDevHeapByteSize() bytes.
   HsaMemoryProperties dev_mem_props = {};
   dev_mem_props.HeapType = HSA_HEAPTYPE_DEVICE_SVM;
   dev_mem_props.SizeInBytes = XdnaDriver::GetDevHeapByteSize();
diff --git a/projects/rocr-runtime/runtime/hsa-runtime/core/runtime/amd_memory_region.cpp b/projects/rocr-runtime/runtime/hsa-runtime/core/runtime/amd_memory_region.cpp
index 21dba2e16b69..286eea71dc31 100644
--- a/projects/rocr-runtime/runtime/hsa-runtime/core/runtime/amd_memory_region.cpp
+++ b/projects/rocr-runtime/runtime/hsa-runtime/core/runtime/amd_memory_region.cpp
@@ -152,7 +152,9 @@ hsa_status_t MemoryRegion::AllocateImpl(size_t& size, AllocateFlags alloc_flags,
   // Skip the per-region cap on Windows/DXG so over-commit requests can
   // reach WDDM; system memory still enforces the cap.
   const bool is_windxg = core::Runtime::runtime_singleton_->thunkLoader()->IsDXG();
-  if (IsSystem() && (size > max_sysmem_alloc_size_)) {
+  const size_t max_alloc_size =
+      IsDeviceSVM() ? max_single_alloc_size_ : max_sysmem_alloc_size_;
+  if (IsSystem() && (size > max_alloc_size)) {
     return HSA_STATUS_ERROR_INVALID_ALLOCATION;
   }
   if (!IsSystem() && !is_windxg && (size > max_single_alloc_size_)) {
@@ -231,9 +233,9 @@ hsa_status_t MemoryRegion::GetInfo(hsa_region_info_t attribute,
     case HSA_REGION_INFO_ALLOC_MAX_SIZE:
       switch (mem_props_.HeapType) {
         case HSA_HEAPTYPE_SYSTEM:
-        case HSA_HEAPTYPE_DEVICE_SVM:
           *((size_t*)value) = max_sysmem_alloc_size_;
           break;
+        case HSA_HEAPTYPE_DEVICE_SVM:
         case HSA_HEAPTYPE_FRAME_BUFFER_PRIVATE:
         case HSA_HEAPTYPE_FRAME_BUFFER_PUBLIC:
         case HSA_HEAPTYPE_GPU_SCRATCH: