CmdStan slower than PyStan?

@mms29 would you mind if we used this model and data to add to our performance test suite so we can test for any future performance regressions to this part of the code?

No problem, feel free to use it! Thanks for responding so quickly, I use CmdStan 2.18 for now and it works perfect

1 Like

fixed via PR-2749 and merged into develop.
performance tests passing.
(@seantalts - what are upstream-cmdstan-tests and should they be run?)


When Math or Stan develop is updated, that triggers an update upstream automatically. At that point the CmdStan Performance Tests job on Jenkins should be triggered, though it didn’t work this time (manually triggered one to verify).

are we releasing a hotfix? if so, when?

The hotfix was pretty difficult last time as things were not really set up for it. I think I’d rather just do a full release now, which will either be 2.20.0 or 2.19.1 depending on if we have new features or just bug fixes in all of the repos, respectively. I’m going to start the release on Thursday at 1pm Eastern time with whatever last passed all the tests up and down the repos.


sounds like it’ll be a 2.19.1 release.

the manuals are only versioned to minor release level - pdfs for the 2.19 manuals are in the docs repo: