llvm dev - Jul 2021 - Enabling Loop Distribution Pass as default in the pipeline of new pass manager

Ping.

Additionally, I was not able to see the pass triggered from llvm-test-suite and
spec benchmark except hmmer.

Thanks
JinGu Kang

From: Sanne Wouda <Sanne.Wouda at arm.com>
Sent: 25 June 2021 11:23
To: Jingu Kang <Jingu.Kang at arm.com>; llvm-dev at lists.llvm.org;
Florian Hahn <florian_hahn at apple.com>
Cc: nikic at php.net
Subject: Re: [llvm-dev] Enabling Loop Distribution Pass as default in the
pipeline of new pass manager

Hi,
Do you have any data on how often LoopDistribute triggers on a larger set of
programs (like llvm-test-suite + SPEC)? AFAIK the implementation is very limited
at the moment (geared towards catching the case in hmmer) and I suspect lack of
generality is one of the reasons why it is not enabled by default yet.
It would be good to have some fresh numbers on how often LoopDistribute
triggers. From what I remember, there are a handful of cases in the test suite,
but nothing that significantly affects performance (other than hmmer,
obviously).
Also, there's been an effort to improve the cost-modeling for LoopDistribute
(https://reviews.llvm.org/D100381) Should we make progress in that direction
first, before enabling by default?
Unfortunately, there were some problems with this effort. First, the current
implementation of LoopDistribute relies heavily on LoopAccessAnalysis, which
made it difficult to adapt.

More importantly though, I'm not convinced that LoopDistribute will be
beneficial other than in cases where it enables more vectorization. (The memcpy
detection gcc might be interesting, I didn't look at that.) It reduces both
ILP and MLP, which in some cases might be made up by lower register or cache
pressure, but this is hard or impossible for the compiler to know.

While working on this, with a more aggressive LoopDistribute across several
benchmarks, I did not see any improvements that didn't turn out to be noise,
and plenty of cases where it was actively degrading performance.

Therefore, I'm not sure this direction is worth pursuing further, and I
believe the current heuristic of "distribute when it enables new
vectorization" is actually pretty reasonable, if not very general.

Cheers,
Sanne
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20210705/257175c7/attachment.html>

Hi,
> On 5 Jul 2021, at 15:40, Jingu Kang <jingu.kang at arm.com> wrote:
> 
> Ping.
>  
> Additionally, I was not able to see the pass triggered from llvm-test-suite
and spec benchmark except hmmer.
>  
Hm, this seems to indicate that the current version of loop-distribute is not
very general at all. I’m not sure if we want to enable the pass by default, as
long as it effectively only benefits a single benchmark?

Taking a step back, the question this raises for me is *why* it does not trigger
anywhere else at all? I doubt there are no other cases in the test-suite and/or
various SPEC benchmarks that would benefit from loop distribution. It would at
least be helpful to understand *why* the pass misses them at the moment and what
kind of changes are needed to make it more general.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20210714/20467287/attachment.html>