[SYCL][CUDA] Enable multiple CUDA arch in the same SYCL build. #2087
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
The patch refactored the SYCL offloading to account for bound arch as well as toolchains. Compilation target are represented as a pair of a toolchain (one toolchain per target triple) and a bound arch (which encodes the target arch). The LTO step is now performed per toolchain/bound arch pair and the produced binary are then bundled per toolchain. The patch also makes the bundler/unbundler use the bound arch when SYCL offloading if enabled. Signed-off-by: Victor Lomuller <[email protected]> Co-authored-by: Alexander Johnston <[email protected]>
Naghasan
force-pushed
the
victor/cuda-multi-arch
branch
from
451bbd3 to
fe2b7e5
Compare
AGindinson
reviewed
Jul 10, 2020
Co-authored-by: Artem Gindinson <[email protected]>
Signed-off-by: Victor Lomuller <[email protected]>
|
@mdtoguchi, @AGindinson, please, take a look at this change. |
bader
approved these changes
Jul 22, 2020
mdtoguchi
reviewed
Jul 22, 2020
mdtoguchi
reviewed
Jul 22, 2020
| auto &LI = LinkInputEnum.value(); | ||
| const ToolChain *TC = SYCLTargetInfoList[LinkInputEnum.index()].TC; | ||
| const char *BoundArch = | ||
| SYCLTargetInfoList[LinkInputEnum.index()].BoundArch; |
There was a problem hiding this comment.
This for loop is pretty big, would it be better to assign BoundArch closer to where it is used for clarity? I guess you could look at it either way due to the fact that LinkInputEnum would be lost at the BoundArch use.
mdtoguchi
reviewed
Jul 22, 2020
Comment on lines
+4164
to
+4165
| // Note: this odd, but the test assert that a integration header | ||
| // is build no matter what. If only fsycl-add-targets is provided, |
There was a problem hiding this comment.
Suggested change
| // Note: this odd, but the test assert that a integration header | |
| // is build no matter what. If only fsycl-add-targets is provided, | |
| // Note: this is odd, but the test asserts that an integration header | |
| // is built no matter what. If only fsycl-add-targets is provided, |
jsji
pushed a commit
that referenced
this pull request
Aug 11, 2023
Workaround unsupported freeze insn by: replacing uses of freeze's result with freeze's source or a random (but compilation reproducible) constant if freeze's source is undef/poison deleting freeze insn. Long term solution is to add a freeze instruction extension in SPIR-V. Issue is tracked in (#1140) Signed-off-by: Lu, John <[email protected]> Original commit: KhronosGroup/SPIRV-LLVM-Translator@ed25856
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.
The patch refactored the SYCL offloading to account for bound arch as well as toolchains.
Compilation target are represented as a pair of a toolchain (one toolchain per target triple) and a bound arch (which encodes the target arch).
The LTO step is now performed per toolchain/bound arch pair and
the produced binary are then bundled per toolchain.
The patch also makes the bundler/unbundler use the bound arch when SYCL offloading if enabled.
Signed-off-by: Victor Lomuller [email protected]
Co-authored-by: Alexander Johnston [email protected]