Print correct variable names in cuda.cmake #153402
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/153402
Note: Links to docs will display an error until the docs builds have been completed. ❌ 1 New Failure, 1 Unrelated FailureAs of commit 51f70b4 with merge base 45df18d ( NEW FAILURE - The following job has failed:
BROKEN TRUNK - The following job failed but were present on the merge base:👉 Rebase onto the `viable/strict` branch to avoid these failures
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
|
This PR needs a
|
| @@ -232,7 +232,7 @@ if(CAFFE2_USE_CUSPARSELT) | |||
| target_link_libraries(torch::cusparselt INTERFACE ${CUSPARSELT_LIBRARY_PATH}) | |||
| endif() | |||
| else() | |||
| message(STATUS "USE_CUSPARSELT is set to 0. Compiling without cuSPARSELt support") | |||
| message(STATUS "CAFFE2_USE_CUSPARSELT is set to 0. Compiling without cuSPARSELt support") | |||
| endif() | |||
|
|
|||
| if(USE_CUDSS) | |||
There was a problem hiding this comment.
That one did not have a CAFFE2_ prefix which is unfortunate, but changing it would break backward compatibility so I didn't update here.
There was a problem hiding this comment.
I'm not sure what do you mean by "correct" names here? I.e. is there documentaiton/or code that says those should be it?
I.e.
Line 241 in 27e9d9b
clearly defines
USE_CUDNN not CAFFE2_USE_CUDNN
|
We have dropped Caffe2 and the names were changed to remove the |
|
@malfet @cyyever : please see following places in
See instances where using the
That's also the recommendation currently given by the very @ptrblck today given the |
Hi @cyyever. There are currently
Removing them sounds fine though it breaks backward comp, might want to keep checking for both (even if only printing without the prefix) |
|
@agirault May you help mitigate the remaining |
Context
https://discuss.pytorch.org/t/use-cudnn-set-to-zero-after-upgrade-to-libtorch-2-0/175413/4
Change
Match status message with the actual variable name
Note: still confusing to me what these impact when importing a pre-built libtorch with
find_package(Torch)