[llvm] [clang] [Instcombine] use zext's nneg flag for icmp folding (PR #70845)

leo-ard wrote: @nikic ping:) https://github.com/llvm/llvm-project/pull/70845 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[llvm] [clang] [Instcombine] use zext's nneg flag for icmp folding (PR #70845)

https://github.com/leo-ard updated https://github.com/llvm/llvm-project/pull/70845 From 00d0c18b5414ffe7222e1ee0ad5ecfdb8783704e Mon Sep 17 00:00:00 2001 From: leo-ard Date: Mon, 30 Oct 2023 18:01:27 -0400 Subject: [PATCH 01/12] Add NonNeg check for InstCombine --- llvm/lib/Transforms/InstCom

[llvm] [clang] [Instcombine] use zext's nneg flag for icmp folding (PR #70845)

@@ -0,0 +1,175 @@ +; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 3 +; See PR-70845 for more details +; RUN: opt < %s -S -passes=instcombine | FileCheck %s + + +define signext i32 @sext_sext(i16 %x, i16 %y) { +; CHECK-LABEL: define s

[llvm] [clang] [Instcombine] use zext's nneg flag for icmp folding (PR #70845)

@@ -247,6 +355,19 @@ define i1 @sext_zext_uge_op0_wide(i16 %x, i8 %y) { ret i1 %c } + +define i1 @sext_zext_nneg_uge_op0_wide(i16 %x, i8 %y) { +; CHECK-LABEL: @sext_zext_nneg_uge_op0_wide( +; CHECK-NEXT:[[TMP1:%.*]] = sext i8 [[Y:%.*]] to i16 +; CHECK-NEXT:[[C:%.*]]

[llvm] [clang] [Instcombine] use zext's nneg flag for icmp folding (PR #70845)

@@ -0,0 +1,175 @@ +; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 3 +; See PR-70845 for more details +; RUN: opt < %s -S -passes=instcombine | FileCheck %s + + +define signext i32 @sext_sext(i16 %x, i16 %y) { leo-ard

[clang] [llvm] [Instcombine] use zext's nneg flag for icmp folding (PR #70845)

https://github.com/leo-ard updated https://github.com/llvm/llvm-project/pull/70845 From 00d0c18b5414ffe7222e1ee0ad5ecfdb8783704e Mon Sep 17 00:00:00 2001 From: leo-ard Date: Mon, 30 Oct 2023 18:01:27 -0400 Subject: [PATCH 01/13] Add NonNeg check for InstCombine --- llvm/lib/Transforms/InstCom

[clang] [llvm] [Instcombine] use zext's nneg flag for icmp folding (PR #70845)

https://github.com/leo-ard edited https://github.com/llvm/llvm-project/pull/70845 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[llvm] [clang] [Instcombine] use zext's nneg flag for icmp folding (PR #70845)

@@ -247,6 +355,19 @@ define i1 @sext_zext_uge_op0_wide(i16 %x, i8 %y) { ret i1 %c } + +define i1 @sext_zext_nneg_uge_op0_wide(i16 %x, i8 %y) { +; CHECK-LABEL: @sext_zext_nneg_uge_op0_wide( +; CHECK-NEXT:[[TMP1:%.*]] = sext i8 [[Y:%.*]] to i16 +; CHECK-NEXT:[[C:%.*]]

[llvm] [clang] [Instcombine] use zext's nneg flag for icmp folding (PR #70845)

@@ -0,0 +1,145 @@ +; RUN: opt < %s --O3 -S | FileCheck %s leo-ard wrote: How did you get the SSA form ? I wasn't able to do it on my side. Here are the commands that I did : ```bash > build_release/bin/clang -S -emit-llvm min_max.c -fno-discard-value-names -o

[clang] [llvm] [Instcombine] use zext's nneg flag for icmp folding (PR #70845)

https://github.com/leo-ard updated https://github.com/llvm/llvm-project/pull/70845 From 00d0c18b5414ffe7222e1ee0ad5ecfdb8783704e Mon Sep 17 00:00:00 2001 From: leo-ard Date: Mon, 30 Oct 2023 18:01:27 -0400 Subject: [PATCH 01/14] Add NonNeg check for InstCombine --- llvm/lib/Transforms/InstCom

[clang] [llvm] [Instcombine] use zext's nneg flag for icmp folding (PR #70845)

@@ -0,0 +1,112 @@ +; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 3 +; RUN: opt < %s -O3 -S | FileCheck %s +; See issue #55013 and PR #70845 for more details. +; This test comes from the following C program, compiled with clang +; +;

[clang] [llvm] [Instcombine] use zext's nneg flag for icmp folding (PR #70845)

https://github.com/leo-ard edited https://github.com/llvm/llvm-project/pull/70845 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [llvm] [Instcombine] use zext's nneg flag for icmp folding (PR #70845)

https://github.com/leo-ard updated https://github.com/llvm/llvm-project/pull/70845 From 00d0c18b5414ffe7222e1ee0ad5ecfdb8783704e Mon Sep 17 00:00:00 2001 From: leo-ard Date: Mon, 30 Oct 2023 18:01:27 -0400 Subject: [PATCH 01/15] Add NonNeg check for InstCombine --- llvm/lib/Transforms/InstCom

