Conversation
| run: | | ||
| echo "SCCACHE_DIR=$env:CCACHE_DIR_INPUT" | Out-File -FilePath $env:GITHUB_ENV -Encoding utf8 -Append | ||
| echo "SCCACHE_MAXSIZE=10G" | Out-File -FilePath $env:GITHUB_ENV -Encoding utf8 -Append | ||
| # Wthout this we get a linking error about not being able to find sycl-devicelib-host.lib | ||
| echo "LIB=$env:LIB;$env:SYCL_DIR_INPUT\lib" | Out-File -FilePath $env:GITHUB_ENV -Encoding utf8 -Append | ||
| echo "CMAKE_EXTRA_ARGS=-DWITH_WINDOWS_SCCACHE=ON" | Out-File -FilePath $env:GITHUB_ENV -Encoding utf8 -Append |
There was a problem hiding this comment.
@KornevNikita Do we need these checks to pass before merging PRs or are we allowed to ignore them?
This warning seems to suggest we aren't allowed to modify any environment variables with non-static strings, and we have to set these vars to get things to work. We already do that heavily in other workflows, it's just the checks were added after those workflows were written. Thanks
There was a problem hiding this comment.
Yeah, this is annoying, I suggest we ignore these :)
There was a problem hiding this comment.
I like that answer, thanks :P
04d162e to
d8e53e1
Compare
Signed-off-by: Nick Sarnie <nick.sarnie@intel.com>
|
do we need to clean up anything afterwards? |
| inputs: | ||
| runner: | ||
| type: string | ||
| required: True |
There was a problem hiding this comment.
| required: True | |
| required: true |
| run: | | ||
| echo "SCCACHE_DIR=$env:CCACHE_DIR_INPUT" | Out-File -FilePath $env:GITHUB_ENV -Encoding utf8 -Append | ||
| echo "SCCACHE_MAXSIZE=10G" | Out-File -FilePath $env:GITHUB_ENV -Encoding utf8 -Append | ||
| # Wthout this we get a linking error about not being able to find sycl-devicelib-host.lib |
There was a problem hiding this comment.
| # Wthout this we get a linking error about not being able to find sycl-devicelib-host.lib | |
| # Without this we get a linking error about not being able to find sycl-devicelib-host.lib |
|
On the next run in the checkout phase, it runs |
Shouldn't we manually clean up the space like we do it for E2E/SYCL-CTS tests? |
Yeah that's cleaner, let me do that, thanks |
Signed-off-by: Nick Sarnie <nick.sarnie@intel.com>
|
Feedback should be addressed in latest commit, thanks Test CI run here https://github.com/intel/llvm/actions/runs/24355748290/job/71130372156 |
Thanks! |
| runs: | ||
| using: "composite" | ||
| steps: | ||
| - uses: ilammy/msvc-dev-cmd@0b201ec74fa43914dc39ae48a89fd1d8cb592756 |
There was a problem hiding this comment.
| - uses: ilammy/msvc-dev-cmd@0b201ec74fa43914dc39ae48a89fd1d8cb592756 | |
| - uses: ilammy/msvc-dev-cmd@0b201ec74fa43914dc39ae48a89fd1d8cb592756 # v1.13.0 |
Same below and below and below. Not a blocker, feel free to ignore/make a follow up. I'm updating "uses: ..." across whole repo, so it'd be nice to have this.
There was a problem hiding this comment.
I'll add this then merge, thanks
Signed-off-by: Nick Sarnie <nick.sarnie@intel.com>
This PR introduces a real world test into CI.
The idea is we build and run Blender using the just-built SYCL compiler (initially, only the nightly) and then run an example workflow on a (multi) GPU system.
The first version of this PR only tracks pass/fail, but in the future we could add a performance threshold to cause a fail.
First we build OIDN and Embree, both of which use SYCL and DPC++. Then we build Blender with the just-built OIDN, Embree and SYCL compiler.
The steps of this workflow (as in, what to do at a high level) were developed with help from internal Blender experts.
Run in nightly example: https://github.com/intel/llvm/actions/runs/24317863959/job/70999916822
Entire workflow takes ~10 minutes.