Skip to content

[UR][SYCL] Add support for zeCommandListAppendLaunchKernelWithArguments() #20316

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

Draft
wants to merge 4 commits into
base: sycl
Choose a base branch
from
+3,596 −448
Draft

[UR][SYCL] Add support for zeCommandListAppendLaunchKernelWithArguments() #20316

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
0df9116
Revert "Revert "[UR][SYCL] Introduce UR api to set kernel args + laun…
ldorau Oct 10, 2025
4b57ee4
[UR][SYCL] Add support for zeCommandListAppendLaunchKernelWithArgumen…
ldorau Oct 15, 2025
e9e3158
Fix urEnqueueKernelLaunchWithArgsExp() in asan_ddi.cpp
ldorau Oct 17, 2025
3e99149
Make UrArgs thread_local
ldorau Oct 16, 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
15 changes: 8 additions & 7 deletions sycl/source/detail/queue_impl.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -467,13 +467,14 @@ EventImplPtr queue_impl::submit_kernel_scheduler_bypass(
BinImage = detail::retrieveKernelBinary(*this, KData.getKernelName());
assert(BinImage && "Failed to obtain a binary image.");
}
enqueueImpKernel(*this, KData.getNDRDesc(), KData.getArgs(),
KernelBundleImpPtr, KernelImplPtr,
*KData.getDeviceKernelInfoPtr(), RawEvents,
ResultEvent.get(), nullptr, KData.getKernelCacheConfig(),
KData.isCooperative(), KData.usesClusterLaunch(),
KData.getKernelWorkGroupMemorySize(), BinImage,
KData.getKernelFuncPtr());
enqueueImpKernel(
*this, KData.getNDRDesc(), KData.getArgs(), KernelBundleImpPtr,
KernelImplPtr, *KData.getDeviceKernelInfoPtr(), RawEvents,
ResultEvent.get(), nullptr, KData.getKernelCacheConfig(),
KData.isCooperative(), KData.usesClusterLaunch(),
KData.getKernelWorkGroupMemorySize(), BinImage,
KData.getKernelFuncPtr(), KData.getKernelNumArgs(),
KData.getKernelParamDescGetter(), KData.hasSpecialCaptures());
#ifdef XPTI_ENABLE_INSTRUMENTATION
if (xptiEnabled) {
// Emit signal only when event is created
Expand Down
244 changes: 178 additions & 66 deletions sycl/source/detail/scheduler/commands.cpp
View file
Open in desktop

Large diffs are not rendered by default.

4 changes: 3 additions & 1 deletion sycl/source/detail/scheduler/commands.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -633,7 +633,9 @@ void enqueueImpKernel(
ur_kernel_cache_config_t KernelCacheConfig, bool KernelIsCooperative,
const bool KernelUsesClusterLaunch, const size_t WorkGroupMemorySize,
const RTDeviceBinaryImage *BinImage = nullptr,
void *KernelFuncPtr = nullptr);
void *KernelFuncPtr = nullptr, int KernelNumArgs = 0,
detail::kernel_param_desc_t (*KernelParamDescGetter)(int) = nullptr,
bool KernelHasSpecialCaptures = true);

/// The exec CG command enqueues execution of kernel or explicit memory
/// operation.
Expand Down
4 changes: 2 additions & 2 deletions sycl/test-e2e/Adapters/level_zero/batch_barrier.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,7 +24,7 @@ int main(int argc, char *argv[]) {
queue q;

submit_kernel(q); // starts a batch
// CHECK: ---> urEnqueueKernelLaunch
// CHECK: ---> urEnqueueKernelLaunchWithArgsExp
// CHECK-NOT: zeCommandQueueExecuteCommandLists

// Initializing Level Zero driver is required if this test is linked
Expand All @@ -42,7 +42,7 @@ int main(int argc, char *argv[]) {
// CHECK-NOT: zeCommandQueueExecuteCommandLists

submit_kernel(q);
// CHECK: ---> urEnqueueKernelLaunch
// CHECK: ---> urEnqueueKernelLaunchWithArgsExp
// CHECK-NOT: zeCommandQueueExecuteCommandLists

