Raghu Raja
Raghu Raja
> it'll need to be added to cmake too. and do we actually want to push the cost of the extra library on to all end users, even if they...
Fixed build failures.
`//test:perf_counters_gtest` and `//test:perf_counters_test` are failing. Are these tests that never ran before in the CI environment, given the perf counter path was not built? ``` [ FAILED ] PerfCountersTest.ReopenExistingCounters (0...
Perhaps I am not following your ask. CMake uses `check_library_exists` to build with the perf counter support when the library is available on the system benchmark is getting built on....
Apologies for letting this sit, I've been taking some time off. I hope to get back to this next week.
It is hard to tell who is maintaining perftests these days, but tagging @HassanKhadour who made the last stable release for review. I do not have triage permissions on this...
Indeed, I will fix that up. I am in favor of completely removing it if we can.. given the explicit configure warning (which we can turn into an error) and...
> Copy-pasting the same concepts every time a new accelerator is introduced isn't very scalable, I would argue it's time to abstract these and have specific "hardware providers" implementations. +1,...
Addressed the CMake situation, removed the explicit CMakeLists.txt change to always turn on PFM... Added a config_setting to make this optional for Bazel builds too. Please take a look @LebedevRI
This is a valuable addition to Lemur for GCP users. Is there a possibility of this getting merged upstream?