Enable compiling for cuda #1454
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
* Implements dpctl.tensor.cbrt * Implements copysign and exp2 elementwise funcs * Adds tests for cbrt, copysign, exp2 * Implements rsqrt and tests for rsqrt * Modified tests for cbrt, copysign, and rsqrt Now test more type combinations/output types
Tweaked test_intel_device_info
Use sycl_complex extension to implement complex-valued trigonometric, hyperbolic functions and their inverses. This works around use of double precision functions/literals in implementations of these functions in MSVC headers, causing failures to offload on Iris Xe for single precision input citing lack of fp64 support by the hardware.
Used functions from sycl::ext::oneapi::experimental context to implement evaluation on data of complex type.
For every CMake target, where add_sycl_to_target is used, we also run
target_compile_options(
${target_name}
PRIVATE
-fysl-targets=spir64-unknown-unknown,nvptx64-nvidia-cuda
)
Replaced them with uses of sycl::ext::oneapi::experimental namespace functions instead.
Also DPCTL_SYCL_TARGETS parameter can be used to specify targets to build for. DPCTL_TARGET_CUDA could be set via cmake option, or via environment variable, e.g. ``` $ DPCTL_TARGET_CUDA=1 python scripts/build_locally.py --verbose ```
|
View rendered docs @ https://intelpython.github.io/dpctl/pulls/1454/index.html |
ndgrigorian
approved these changes
Oct 25, 2023
|
Array API standard conformance tests for dpctl=0.15.1dev0=py310ha25a700_71 ran successfully. |
|
Array API standard conformance tests for dpctl=0.15.1dev0=py310ha25a700_72 ran successfully. |
oleksandr-pavlyk
merged commit 479a969
into
use-sycl-ext-oneapi-experimental-for-complex
23 of 26 checks passed
|
Deleted rendered PR docs from intelpython.github.com/dpctl, latest should be updated shortly. 🤞 |
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.
This PR fixes build for CUDA by replacing overlooked instances of uses
stdnamespace functions for complex type inputs with uses ofsycl::ext::oneapi::experimentalnamespace functions.Cmake scripts are modified to allow building multi sycl-targets binaries. This can be done by specifying
-DDPCTL_TARGET_CUDA:BOOL=ON, or by setting environment variableDPCTL_TARGET_CUDA=1.It is also possible to manually specify sycl targets string via
-DDPCTL_SYCL_TARGETS=nvptx64-nvidia-cuda,spir64-unknown-unknown.Test suite runs to completion, when
ONEAPI_DEVICE_SELECTOR=cuda:gpuis set, but with many test failures which must be investigated, e.g.This may be caused by compiler support not yet implemented for these types.