[Hexagon] Optimize post-increment load and stores in loops. #82418
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
sgundapa
requested review from
androm3da,
krasimirgg,
iajbar and
SundeepKushwaha
sgundapa
force-pushed
the
postinc_new
branch
from
1f7a200 to
fe22e29
Compare
|
One of the issues that was exposed with this patch is fixed here : #82386 |
sgundapa
force-pushed
the
postinc_new
branch
4 times, most recently
from
2f40a7c to
52d7dcd
Compare
This patch optimizes the post-increment instructions so that we can packetize them together. v1 = phi(v0, v3') v2,v3 = post_load v1, 4 v2',v3'= post_load v3, 4 This can be optimized in two ways v1 = phi(v0, v3') v2,v3' = post_load v1, 8 v2' = load v1, 4
sgundapa
force-pushed
the
postinc_new
branch
from
52d7dcd to
3247c0e
Compare
|
This appears to have injected a failure in HVX codegen in Halide, e.g. |
|
Ping -- this is still broken. Is a fix (or rollback) on the way? Can we assist in some way? |
|
Hi Steven,
Is there a way you can provide me a test that triggers the run time failure ?
From: Steven Johnson ***@***.***>
Sent: Monday, February 26, 2024 11:53 AM
To: llvm/llvm-project ***@***.***>
Cc: Sumanth Gundapaneni ***@***.***>; State change ***@***.***>
Subject: Re: [llvm/llvm-project] [Hexagon] Optimize post-increment load and stores in loops. (PR #82418)
WARNING: This email originated from outside of Qualcomm. Please be wary of any links or attachments, and do not enable macros.
Ping -- this is still broken. Is a fix (or rollback) on the way? Can we assist in some way?
—
Reply to this email directly, view it on GitHub<#82418 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ABP7TI3JLGGXSNMEI2OGTE3YVTDWBAVCNFSM6AAAAABDR4AN5WVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSNRUG42TONJQG4>.
You are receiving this because you modified the open/close state.Message ID: ***@***.******@***.***>>
|
|
Please feel free to revert the change meanwhile. |
|
I don't have the privileges to revert this, but if you do so I will offer my approval :-) |
|
The standard Halide tests will repro it when configured for Hexagon; this
is a bit of work to set up, but honestly, testing Hexagon changes against
Halide is probably worth your while anyway, as we have been the first
project to report the last few breakages. See
https://github.com/halide/Halide/blob/main/README.md#halide-for-hexagon-hvx
for details (and please ping me if you have trouble getting it set up).
…On Mon, Feb 26, 2024 at 11:12 AM Sumanth Gundapaneni < ***@***.***> wrote:
Hi Steven,
Is there a way you can provide me a test that triggers the run time
failure ?
From: Steven Johnson ***@***.***>
Sent: Monday, February 26, 2024 11:53 AM
To: llvm/llvm-project ***@***.***>
Cc: Sumanth Gundapaneni ***@***.***>; State change ***@***.***>
Subject: Re: [llvm/llvm-project] [Hexagon] Optimize post-increment load
and stores in loops. (PR #82418)
WARNING: This email originated from outside of Qualcomm. Please be wary of
any links or attachments, and do not enable macros.
Ping -- this is still broken. Is a fix (or rollback) on the way? Can we
assist in some way?
—
Reply to this email directly, view it on GitHub<
#82418 (comment)>,
or unsubscribe<
https://github.com/notifications/unsubscribe-auth/ABP7TI3JLGGXSNMEI2OGTE3YVTDWBAVCNFSM6AAAAABDR4AN5WVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSNRUG42TONJQG4>.
You are receiving this because you modified the open/close state.Message
ID: ***@***.******@***.***>>
—
Reply to this email directly, view it on GitHub
<#82418 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AACQJ6Y65R4LBL2YC6RY3L3YVTNCPAVCNFSM6AAAAABDR4AN5WVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSNRVGAYDKNBRG4>
.
You are receiving this because you commented.Message ID:
***@***.***>
|
sgundapa
pushed a commit
to sgundapa/upstream_effort
that referenced
this pull request
Feb 27, 2024
…lvm#82418)" This reverts commit d62ca8d.
|
#83151 - to revert the change. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This patch optimizes the post-increment instructions so that we can packetize them together.
v1 = phi(v0, v3')
v2,v3 = post_load v1, 4
v2',v3'= post_load v3, 4
This can be optimized in two ways
v1 = phi(v0, v3')
v2,v3' = post_load v1, 8
v2' = load v1, 4