[SYCL][UR][L0 v2] Use discrete buffer implementation for integrated GPUs (#18354)

Author: igchor

unified-runtime/source/adapters/level_zero/v2/memory.cpp

@@ -114,11 +114,16 @@ void *ur_integrated_buffer_handle_t::getDevicePtr(
 void *ur_integrated_buffer_handle_t::mapHostPtr(
     ur_map_flags_t /*flags*/, size_t offset, size_t /*size*/,
     std::function<void(void *src, void *dst, size_t)> /*migrate*/) {
+  // TODO: if writeBackPtr is set, we should map to that pointer
+  // because that's what SYCL expects, SYCL will attempt to call free
+  // on the resulting pointer leading to double free with the current
+  // implementation. Investigate the SYCL implementation.
   return ur_cast<char *>(ptr.get()) + offset;
 }
 
 void ur_integrated_buffer_handle_t::unmapHostPtr(
     void * /*pMappedPtr*/, std::function<void(void *src, void *dst, size_t)>) {
+  // TODO: if writeBackPtr is set, we should copy the data back
   /* nop */
 }
 
@@ -371,15 +376,19 @@ void ur_shared_buffer_handle_t::unmapHostPtr(
   // nop
 }
 
-static bool useHostBuffer(ur_context_handle_t hContext) {
+static bool useHostBuffer(ur_context_handle_t /* hContext */) {
   // We treat integrated devices (physical memory shared with the CPU)
   // differently from discrete devices (those with distinct memories).
   // For integrated devices, allocating the buffer in the host memory
   // enables automatic access from the device, and makes copying
   // unnecessary in the map/unmap operations. This improves performance.
-  return hContext->getDevices().size() == 1 &&
-         hContext->getDevices()[0]->ZeDeviceProperties->flags &
-             ZE_DEVICE_PROPERTY_FLAG_INTEGRATED;
+
+  // TODO: fix integrated buffer implementation
+  return false;
+
+  // return hContext->getDevices().size() == 1 &&
+  //        hContext->getDevices()[0]->ZeDeviceProperties->flags &
+  //            ZE_DEVICE_PROPERTY_FLAG_INTEGRATED;
 }
 
 ur_mem_sub_buffer_t::ur_mem_sub_buffer_t(ur_mem_handle_t hParent, size_t offset,