// interop should close the batch
Expand Down
65 changes: 32 additions & 33 deletions sycl/test-e2e/Adapters/level_zero/batch_test.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -55,55 +55,54 @@
// variable SYCL_PI_LEVEL_ZEOR+BATCH_SIZE=N.
// This test enqueues 8 kernels and then does a wait. And it does this 3 times.
// Expected output is that for batching =1 you will see zeCommandListClose,
// and zeCommandQueueExecuteCommandLists after every urEnqueueKernelLaunch.
// For batching=3 you will see that after 3rd and 6th enqueues, and then after
// urQueueFinish. For 5, after 5th urEnqueue, and then after urQueueFinish. For
// 4 you will see these after 4th and 8th Enqueue, and for 8, only after the
// 8th enqueue. And lastly for 9, you will see the Close and Execute calls
// only after the urQueueFinish.
// Since the test does this 3 times, this pattern will repeat 2 more times,
// and then the test will print Test Passed 8 times, once for each kernel
// validation check.
// and zeCommandQueueExecuteCommandLists after every
// urEnqueueKernelLaunchWithArgsExp. For batching=3 you will see that after 3rd
// and 6th enqueues, and then after urQueueFinish. For 5, after 5th urEnqueue,
// and then after urQueueFinish. For 4 you will see these after 4th and 8th
// Enqueue, and for 8, only after the 8th enqueue. And lastly for 9, you will
// see the Close and Execute calls only after the urQueueFinish. Since the test
// does this 3 times, this pattern will repeat 2 more times, and then the test
// will print Test Passed 8 times, once for each kernel validation check.
// Pattern starts first set of kernel executions.
// CKALL: ---> urEnqueueKernelLaunch
// CKALL: ---> urEnqueueKernelLaunchWithArgsExp
// CKALL: zeCommandListAppendLaunchKernel(
// CKB1: zeCommandListClose(
// CKB1: zeCommandQueueExecuteCommandLists(
// CKALL: ---> urEnqueueKernelLaunch
// CKALL: ---> urEnqueueKernelLaunchWithArgsExp
// CKALL: zeCommandListAppendLaunchKernel(
// CKB1: zeCommandListClose(
// CKB1: zeCommandQueueExecuteCommandLists(
// CKALL: ---> urEnqueueKernelLaunch
// CKALL: ---> urEnqueueKernelLaunchWithArgsExp
// CKALL: zeCommandListAppendLaunchKernel(
// CKB1: zeCommandListClose(
// CKB1: zeCommandQueueExecuteCommandLists(
// CKB3: zeCommandListClose(
// CKB3: zeCommandQueueExecuteCommandLists(
// CKALL: ---> urEnqueueKernelLaunch
// CKALL: ---> urEnqueueKernelLaunchWithArgsExp
// CKALL: zeCommandListAppendLaunchKernel(
// CKB1: zeCommandListClose(
// CKB1: zeCommandQueueExecuteCommandLists(
// CKB4: zeCommandListClose(
// CKB4: zeCommandQueueExecuteCommandLists(
// CKALL: ---> urEnqueueKernelLaunch
// CKALL: ---> urEnqueueKernelLaunchWithArgsExp
// CKALL: zeCommandListAppendLaunchKernel(
// CKB1: zeCommandListClose(
// CKB1: zeCommandQueueExecuteCommandLists(
// CKB5: zeCommandListClose(
// CKB5: zeCommandQueueExecuteCommandLists(
// CKALL: ---> urEnqueueKernelLaunch
// CKALL: ---> urEnqueueKernelLaunchWithArgsExp
// CKALL: zeCommandListAppendLaunchKernel(
// CKB1: zeCommandListClose(
// CKB1: zeCommandQueueExecuteCommandLists(
// CKB3: zeCommandListClose(
// CKB3: zeCommandQueueExecuteCommandLists(
// CKALL: ---> urEnqueueKernelLaunch
// CKALL: ---> urEnqueueKernelLaunchWithArgsExp
// CKALL: zeCommandListAppendLaunchKernel(
// CKB1: zeCommandListClose(
// CKB1: zeCommandQueueExecuteCommandLists(
// CKB7: zeCommandListClose(
// CKB7: zeCommandQueueExecuteCommandLists(
// CKALL: ---> urEnqueueKernelLaunch
// CKALL: ---> urEnqueueKernelLaunchWithArgsExp
// CKALL: zeCommandListAppendLaunchKernel(
// CKB1: zeCommandListClose(
// CKB1: zeCommandQueueExecuteCommandLists(
Expand All @@ -121,45 +120,45 @@
// CKB9: zeCommandListClose(
// CKB9: zeCommandQueueExecuteCommandLists(
// Pattern starts 2nd set of kernel executions
// CKALL: ---> urEnqueueKernelLaunch
// CKALL: ---> urEnqueueKernelLaunchWithArgsExp
// CKALL: zeCommandListAppendLaunchKernel(
// CKB1: zeCommandListClose(
