[PATCH] D38154: [PassManager] Run global opts after the inliner

Chandler Carruth via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Mon Oct 2 22:23:53 PDT 2017


chandlerc added a comment.

In https://reviews.llvm.org/D38154#886713, @mehdi_amini wrote:

> In https://reviews.llvm.org/D38154#886712, @chandlerc wrote:
>
> > In https://reviews.llvm.org/D38154#886708, @mzolotukhin wrote:
> >
> > > I'd recommend using LNT,
> >
> >
> > This seems like a really cumbersome process to expect folks to go through for changes intending to improve compile time. Is there no way to simplify it?
>
>
> Which part is cumbersome?


I already have test-suite checked out and correctly hooked up to SPEC and other things. LNT is a large Python project including server code and other things. It all runs in a Python virtualenv which I don't know anything about and don't use for any other part of development.

> I use to do this all the time, and it is actually pretty nice when you get use to it. Especially it can build once and run multiple time each test to get more sample and reduce the measurement noise.

Can definitely get multiple runs with lit directly, no need for virtualenv or LNT... In fact, AFAIK, LNT just uses lit to do all of this already. LNT seems really focused on running a server and collecting and aggregating results over time, rather than checking a single commit.

> The comparison script is easy to use as well.

Yeah, I have used the comparison script myself with lit directly, seems quite nice.

> Note, in case there is any confusion: this does not require to setup or use an LNT *instance* (as in the web application).

Yes, but you're still paying for a lot of this infrastructure.

> On the other hand I never found any similar convenience with the cmake/lit use of the test-suite (maybe too young to have proper support?).

Possibly, dunno when this support landed.


https://reviews.llvm.org/D38154





More information about the llvm-commits mailing list