[RISCV][VLOPT] Add Vector Single-Width Averaging Add and Subtract to isSupportedInstr #122351

michaelmaitland · 2025-01-09T19:39:02Z

No description provided.

llvmbot · 2025-01-09T19:39:25Z

@llvm/pr-subscribers-backend-risc-v

Author: Michael Maitland (michaelmaitland)

Changes

Full diff: https://github.com/llvm/llvm-project/pull/122351.diff

2 Files Affected:

(modified) llvm/lib/Target/RISCV/RISCVVLOptimizer.cpp (+9)
(modified) llvm/test/CodeGen/RISCV/rvv/vl-opt-instrs.ll (+176)

diff --git a/llvm/lib/Target/RISCV/RISCVVLOptimizer.cpp b/llvm/lib/Target/RISCV/RISCVVLOptimizer.cpp
index 6661921d66f95e..448f0645e08395 100644
--- a/llvm/lib/Target/RISCV/RISCVVLOptimizer.cpp
+++ b/llvm/lib/Target/RISCV/RISCVVLOptimizer.cpp
@@ -959,6 +959,15 @@ static bool isSupportedInstr(const MachineInstr &MI) {
   case RISCV::VMV_V_I:
   case RISCV::VMV_V_X:
   case RISCV::VMV_V_V:
+  // Vector Single-Width Averaging Add and Subtract
+  case RISCV::VAADDU_VV:
+  case RISCV::VAADDU_VX:
+  case RISCV::VAADD_VV:
+  case RISCV::VAADD_VX:
+  case RISCV::VASUBU_VV:
+  case RISCV::VASUBU_VX:
+  case RISCV::VASUB_VV:
+  case RISCV::VASUB_VX:
 