// CKB1: zeCommandQueueExecuteCommandLists(
// CKALL: ---> urEnqueueKernelLaunch
// CKALL: ---> urEnqueueKernelLaunchWithArgsExp
// CKALL: zeCommandListAppendLaunchKernel(
// CKB1: zeCommandListClose(
// CKB1: zeCommandQueueExecuteCommandLists(
// CKALL: ---> urEnqueueKernelLaunch
// CKALL: ---> urEnqueueKernelLaunchWithArgsExp
// CKALL: zeCommandListAppendLaunchKernel(
// CKB1: zeCommandListClose(
// CKB1: zeCommandQueueExecuteCommandLists(
// CKB3: zeCommandListClose(
// CKB3: zeCommandQueueExecuteCommandLists(
// CKALL: ---> urEnqueueKernelLaunch
// CKALL: ---> urEnqueueKernelLaunchWithArgsExp
// CKALL: zeCommandListAppendLaunchKernel(
// CKB1: zeCommandListClose(
// CKB1: zeCommandQueueExecuteCommandLists(
// CKB4: zeCommandListClose(
// CKB4: zeCommandQueueExecuteCommandLists(
// CKALL: ---> urEnqueueKernelLaunch
// CKALL: ---> urEnqueueKernelLaunchWithArgsExp
// CKALL: zeCommandListAppendLaunchKernel(
// CKB1: zeCommandListClose(
// CKB1: zeCommandQueueExecuteCommandLists(
// CKB5: zeCommandListClose(
// CKB5: zeCommandQueueExecuteCommandLists(
// CKALL: ---> urEnqueueKernelLaunch
// CKALL: ---> urEnqueueKernelLaunchWithArgsExp
// CKALL: zeCommandListAppendLaunchKernel(
// CKB1: zeCommandListClose(
// CKB1: zeCommandQueueExecuteCommandLists(
// CKB3: zeCommandListClose(
// CKB3: zeCommandQueueExecuteCommandLists(
// CKALL: ---> urEnqueueKernelLaunch
// CKALL: ---> urEnqueueKernelLaunchWithArgsExp
// CKALL: zeCommandListAppendLaunchKernel(
// CKB1: zeCommandListClose(
// CKB1: zeCommandQueueExecuteCommandLists(
// CKB7: zeCommandListClose(
// CKB7: zeCommandQueueExecuteCommandLists(
// CKALL: ---> urEnqueueKernelLaunch
// CKALL: ---> urEnqueueKernelLaunchWithArgsExp
// CKALL: zeCommandListAppendLaunchKernel(
// CKB1: zeCommandListClose(
// CKB1: zeCommandQueueExecuteCommandLists(
Expand All @@ -177,45 +176,45 @@
// CKB9: zeCommandListClose(
// CKB9: zeCommandQueueExecuteCommandLists(
// Pattern starts 3rd set of kernel executions
// CKALL: ---> urEnqueueKernelLaunch
// CKALL: ---> urEnqueueKernelLaunchWithArgsExp
// CKALL: zeCommandListAppendLaunchKernel(
// CKB1: zeCommandListClose(
// CKB1: zeCommandQueueExecuteCommandLists(
// CKALL: ---> urEnqueueKernelLaunch
// CKALL: ---> urEnqueueKernelLaunchWithArgsExp
// CKALL: zeCommandListAppendLaunchKernel(
// CKB1: zeCommandListClose(
// CKB1: zeCommandQueueExecuteCommandLists(
// CKALL: ---> urEnqueueKernelLaunch
// CKALL: ---> urEnqueueKernelLaunchWithArgsExp
// CKALL: zeCommandListAppendLaunchKernel(
// CKB1: zeCommandListClose(
// CKB1: zeCommandQueueExecuteCommandLists(
// CKB3: zeCommandListClose(
// CKB3: zeCommandQueueExecuteCommandLists(
// CKALL: ---> urEnqueueKernelLaunch
// CKALL: ---> urEnqueueKernelLaunchWithArgsExp
// CKALL: zeCommandListAppendLaunchKernel(
// CKB1: zeCommandListClose(
// CKB1: zeCommandQueueExecuteCommandLists(
// CKB4: zeCommandListClose(
// CKB4: zeCommandQueueExecuteCommandLists(
// CKALL: ---> urEnqueueKernelLaunch
// CKALL: ---> urEnqueueKernelLaunchWithArgsExp
// CKALL: zeCommandListAppendLaunchKernel(
// CKB1: zeCommandListClose(
// CKB1: zeCommandQueueExecuteCommandLists(
// CKB5: zeCommandListClose(
// CKB5: zeCommandQueueExecuteCommandLists(
// CKALL: ---> urEnqueueKernelLaunch
// CKALL: ---> urEnqueueKernelLaunchWithArgsExp
// CKALL: zeCommandListAppendLaunchKernel(
// CKB1: zeCommandListClose(
// CKB1: zeCommandQueueExecuteCommandLists(
// CKB3: zeCommandListClose(
// CKB3: zeCommandQueueExecuteCommandLists(
// CKALL: ---> urEnqueueKernelLaunch
// CKALL: ---> urEnqueueKernelLaunchWithArgsExp
// CKALL: zeCommandListAppendLaunchKernel(
// CKB1: zeCommandListClose(
// CKB1: zeCommandQueueExecuteCommandLists(
// CKB7: zeCommandListClose(
// CKB7: zeCommandQueueExecuteCommandLists(
// CKALL: ---> urEnqueueKernelLaunch
// CKALL: ---> urEnqueueKernelLaunchWithArgsExp
// CKALL: zeCommandListAppendLaunchKernel(
// CKB1: zeCommandListClose(
// CKB1: zeCommandQueueExecuteCommandLists(
Expand Down
Loading
Loading