One third is:
The CI machine only achieving 1.4 FPS when we fake-render at 10 FPS. Possibly beyond help :P
One third is:
nbuffers expected to be 3, but measured to be 2. This might be one of the NBS regressions kdub is about to land a fix for. In theory the test should always hit 3 buffers because it renders in bursts that flood the queue.
I'll log new bugs and see if they reoccur. But seriously, if the CI machine can only manage 1 FPS, we need to restructure CI. Because that means there's no way to automate performance testing by any name.
The failures are unrelated to this branch.
One third is: /code.launchpad .net/~vanvugt/ mir/fix- 1583970/ +merge/ 295294
https:/
One third is:
The CI machine only achieving 1.4 FPS when we fake-render at 10 FPS. Possibly beyond help :P
One third is:
nbuffers expected to be 3, but measured to be 2. This might be one of the NBS regressions kdub is about to land a fix for. In theory the test should always hit 3 buffers because it renders in bursts that flood the queue.
I'll log new bugs and see if they reoccur. But seriously, if the CI machine can only manage 1 FPS, we need to restructure CI. Because that means there's no way to automate performance testing by any name.