   // Vector Crypto
   case RISCV::VWSLL_VI:
diff --git a/llvm/test/CodeGen/RISCV/rvv/vl-opt-instrs.ll b/llvm/test/CodeGen/RISCV/rvv/vl-opt-instrs.ll
index 46fbba35c35a2b..3040d7e691b32a 100644
--- a/llvm/test/CodeGen/RISCV/rvv/vl-opt-instrs.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/vl-opt-instrs.ll
@@ -3535,3 +3535,179 @@ define <vscale x 4 x i1> @vmfgt_vv(<vscale x 4 x float> %a, <vscale x 4 x i1> %b
   %2 = call <vscale x 4 x i1> @llvm.riscv.vmand.nxv4i1(<vscale x 4 x i1> %1, <vscale x 4 x i1> %b, iXLen %vl)
   ret <vscale x 4 x i1> %2
 }
+
+define <vscale x 4 x i32> @vaadd_vv(<vscale x 4 x i32> %a, <vscale x 4 x i32> %b, iXLen %vl) {
+; NOVLOPT-LABEL: vaadd_vv:
+; NOVLOPT:       # %bb.0:
+; NOVLOPT-NEXT:    csrwi vxrm, 0
+; NOVLOPT-NEXT:    vsetvli a1, zero, e32, m2, ta, ma
+; NOVLOPT-NEXT:    vaadd.vv v8, v8, v10
+; NOVLOPT-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; NOVLOPT-NEXT:    vadd.vv v8, v8, v10
+; NOVLOPT-NEXT:    ret
+;
+; VLOPT-LABEL: vaadd_vv:
+; VLOPT:       # %bb.0:
+; VLOPT-NEXT:    csrwi vxrm, 0
+; VLOPT-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; VLOPT-NEXT:    vaadd.vv v8, v8, v10
+; VLOPT-NEXT:    vadd.vv v8, v8, v10
+; VLOPT-NEXT:    ret
+  %1 = call <vscale x 4 x i32> @llvm.riscv.vaadd.nxv4i32.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i32> %a, <vscale x 4 x i32> %b, iXLen 0, iXLen -1)
+  %2 = call <vscale x 4 x i32> @llvm.riscv.vadd.nxv4i32.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i32> %1, <vscale x 4 x i32> %b, iXLen %vl)
+  ret <vscale x 4 x i32> %2
+}
+
+define <vscale x 4 x i32> @vaadd_vx(<vscale x 4 x i32> %a, i32 %b, iXLen %vl) {
+; NOVLOPT-LABEL: vaadd_vx:
+; NOVLOPT:       # %bb.0:
+; NOVLOPT-NEXT:    csrwi vxrm, 0
+; NOVLOPT-NEXT:    vsetvli a2, zero, e32, m2, ta, ma
+; NOVLOPT-NEXT:    vaadd.vx v10, v8, a0
+; NOVLOPT-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; NOVLOPT-NEXT:    vadd.vv v8, v10, v8
+; NOVLOPT-NEXT:    ret
+;
+; VLOPT-LABEL: vaadd_vx:
+; VLOPT:       # %bb.0:
+; VLOPT-NEXT:    csrwi vxrm, 0
+; VLOPT-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; VLOPT-NEXT:    vaadd.vx v10, v8, a0
+; VLOPT-NEXT:    vadd.vv v8, v10, v8
+; VLOPT-NEXT:    ret
+  %1 = call <vscale x 4 x i32> @llvm.riscv.vaadd.nxv4i32.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i32> %a, i32 %b, iXLen 0, iXLen -1)
+  %2 = call <vscale x 4 x i32> @llvm.riscv.vadd.nxv4i32.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i32> %1, <vscale x 4 x i32> %a, iXLen %vl)
+  ret <vscale x 4 x i32> %2
+}
+
+define <vscale x 4 x i32> @vasub_vv(<vscale x 4 x i32> %a, <vscale x 4 x i32> %b, iXLen %vl) {
+; NOVLOPT-LABEL: vasub_vv:
+; NOVLOPT:       # %bb.0:
+; NOVLOPT-NEXT:    csrwi vxrm, 0
+; NOVLOPT-NEXT:    vsetvli a1, zero, e32, m2, ta, ma
+; NOVLOPT-NEXT:    vasub.vv v8, v8, v10
+; NOVLOPT-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; NOVLOPT-NEXT:    vadd.vv v8, v8, v10
+; NOVLOPT-NEXT:    ret
+;
+; VLOPT-LABEL: vasub_vv:
+; VLOPT:       # %bb.0:
+; VLOPT-NEXT:    csrwi vxrm, 0
+; VLOPT-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; VLOPT-NEXT:    vasub.vv v8, v8, v10
+; VLOPT-NEXT:    vadd.vv v8, v8, v10
+; VLOPT-NEXT:    ret
+  %1 = call <vscale x 4 x i32> @llvm.riscv.vasub.nxv4i32.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i32> %a, <vscale x 4 x i32> %b, iXLen 0, iXLen -1)
+  %2 = call <vscale x 4 x i32> @llvm.riscv.vadd.nxv4i32.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i32> %1, <vscale x 4 x i32> %b, iXLen %vl)
+  ret <vscale x 4 x i32> %2
+}
+
+define <vscale x 4 x i32> @vasub_vx(<vscale x 4 x i32> %a, i32 %b, iXLen %vl) {
+; NOVLOPT-LABEL: vasub_vx:
+; NOVLOPT:       # %bb.0:
+; NOVLOPT-NEXT:    csrwi vxrm, 0
+; NOVLOPT-NEXT:    vsetvli a2, zero, e32, m2, ta, ma
+; NOVLOPT-NEXT:    vasub.vx v10, v8, a0
+; NOVLOPT-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; NOVLOPT-NEXT:    vadd.vv v8, v10, v8
