Davide Valsecchi
Davide Valsecchi
Do you have news about when this nice feature will land in master? > > Anyway, I found the way to make `org-ql`'s sorting work within an `org-ql-block` block. >...
Hi @tvami sorry for the late reply. I'm investigating with experts. I will let you know.
Hi @tvami, I have checked our scripts and they are writing on local sqlite files, having no problems when running on my lxplus CMSSW instance. Can the issue be related...
Ok, sorry for the pain, probably testing locally one thing at a time I missed the issue before.. Preparing the small PR to fix this.
The change on the ONNX version was merged yesterday in https://github.com/cms-sw/cmsdist/pull/9046 at 15:00. Here it is using IB `CMSSW_14_1_X_2024-03-13-1100`. Don't we have to refresh the tests?
enable gpu
please test
+ml technical. Avoid running cuda tests if tensorflow is not compiled for Cuda. Making sure cuda tests fail if TF does not recognize the card.
> @valsdav , any idea why only `testTFVisibleDevicesCUDA` fails for GPU IBs (https://cmssdt.cern.ch/SDT/cgi-bin/logreader/el8_amd64_gcc12/CMSSW_14_1_GPU_X_2024-03-11-2300/unitTestLogs/PhysicsTools/TensorFlow#/89-89 ) ? Hi @smuzaffar I realized this morning testing locally that only that test is failing because...
> > all tests which requires TF to be build with GPU support > That's a good idea! Thanks!