[llvm] [clang] [Instcombine] use zext's nneg flag for icmp folding (PR #70845)

https://github.com/leo-ard updated https://github.com/llvm/llvm-project/pull/70845 From 00d0c18b5414ffe7222e1ee0ad5ecfdb8783704e Mon Sep 17 00:00:00 2001 From: leo-ard Date: Mon, 30 Oct 2023 18:01:27 -0400 Subject: [PATCH 01/16] Add NonNeg check for InstCombine --- llvm/lib/Transforms/InstCom

[llvm] [clang] [Instcombine] use zext's nneg flag for icmp folding (PR #70845)

@@ -0,0 +1,111 @@ +; RUN: opt < %s -O3 -S | FileCheck %s +; See issue #55013 and PR #70845 for more details. +; This test comes from the following C program, compiled with clang +; +;; short vecreduce_smin_v2i16(int n, short* v) +;; { +;; short p = 0; +;; for (int i = 0; i < n

[llvm] [clang] [InstCombine] Use zext's nneg flag for icmp folding (PR #70845)

leo-ard wrote: > Do you have the access to merge PR? No I don't. Thanks for the time and insightful comments ! https://github.com/llvm/llvm-project/pull/70845 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/ma

[clang] [llvm] [Instcombine] use zext's nneg flag for icmp folding (PR #70845)

https://github.com/leo-ard created https://github.com/llvm/llvm-project/pull/70845 This PR fixes https://github.com/llvm/llvm-project/issues/55013 : the max intrinsics is not generated for this simple loop case : https://godbolt.org/z/hxz1xhMPh. This is caused by a ICMP not being folded into

[clang] [llvm] [SCCP] [Transform] Adding ICMP folding for zext and sext in SCCPSolver (PR #67594)

https://github.com/leo-ard closed https://github.com/llvm/llvm-project/pull/67594 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[llvm] [clang] [SCCP] [Transform] Adding ICMP folding for zext and sext in SCCPSolver (PR #67594)

leo-ard wrote: I'm closing this PR in favour of #70845 https://github.com/llvm/llvm-project/pull/67594 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [llvm] [Instcombine] use zext's nneg flag for icmp folding (PR #70845)

https://github.com/leo-ard updated https://github.com/llvm/llvm-project/pull/70845 From 6bb97fd48d59b7f79fdf90a2b27e9220f417fac7 Mon Sep 17 00:00:00 2001 From: leo-ard Date: Mon, 30 Oct 2023 18:01:27 -0400 Subject: [PATCH 1/8] Add NonNeg check for InstCombine --- llvm/lib/Transforms/InstCombi

[llvm] [clang] [Instcombine] use zext's nneg flag for icmp folding (PR #70845)

@@ -0,0 +1,19 @@ +// RUN: %clang_cc1 %s -O2 -triple=x86_64-apple-darwin -emit-llvm -o - | FileCheck %s leo-ard wrote: Out of curiosity, why is end-to-end codegen with clang not used ? Is it too unstable ? https://github.com/llvm/llvm-project/pull/70845 ___

[clang] [llvm] [Instcombine] use zext's nneg flag for icmp folding (PR #70845)

@@ -0,0 +1,185 @@ +; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --tool ./bin/opt --version 3 +; See PRXXX for more details +; RUN-./bin/opt: opt < %s -S -passes=ipsccp | FileCheck %s leo-ard wrote: yep, a typo on my end ht

[llvm] [clang] [Instcombine] use zext's nneg flag for icmp folding (PR #70845)

leo-ard wrote: Thanks for taking the time to review the PR. I just added another test in PhaseOrdering to make sure that the min/max intrinsics are generated https://github.com/llvm/llvm-project/pull/70845 ___ cfe-commits mailing list cfe-commits@list

[llvm] [clang] [Instcombine] use zext's nneg flag for icmp folding (PR #70845)

https://github.com/leo-ard updated https://github.com/llvm/llvm-project/pull/70845 From 6bb97fd48d59b7f79fdf90a2b27e9220f417fac7 Mon Sep 17 00:00:00 2001 From: leo-ard Date: Mon, 30 Oct 2023 18:01:27 -0400 Subject: [PATCH 1/9] Add NonNeg check for InstCombine --- llvm/lib/Transforms/InstCombi

[clang] [SCCP] [Transform] Adding ICMP folding for zext and sext in SCCPSolver (PR #67594)

https://github.com/leo-ard created https://github.com/llvm/llvm-project/pull/67594 This PR fixes #55013 : the max intrinsics is not generated for this simple loop case : https://godbolt.org/z/hxz1xhMPh. This is caused by a ICMP not being folded into a select, thus not generating the max intrin

[clang] [SCCP] [Transform] Adding ICMP folding for zext and sext in SCCPSolver (PR #67594)

https://github.com/leo-ard updated https://github.com/llvm/llvm-project/pull/67594 From 399f9d64cfde0761ac8278dd05ba704d879b1f5a Mon Sep 17 00:00:00 2001 From: leo-ard Date: Wed, 27 Sep 2023 13:35:53 -0400 Subject: [PATCH 1/2] Adding ICMP folding for SCCPSolver --- clang/test/CodeGen/X86/min_

[clang] [SCCP] [Transform] Adding ICMP folding for zext and sext in SCCPSolver (PR #67594)

leo-ard wrote: > If I understood correctly, what you're trying to do here is to apply an icmp > fold early before the sext -> zext information gets lost. I don't think this > is the correct way to approach the problem. The correct way is to preserve > the fact that the operand is non-negative