[SYCL] Switch from COPY_HOST_PTR to write for devices w/o HUM #3105
Conversation
This file contains hidden or 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
Switch from initializing native memory objects with COPY_HOST_PTR to creating them without a host ptr and following up with a write operation. The reasoning behind this change is that some OpenCL runtimes (e. g. FPGA) use the queue the memory object write is enqueued to as a hint to which device from the context this memory should be copied to, as opposed to COPY_HOST_PTR where only the context itself is known. This led to unnecessary temporary host copy made by the FPGA runtime during buffer creation.
|
/summary:run |
romanovvlad
approved these changes
Feb 3, 2021
bader
reviewed
Feb 4, 2021
sergey-semenov
added a commit
to sergey-semenov/llvm
that referenced
this pull request
Feb 4, 2021
Align the description of how native memory object creation is handled for unlinked allocation commands with the changes made in intel#3105.
iclsrc
pushed a commit
that referenced
this pull request
Apr 21, 2025
After KhronosGroup/SPIRV-Tools#6027 spirv-val is now rejecting `Aligned 0` Memory Operands. This caused various tests to fail validation with a recent SPIRV-Tools version. Original commit: KhronosGroup/SPIRV-LLVM-Translator@ec5d4d6a5a6594d
AlexeySachkov
pushed a commit
that referenced
this pull request
May 28, 2025
After KhronosGroup/SPIRV-Tools#6027 spirv-val is now rejecting `Aligned 0` Memory Operands. This caused various tests to fail validation with a recent SPIRV-Tools version. Original commit: KhronosGroup/SPIRV-LLVM-Translator@ec5d4d6a5a6594d
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Outdated suggestions cannot be applied.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.
Switch from initializing native memory objects with COPY_HOST_PTR to
creating them without a host ptr and following up with a write
operation for devices without host unified memory.
The reasoning behind this change is that some OpenCL runtimes (e. g.
FPGA) use the queue the memory object write is enqueued to as a hint to
which device from the context this memory should be copied to, as
opposed to COPY_HOST_PTR where only the context itself is known. This
led to unnecessary temporary host copy made by the FPGA runtime during
buffer creation.