Skip to content

[SYCL][UR][L0 v2] import host ptr when SYCL_USM_HOSTPTR_IMPORT=1 and fix mapHostPtr #19297

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 3 commits into from
Jul 17, 2025
Merged

[SYCL][UR][L0 v2] import host ptr when SYCL_USM_HOSTPTR_IMPORT=1 and fix mapHostPtr #19297

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
a4692a3
[SYCL][UR][L0 v2] import host ptr when SYCL_USM_HOSTPTR_IMPORT=1
igchor Jul 3, 2025
8571499
[SYCL][UR][L0 v2] fix ur_discrete_buffer_handle_t::mapHostPtr
igchor Jul 3, 2025
fca47e6
fixes
igchor Jul 7, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
43 changes: 27 additions & 16 deletions unified-runtime/source/adapters/level_zero/v2/memory.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -55,14 +55,11 @@ void ur_usm_handle_t::unmapHostPtr(void * /*pMappedPtr*/,

ur_integrated_buffer_handle_t::ur_integrated_buffer_handle_t(
ur_context_handle_t hContext, void *hostPtr, size_t size,
host_ptr_action_t hostPtrAction, device_access_mode_t accessMode)
device_access_mode_t accessMode)
: ur_mem_buffer_t(hContext, size, accessMode) {
bool hostPtrImported = false;
if (hostPtrAction == host_ptr_action_t::import) {
hostPtrImported =
maybeImportUSM(hContext->getPlatform()->ZeDriverHandleExpTranslated,
hContext->getZeHandle(), hostPtr, size);
}
bool hostPtrImported =
maybeImportUSM(hContext->getPlatform()->ZeDriverHandleExpTranslated,
hContext->getZeHandle(), hostPtr, size);

if (hostPtrImported) {
this->ptr = usm_unique_ptr_t(hostPtr, [hContext](void *ptr) {
Expand Down Expand Up @@ -201,8 +198,23 @@ ur_discrete_buffer_handle_t::ur_discrete_buffer_handle_t(
device_access_mode_t accessMode)
: ur_mem_buffer_t(hContext, size, accessMode),
deviceAllocations(hContext->getPlatform()->getNumDevices()),
activeAllocationDevice(nullptr), mapToPtr(hostPtr), hostAllocations() {
activeAllocationDevice(nullptr), mapToPtr(nullptr, nullptr),
hostAllocations() {
if (hostPtr) {
// Try importing the pointer to speed up memory copies for map/unmap
bool hostPtrImported =
maybeImportUSM(hContext->getPlatform()->ZeDriverHandleExpTranslated,
hContext->getZeHandle(), hostPtr, size);

if (hostPtrImported) {
mapToPtr = usm_unique_ptr_t(hostPtr, [hContext](void *ptr) {
ZeUSMImport.doZeUSMRelease(
hContext->getPlatform()->ZeDriverHandleExpTranslated, ptr);
});
} else {
mapToPtr = usm_unique_ptr_t(hostPtr, [](void *) {});
}

auto initialDevice = hContext->getDevices()[0];
UR_CALL_THROWS(migrateBufferTo(initialDevice, hostPtr, size));
}
Expand Down Expand Up @@ -305,18 +317,18 @@ void *ur_discrete_buffer_handle_t::mapHostPtr(ur_map_flags_t flags,
TRACK_SCOPE_LATENCY("ur_discrete_buffer_handle_t::mapHostPtr");
// TODO: use async alloc?

void *ptr = mapToPtr;
void *ptr = mapToPtr.get();
if (!ptr) {
UR_CALL_THROWS(hContext->getDefaultUSMPool()->allocate(
hContext, nullptr, nullptr, UR_USM_TYPE_HOST, size, &ptr));
}

usm_unique_ptr_t mappedPtr =
usm_unique_ptr_t(ptr, [ownsAlloc = bool(mapToPtr), this](void *p) {
usm_unique_ptr_t(ptr, [ownsAlloc = !bool(mapToPtr), this](void *p) {
if (ownsAlloc) {
auto ret = hContext->getDefaultUSMPool()->free(p);
if (ret != UR_RESULT_SUCCESS) {
UR_LOG(ERR, "Failed to mapped memory: {}", ret);
UR_LOG(ERR, "Failed to free mapped memory: {}", ret);
}
}
});
Expand Down Expand Up @@ -542,15 +554,14 @@ ur_result_t urMemBufferCreate(ur_context_handle_t hContext,
}

void *hostPtr = pProperties ? pProperties->pHost : nullptr;
bool useHostPointer = flags & UR_MEM_FLAG_USE_HOST_POINTER;
auto accessMode = ur_mem_buffer_t::getDeviceAccessMode(flags);

assert(useHostPointer == bool(hostPtr));

if (useHostBuffer(hContext)) {
auto hostPtrAction =
flags & UR_MEM_FLAG_USE_HOST_POINTER
? ur_integrated_buffer_handle_t::host_ptr_action_t::import
: ur_integrated_buffer_handle_t::host_ptr_action_t::copy;
*phBuffer = ur_mem_handle_t_::create<ur_integrated_buffer_handle_t>(
hContext, hostPtr, size, hostPtrAction, accessMode);
hContext, hostPtr, size, accessMode);
} else {
*phBuffer = ur_mem_handle_t_::create<ur_discrete_buffer_handle_t>(
hContext, hostPtr, size, accessMode);
Expand Down
7 changes: 2 additions & 5 deletions unified-runtime/source/adapters/level_zero/v2/memory.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -89,11 +89,8 @@ struct ur_usm_handle_t : ur_mem_buffer_t {
// For integrated devices the buffer has been allocated in host memory
// and can be accessed by the device without copying.
struct ur_integrated_buffer_handle_t : ur_mem_buffer_t {
enum class host_ptr_action_t { import, copy };

ur_integrated_buffer_handle_t(ur_context_handle_t hContext, void *hostPtr,
size_t size, host_ptr_action_t useHostPtr,
device_access_mode_t accesMode);
size_t size, device_access_mode_t accesMode);

ur_integrated_buffer_handle_t(ur_context_handle_t hContext, void *hostPtr,
size_t size, device_access_mode_t accesMode,
Expand Down Expand Up @@ -165,7 +162,7 @@ struct ur_discrete_buffer_handle_t : ur_mem_buffer_t {
void *writeBackPtr = nullptr;

// If not null, mapHostPtr should map memory to this ptr
void *mapToPtr = nullptr;
usm_unique_ptr_t mapToPtr;

std::vector<host_allocation_desc_t> hostAllocations;

Expand Down
Loading