[SYCL][CUDA] Pass device from context in create queue. #10491
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Signed-off-by: JackAKirk <[email protected]>
Signed-off-by: JackAKirk <[email protected]>
do we need then the device argument in urQueueCreateWithNativeHandle? I think the reason why make_queue passes nullptr for device is that the native queue has been created already in a given device, so it's not like you can create the new queue with the native handle in another device. |
jandres742
approved these changes
Jul 21, 2023
jchlanda
approved these changes
Aug 4, 2023
|
@jandres742 @kbenzie @pbalcer @rdeodhar I've created an issue for the question posed regarding the device argument in the appropriate place: oneapi-src/unified-runtime#778 This PR is a simple bug fix so |
steffenlarsen
pushed a commit
that referenced
this pull request
Aug 30, 2023
This PR adds missing functions in the hip backend to allow for interoperability in programs that create sycl objects from native hip objects. The new function implementations are: - `make_device` - `make_queue` - `make_event` Note that it would really make sense for #10491 to be merged first because this PR makes the same code change in pi2ur, for a fix that is attributed to #10491. --------- Signed-off-by: Jack Kirk <[email protected]>
fabiomestre
pushed a commit
to fabiomestre/llvm
that referenced
this pull request
Sep 26, 2023
Recently in the switch to UR `urQueueCreateFromNativeHandle` changed the previous behaviour whereby a queue was created with a device taken as the default device from the context. It changed it so that the queue was created with the device argument instead. Since the sycl runtime always passes a nullptr for the device when programmers call `make_queue(nativeStream, context)`, this broke `make_queue`. This patch reverts to the previous behaviour before the switch from pi cuda to ur cuda. Note that this should also fix `make_queue` for l0 which I also guess was broken due to the asserts meaning that this line was never reached: https://github.com/intel/llvm/blob/sycl/sycl/plugins/unified_runtime/ur/adapters/level_zero/queue.cpp#L574. But I have not tested this. --------- Signed-off-by: JackAKirk <[email protected]>
fabiomestre
pushed a commit
to fabiomestre/llvm
that referenced
this pull request
Sep 26, 2023
) This PR adds missing functions in the hip backend to allow for interoperability in programs that create sycl objects from native hip objects. The new function implementations are: - `make_device` - `make_queue` - `make_event` Note that it would really make sense for intel#10491 to be merged first because this PR makes the same code change in pi2ur, for a fix that is attributed to intel#10491. --------- Signed-off-by: Jack Kirk <[email protected]>
fabiomestre
pushed a commit
to fabiomestre/unified-runtime
that referenced
this pull request
Sep 26, 2023
This PR adds missing functions in the hip backend to allow for interoperability in programs that create sycl objects from native hip objects. The new function implementations are: - `make_device` - `make_queue` - `make_event` Note that it would really make sense for intel/llvm#10491 to be merged first because this PR makes the same code change in pi2ur, for a fix that is attributed to #10491. --------- Signed-off-by: Jack Kirk <[email protected]>
fabiomestre
pushed a commit
to oneapi-src/unified-runtime
that referenced
this pull request
Sep 27, 2023
This PR adds missing functions in the hip backend to allow for interoperability in programs that create sycl objects from native hip objects. The new function implementations are: - `make_device` - `make_queue` - `make_event` Note that it would really make sense for intel/llvm#10491 to be merged first because this PR makes the same code change in pi2ur, for a fix that is attributed to #10491. --------- Signed-off-by: Jack Kirk <[email protected]>
veselypeta
pushed a commit
to veselypeta/llvm
that referenced
this pull request
Sep 28, 2023
) This PR adds missing functions in the hip backend to allow for interoperability in programs that create sycl objects from native hip objects. The new function implementations are: - `make_device` - `make_queue` - `make_event` Note that it would really make sense for intel#10491 to be merged first because this PR makes the same code change in pi2ur, for a fix that is attributed to intel#10491. --------- Signed-off-by: Jack Kirk <[email protected]>
szadam
pushed a commit
to szadam/unified-runtime
that referenced
this pull request
Oct 13, 2023
This PR adds missing functions in the hip backend to allow for interoperability in programs that create sycl objects from native hip objects. The new function implementations are: - `make_device` - `make_queue` - `make_event` Note that it would really make sense for intel/llvm#10491 to be merged first because this PR makes the same code change in pi2ur, for a fix that is attributed to #10491. --------- Signed-off-by: Jack Kirk <[email protected]>
mdtoguchi
pushed a commit
to mdtoguchi/llvm
that referenced
this pull request
Oct 18, 2023
Recently in the switch to UR `urQueueCreateFromNativeHandle` changed the previous behaviour whereby a queue was created with a device taken as the default device from the context. It changed it so that the queue was created with the device argument instead. Since the sycl runtime always passes a nullptr for the device when programmers call `make_queue(nativeStream, context)`, this broke `make_queue`. This patch reverts to the previous behaviour before the switch from pi cuda to ur cuda. Note that this should also fix `make_queue` for l0 which I also guess was broken due to the asserts meaning that this line was never reached: https://github.com/intel/llvm/blob/sycl/sycl/plugins/unified_runtime/ur/adapters/level_zero/queue.cpp#L574. But I have not tested this. --------- Signed-off-by: JackAKirk <[email protected]>
omarahmed1111
pushed a commit
to omarahmed1111/unified-runtime
that referenced
this pull request
Oct 23, 2023
This PR adds missing functions in the hip backend to allow for interoperability in programs that create sycl objects from native hip objects. The new function implementations are: - `make_device` - `make_queue` - `make_event` Note that it would really make sense for intel/llvm#10491 to be merged first because this PR makes the same code change in pi2ur, for a fix that is attributed to #10491. --------- Signed-off-by: Jack Kirk <[email protected]>
omarahmed1111
pushed a commit
to omarahmed1111/unified-runtime
that referenced
this pull request
Oct 23, 2023
This PR adds missing functions in the hip backend to allow for interoperability in programs that create sycl objects from native hip objects. The new function implementations are: - `make_device` - `make_queue` - `make_event` Note that it would really make sense for intel/llvm#10491 to be merged first because this PR makes the same code change in pi2ur, for a fix that is attributed to #10491. --------- Signed-off-by: Jack Kirk <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Recently in the switch to UR
urQueueCreateFromNativeHandlechanged the previous behaviour whereby a queue was created with a device taken as the default device from the context. It changed it so that the queue was created with the device argument instead. Since the sycl runtime always passes a nullptr for the device when programmers callmake_queue(nativeStream, context), this brokemake_queue. This patch reverts to the previous behaviour before the switch from pi cuda to ur cuda.Note that this should also fix
make_queuefor l0 which I also guess was broken due to the asserts meaning that this line was never reached: https://github.com/intel/llvm/blob/sycl/sycl/plugins/unified_runtime/ur/adapters/level_zero/queue.cpp#L574. But I have not tested this.