Skip to content

Conversation

@AlexeySachkov
Copy link
Contributor

This PR intended to reduce amount of template instantiations performed during both host & device compilation passes, as well as reduce amount of LLVM IR emitted during host compilation pass.

handler::verifyUsedKernelBundle is a one-line helper which was inlined to avoid one instantiation per-kernel and emission of associated LLVM IR. Its uses (now inlined) were also moved under #ifndef __SYCL_DEVICE_ONLY__ to reduce amount of code device compiler has to parse.

This PR intended to reduce amount of template instantiations performed
during both host & device compilation passes, as well as reduce amount
of LLVM IR emitted during host compilation pass.

`handler::verifyUsedKernelBundle` is a one-line helper which was
inlined to avoid one instantiation per-kernel and emission of associated
LLVM IR. Its uses (now inlined) were also moved under
`#ifndef __SYCL_DEVICE_ONLY__` to reduce amount of code device compiler
has to parse.
@AlexeySachkov AlexeySachkov merged commit f126cf5 into intel:sycl Sep 27, 2024
12 checks passed
@AlexeySachkov AlexeySachkov deleted the private/asachkov/optimize-verifykernelbundle-usage branch October 9, 2024 08:01
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants