<feed xmlns='http://www.w3.org/2005/Atom'>
<title>llvm-project.git/llvm/test/Transforms/LoopVectorize/forked-pointers.ll, branch users/nico/python-2</title>
<subtitle>Unnamed repository; edit this file 'description' to name the repository.
</subtitle>
<link rel='alternate' type='text/html' href='https://git.belthelziquor.com/llvm-project.git/'/>
<entry>
<title>[InstCombine] Handle "add like" in ADD+GEP-&gt;GEP+GEP rewrites (#135156)</title>
<updated>2025-04-14T15:11:13+00:00</updated>
<author>
<name>Björn Pettersson</name>
<email>bjorn.a.pettersson@ericsson.com</email>
</author>
<published>2025-04-14T15:11:13+00:00</published>
<link rel='alternate' type='text/html' href='https://git.belthelziquor.com/llvm-project.git/commit/?id=092b6e73e651469527662443b592f98f442ece72'/>
<id>092b6e73e651469527662443b592f98f442ece72</id>
<content type='text'>
Considering that "or disjoint" is the canonical for certain add
operations, then I think we want to support such "add like" operations
when doing ADD+GEP-&gt;GEP+GEP rewrites to make things more consistent.

Problem was found when improving ValueTracking, which turned an ADD into
OR, and then suddenly optimizations got worse due to these rewrites no
longer triggering.</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
Considering that "or disjoint" is the canonical for certain add
operations, then I think we want to support such "add like" operations
when doing ADD+GEP-&gt;GEP+GEP rewrites to make things more consistent.

Problem was found when improving ValueTracking, which turned an ADD into
OR, and then suddenly optimizations got worse due to these rewrites no
longer triggering.</pre>
</div>
</content>
</entry>
<entry>
<title>[InstCombine] Infer nusw + nneg -&gt; nuw for getelementptr (#111144)</title>
<updated>2024-12-05T13:36:40+00:00</updated>
<author>
<name>Nikita Popov</name>
<email>npopov@redhat.com</email>
</author>
<published>2024-12-05T13:36:40+00:00</published>
<link rel='alternate' type='text/html' href='https://git.belthelziquor.com/llvm-project.git/commit/?id=462cb3cd6cecd0511ecaf0e3ebcaba455ece587d'/>
<id>462cb3cd6cecd0511ecaf0e3ebcaba455ece587d</id>
<content type='text'>
If the gep is nusw (usually via inbounds) and the offset is
non-negative, we can infer nuw.

Proof: https://alive2.llvm.org/ce/z/ihztLy</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
If the gep is nusw (usually via inbounds) and the offset is
non-negative, we can infer nuw.

Proof: https://alive2.llvm.org/ce/z/ihztLy</pre>
</div>
</content>
</entry>
<entry>
<title>[InstCombine] Set disjoint flag when turning Add into Or. (#72702)</title>
<updated>2023-11-27T20:54:11+00:00</updated>
<author>
<name>Craig Topper</name>
<email>craig.topper@sifive.com</email>
</author>
<published>2023-11-27T20:54:11+00:00</published>
<link rel='alternate' type='text/html' href='https://git.belthelziquor.com/llvm-project.git/commit/?id=03d4a9d94da30590ebfc444cf13a8763f47b7bb9'/>
<id>03d4a9d94da30590ebfc444cf13a8763f47b7bb9</id>
<content type='text'>
The disjoint flag was recently added to IR in #72583 </content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
The disjoint flag was recently added to IR in #72583 </pre>
</div>
</content>
</entry>
<entry>
<title>[LoopUtils] Freeze compare results for diff checks instead of pointers.</title>
<updated>2023-11-21T10:54:36+00:00</updated>
<author>
<name>Florian Hahn</name>
<email>flo@fhahn.com</email>
</author>
<published>2023-11-21T10:54:35+00:00</published>
<link rel='alternate' type='text/html' href='https://git.belthelziquor.com/llvm-project.git/commit/?id=ead35564c0136fef0c1efface7b9857ad87af9de'/>
<id>ead35564c0136fef0c1efface7b9857ad87af9de</id>
<content type='text'>
THe freezes are introduced to avoid branch on undef/poison, if any of
the pointers may be poison. The same can be achieved by just freezing
the compare, which reduces the number of freezes needed. See
https://alive2.llvm.org/ce/z/NHa_ud

Note that the individual compares need to be frozen and it is not
sufficient to only freeze the resulting OR:

Result OR frozen only (UNSOUND): https://alive2.llvm.org/ce/z/YzFHQY
Individual conds frozen (SOUND): https://alive2.llvm.org/ce/z/5L6Z3f
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
THe freezes are introduced to avoid branch on undef/poison, if any of
the pointers may be poison. The same can be achieved by just freezing
the compare, which reduces the number of freezes needed. See
https://alive2.llvm.org/ce/z/NHa_ud

Note that the individual compares need to be frozen and it is not
sufficient to only freeze the resulting OR:

Result OR frozen only (UNSOUND): https://alive2.llvm.org/ce/z/YzFHQY
Individual conds frozen (SOUND): https://alive2.llvm.org/ce/z/5L6Z3f
</pre>
</div>
</content>
</entry>
<entry>
<title>[InstCombine] Set dead phi inputs to poison in more cases</title>
<updated>2023-08-01T09:53:47+00:00</updated>
<author>
<name>Nikita Popov</name>
<email>npopov@redhat.com</email>
</author>
<published>2023-08-01T08:37:42+00:00</published>
<link rel='alternate' type='text/html' href='https://git.belthelziquor.com/llvm-project.git/commit/?id=d01aec4c769d50fb92e86decd41d077c94105841'/>
<id>d01aec4c769d50fb92e86decd41d077c94105841</id>
<content type='text'>
Set phi inputs to poison whenever we find a dead edge (either
during initial worklist population or the main InstCombine run),
instead of only doing this for successors of dead blocks.

