Skip to content

Conversation

@Naghasan
Copy link

The patch adds the generic SPIR-V implementation for the math builtins (so nvptx specific implementation yet).

Signed-off-by: Victor Lomuller [email protected]
Co-authored-by: David Wood [email protected]

@Naghasan Naghasan requested review from a team and bader as code owners June 26, 2020 09:27
@Naghasan Naghasan requested a review from sergey-semenov June 26, 2020 09:27
Signed-off-by: Victor Lomuller <[email protected]>
Co-authored-by: David Wood <[email protected]>
@Naghasan Naghasan force-pushed the victor/spirv-ptx-math branch from 64e9720 to 35fbd3f Compare June 26, 2020 12:30
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This probably should belong to other PR.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I assume these "common" built-ins are used in "math" built-ins implementation, right?

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This probably should belong to other PR.

That's a change that relates to the refactoring prior to this patch. This wasn't part of it as it directly conflicted with the PR #1832. This avoided a dependency.

The header file included is the same, ../../lib/clcmacro.h is a link to the one in include. So there is no functional change, just consistency.

I assume these "common" built-ins are used in "math" built-ins implementation, right?

Some common builtins are indeed used by the math one.

*/

#include <clc/clc.h>
//===----------------------------------------------------------------------===//
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Hm... shouldn't we keep old AMD copyright header?

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I can revert if you prefer.

Technically, the LLVM relicensed all sub-projects under this license, but this one never updated the headers. But I'm no lawyer.

@bader bader added the cuda CUDA back-end label Jun 29, 2020
@bader bader merged commit 0c9c9c0 into intel:sycl Jun 30, 2020
@bader bader added the libclc libclc project related issues label Jul 3, 2020
kchusha pushed a commit to kchusha/llvm that referenced this pull request May 10, 2023
…tel#1990)

Add the library `BINARY_DIR` to `LD_LIBRARY_PATH` to ensure that
the freshly built `libLLVMSPIRVLib.so` is tested.  Otherwise, llvm-spirv
spawned by the test suite may use the previously installed
`libLLVMSPIRVLib.so`.

I have noticed the problem after rebuilding LLVM with
`-DLLVM_ENABLE_ASSSERTIONS=ON`.  This meant that the previous version
of `libLLVMSPIRVLib.so` now crashed, effectively causing the test suite
to fail incorrectly.

Signed-off-by: Michał Górny <[email protected]>

Original commit:
KhronosGroup/SPIRV-LLVM-Translator@ba965cd
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

cuda CUDA back-end libclc libclc project related issues

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants