kasuga-fj wrote:
I misunderstood GitHub behavior and cannot reopen this PR... Submitted another
PR #139690 with same contents.
https://github.com/llvm/llvm-project/pull/118267
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
htt
kasuga-fj wrote:
Thanks for your review.
https://github.com/llvm/llvm-project/pull/118267
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
kasuga-fj wrote:
@Meinersbur Could you please take a look?
I have submitted another PR #139254 as well, on which this PR depends, to
address the following problems: I first tried to simply replace the function
`isLexicographicallyPositive(std::vector &DV)` to take an additional
argument like
https://github.com/kasuga-fj edited
https://github.com/llvm/llvm-project/pull/118267
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
https://github.com/kasuga-fj updated
https://github.com/llvm/llvm-project/pull/118267
Rate limit ยท GitHub
body {
background-color: #f6f8fa;
color: #24292e;
font-family: -apple-system,BlinkMacSystemFont,Segoe
UI,Helvetica,Arial,sans
https://github.com/kasuga-fj edited
https://github.com/llvm/llvm-project/pull/118267
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
kasuga-fj wrote:
This PR is a part of [Stacked Pull
Requests](https://llvm.org/docs/GitHub.html#stacked-pull-requests) and depends
on #137662. The target branch will automatically change to main after the
dependent PR is merged. Could you please take a look at #137662 at first?
https://github
https://github.com/kasuga-fj created
https://github.com/llvm/llvm-project/pull/137663
In MachinePipeliner, loop-carried memory dependencies are represented by DAG,
which makes things complicated and causes some necessary dependencies to be
missing. This patch introduces a new class to manage l
https://github.com/kasuga-fj updated
https://github.com/llvm/llvm-project/pull/133667
>From ae5d9cff055000480e7e71205265b18911440e29 Mon Sep 17 00:00:00 2001
From: Ryotaro Kasuga
Date: Thu, 27 Mar 2025 09:52:16 +
Subject: [PATCH] [LoopInterchange] Fix the vectorizable check for a loop
In t
https://github.com/kasuga-fj updated
https://github.com/llvm/llvm-project/pull/133672
>From 72b48ba6d6b70eb9a65abdc516697f3dee9c7a2e Mon Sep 17 00:00:00 2001
From: Ryotaro Kasuga
Date: Thu, 27 Mar 2025 10:45:26 +
Subject: [PATCH] [LoopInterchange] Improve profitability check for
vectorizat
https://github.com/kasuga-fj edited
https://github.com/llvm/llvm-project/pull/133665
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
@@ -80,6 +80,21 @@ enum class RuleTy {
ForVectorization,
};
+/// Store the information about if corresponding direction vector was negated
kasuga-fj wrote:
> But I now guess that the complication here is the unique entries in the
> dependency matrix, is th
https://github.com/kasuga-fj created
https://github.com/llvm/llvm-project/pull/133667
In the profitability check for vectorization, the dependency matrix was not
handled correctly. This can result to make a wrong decision: It may say "this
loop can be vectorized" when in fact it cannot. The ro
https://github.com/kasuga-fj updated
https://github.com/llvm/llvm-project/pull/133667
>From bd84ddc9e4dc645e965b2a6dc535a3023e0d7e45 Mon Sep 17 00:00:00 2001
From: Ryotaro Kasuga
Date: Thu, 27 Mar 2025 09:52:16 +
Subject: [PATCH] [LoopInterchange] Fix the vectorizable check for a loop
In t
https://github.com/kasuga-fj updated
https://github.com/llvm/llvm-project/pull/133672
>From 692e4de4f84281f8c2bc5f7278f8066929df3cd6 Mon Sep 17 00:00:00 2001
From: Ryotaro Kasuga
Date: Thu, 27 Mar 2025 10:45:26 +
Subject: [PATCH] [LoopInterchange] Improve profitability check for
vectorizat
kasuga-fj wrote:
Depends on #133664 (Sorry for inconvenience, I tried using
[Graphite](https://app.graphite.dev/), but it didn't work fine due to my
network problem).
https://github.com/llvm/llvm-project/pull/133665
___
llvm-branch-commits mailing li
@@ -80,6 +80,21 @@ enum class RuleTy {
ForVectorization,
};
+/// Store the information about if corresponding direction vector was negated
kasuga-fj wrote:
> I think duplicated direction vectors are always allowed. They don't add new
> or different informa
https://github.com/kasuga-fj updated
https://github.com/llvm/llvm-project/pull/133672
>From 1a1c1f61a8cb179443d782127c157695bd21f6cc Mon Sep 17 00:00:00 2001
From: Ryotaro Kasuga
Date: Thu, 27 Mar 2025 10:45:26 +
Subject: [PATCH] [LoopInterchange] Improve profitability check for
vectorizat
kasuga-fj wrote:
Depends on #133667
https://github.com/llvm/llvm-project/pull/133672
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
kasuga-fj wrote:
Depends on #133665
https://github.com/llvm/llvm-project/pull/133667
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
https://github.com/kasuga-fj created
https://github.com/llvm/llvm-project/pull/133672
The vectorization profitability has a process to check whether a given loop can
be vectorized or not. Since the process is conservative, a loop that can be
vectorized may be deemed not to be possible. This ca
https://github.com/kasuga-fj created
https://github.com/llvm/llvm-project/pull/133665
There is a problem with the current profitability check for vectorization in
LoopInterchange. There are both false positives and false negatives. The former
means that the heuristic may say that "an exchange
22 matches
Mail list logo