This means that the phi operand is set to poison even if for
critical edges without an intermediate block.

There are quite a few test changes, because the pattern is fairly
common in vectorizer output, for cases where we know the vectorized
loop will be entered.
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
Set phi inputs to poison whenever we find a dead edge (either
during initial worklist population or the main InstCombine run),
instead of only doing this for successors of dead blocks.

This means that the phi operand is set to poison even if for
critical edges without an intermediate block.

There are quite a few test changes, because the pattern is fairly
common in vectorizer output, for cases where we know the vectorized
loop will be entered.
</pre>
</div>
</content>
</entry>
<entry>
<title>[LoopVectorize] Regenerate test checks (NFC)</title>
<updated>2023-08-01T09:30:55+00:00</updated>
<author>
<name>Nikita Popov</name>
<email>npopov@redhat.com</email>
</author>
<published>2023-08-01T09:30:55+00:00</published>
<link rel='alternate' type='text/html' href='https://git.belthelziquor.com/llvm-project.git/commit/?id=7c64449e44c871102a38313aa4203972e36a930a'/>
<id>7c64449e44c871102a38313aa4203972e36a930a</id>
<content type='text'>
To reduce spurious diffs in future changes.
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
To reduce spurious diffs in future changes.
</pre>
</div>
</content>
</entry>
<entry>
<title>[LoopVectorize] Convert some tests to opaque pointers (NFC)</title>
<updated>2022-12-14T14:16:59+00:00</updated>
<author>
<name>Nikita Popov</name>
<email>npopov@redhat.com</email>
</author>
<published>2022-12-14T14:15:18+00:00</published>
<link rel='alternate' type='text/html' href='https://git.belthelziquor.com/llvm-project.git/commit/?id=7d7577256b76e4293f455b8093504d5f7044ab4b'/>
<id>7d7577256b76e4293f455b8093504d5f7044ab4b</id>
<content type='text'>
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
</pre>
</div>
</content>
</entry>
<entry>
<title>[NFC] Port all runlines for LoopVectorize pass tests to -passes syntax</title>
<updated>2022-12-05T19:17:30+00:00</updated>
<author>
<name>Roman Lebedev</name>
<email>lebedev.ri@gmail.com</email>
</author>
<published>2022-12-05T18:44:25+00:00</published>
<link rel='alternate' type='text/html' href='https://git.belthelziquor.com/llvm-project.git/commit/?id=be51fa4580b3274d35cc77bedb238a9ad5fb04bc'/>
<id>be51fa4580b3274d35cc77bedb238a9ad5fb04bc</id>
<content type='text'>
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
</pre>
</div>
</content>
</entry>
<entry>
<title>[LAA] Add recursive IR walker for forked pointers</title>
<updated>2022-07-18T11:06:17+00:00</updated>
<author>
<name>Graham Hunter</name>
<email>graham.hunter@arm.com</email>
</author>
<published>2022-06-16T08:59:29+00:00</published>
<link rel='alternate' type='text/html' href='https://git.belthelziquor.com/llvm-project.git/commit/?id=db8fcb2c2537fa1aa71d7d8fb94545408a8085d2'/>
<id>db8fcb2c2537fa1aa71d7d8fb94545408a8085d2</id>
<content type='text'>
This builds on the previous forked pointers patch, which only accepted
a single select as the pointer to check. A recursive function to walk
through IR has been added, which searches for either a loop-invariant
or addrec SCEV.

This will only handle a single fork at present, so selects of selects
or a GEP with a select for both the base and offset will be rejected.

There is also a recursion limit with a cli option to change it.

Reviewed By: fhahn, david-arm

Differential Revision: https://reviews.llvm.org/D108699
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
This builds on the previous forked pointers patch, which only accepted
a single select as the pointer to check. A recursive function to walk
through IR has been added, which searches for either a loop-invariant
or addrec SCEV.

This will only handle a single fork at present, so selects of selects
or a GEP with a select for both the base and offset will be rejected.

There is also a recursion limit with a cli option to change it.

Reviewed By: fhahn, david-arm

Differential Revision: https://reviews.llvm.org/D108699
</pre>
</div>
</content>
</entry>
<entry>
<title>[NFC][LAA] Precommit tests for forked pointers</title>
<updated>2021-11-24T16:20:35+00:00</updated>
<author>
<name>Graham Hunter</name>
<email>graham.hunter@arm.com</email>
</author>
<published>2021-08-12T14:38:13+00:00</published>
<link rel='alternate' type='text/html' href='https://git.belthelziquor.com/llvm-project.git/commit/?id=dee810e117ad442c426a1213277faa8b028751d5'/>
<id>dee810e117ad442c426a1213277faa8b028751d5</id>
<content type='text'>
Precommit for https://reviews.llvm.org/D108699
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
Precommit for https://reviews.llvm.org/D108699
</pre>
</div>
</content>
</entry>
</feed>