+; NOVLOPT-NEXT:    ret
+;
+; VLOPT-LABEL: vasub_vx:
+; VLOPT:       # %bb.0:
+; VLOPT-NEXT:    csrwi vxrm, 0
+; VLOPT-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; VLOPT-NEXT:    vasub.vx v10, v8, a0
+; VLOPT-NEXT:    vadd.vv v8, v10, v8
+; VLOPT-NEXT:    ret
+  %1 = call <vscale x 4 x i32> @llvm.riscv.vasub.nxv4i32.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i32> %a, i32 %b, iXLen 0, iXLen -1)
+  %2 = call <vscale x 4 x i32> @llvm.riscv.vadd.nxv4i32.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i32> %1, <vscale x 4 x i32> %a, iXLen %vl)
+  ret <vscale x 4 x i32> %2
+}
+
+define <vscale x 4 x i32> @vaaddu_vv(<vscale x 4 x i32> %a, <vscale x 4 x i32> %b, iXLen %vl) {
+; NOVLOPT-LABEL: vaaddu_vv:
+; NOVLOPT:       # %bb.0:
+; NOVLOPT-NEXT:    csrwi vxrm, 0
+; NOVLOPT-NEXT:    vsetvli a1, zero, e32, m2, ta, ma
+; NOVLOPT-NEXT:    vaaddu.vv v8, v8, v10
+; NOVLOPT-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; NOVLOPT-NEXT:    vadd.vv v8, v8, v10
+; NOVLOPT-NEXT:    ret
+;
+; VLOPT-LABEL: vaaddu_vv:
+; VLOPT:       # %bb.0:
+; VLOPT-NEXT:    csrwi vxrm, 0
+; VLOPT-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; VLOPT-NEXT:    vaaddu.vv v8, v8, v10
+; VLOPT-NEXT:    vadd.vv v8, v8, v10
+; VLOPT-NEXT:    ret
+  %1 = call <vscale x 4 x i32> @llvm.riscv.vaaddu.nxv4i32.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i32> %a, <vscale x 4 x i32> %b, iXLen 0, iXLen -1)
+  %2 = call <vscale x 4 x i32> @llvm.riscv.vadd.nxv4i32.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i32> %1, <vscale x 4 x i32> %b, iXLen %vl)
+  ret <vscale x 4 x i32> %2
+}
+
+define <vscale x 4 x i32> @vaaddu_vx(<vscale x 4 x i32> %a, i32 %b, iXLen %vl) {
+; NOVLOPT-LABEL: vaaddu_vx:
+; NOVLOPT:       # %bb.0:
+; NOVLOPT-NEXT:    csrwi vxrm, 0
+; NOVLOPT-NEXT:    vsetvli a2, zero, e32, m2, ta, ma
+; NOVLOPT-NEXT:    vaaddu.vx v10, v8, a0
+; NOVLOPT-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; NOVLOPT-NEXT:    vadd.vv v8, v10, v8
+; NOVLOPT-NEXT:    ret
+;
+; VLOPT-LABEL: vaaddu_vx:
+; VLOPT:       # %bb.0:
+; VLOPT-NEXT:    csrwi vxrm, 0
+; VLOPT-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; VLOPT-NEXT:    vaaddu.vx v10, v8, a0
+; VLOPT-NEXT:    vadd.vv v8, v10, v8
+; VLOPT-NEXT:    ret
+  %1 = call <vscale x 4 x i32> @llvm.riscv.vaaddu.nxv4i32.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i32> %a, i32 %b, iXLen 0, iXLen -1)
+  %2 = call <vscale x 4 x i32> @llvm.riscv.vadd.nxv4i32.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i32> %1, <vscale x 4 x i32> %a, iXLen %vl)
+  ret <vscale x 4 x i32> %2
+}
+
+define <vscale x 4 x i32> @vasubu_vv(<vscale x 4 x i32> %a, <vscale x 4 x i32> %b, iXLen %vl) {
+; NOVLOPT-LABEL: vasubu_vv:
+; NOVLOPT:       # %bb.0:
+; NOVLOPT-NEXT:    csrwi vxrm, 0
+; NOVLOPT-NEXT:    vsetvli a1, zero, e32, m2, ta, ma
+; NOVLOPT-NEXT:    vasubu.vv v8, v8, v10
+; NOVLOPT-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; NOVLOPT-NEXT:    vadd.vv v8, v8, v10
+; NOVLOPT-NEXT:    ret
+;
+; VLOPT-LABEL: vasubu_vv:
+; VLOPT:       # %bb.0:
+; VLOPT-NEXT:    csrwi vxrm, 0
+; VLOPT-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; VLOPT-NEXT:    vasubu.vv v8, v8, v10
+; VLOPT-NEXT:    vadd.vv v8, v8, v10
+; VLOPT-NEXT:    ret
+  %1 = call <vscale x 4 x i32> @llvm.riscv.vasubu.nxv4i32.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i32> %a, <vscale x 4 x i32> %b, iXLen 0, iXLen -1)
+  %2 = call <vscale x 4 x i32> @llvm.riscv.vadd.nxv4i32.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i32> %1, <vscale x 4 x i32> %b, iXLen %vl)
+  ret <vscale x 4 x i32> %2
+}
+
+define <vscale x 4 x i32> @vasubu_vx(<vscale x 4 x i32> %a, i32 %b, iXLen %vl) {
+; NOVLOPT-LABEL: vasubu_vx:
+; NOVLOPT:       # %bb.0:
+; NOVLOPT-NEXT:    csrwi vxrm, 0
+; NOVLOPT-NEXT:    vsetvli a2, zero, e32, m2, ta, ma
+; NOVLOPT-NEXT:    vasubu.vx v10, v8, a0
+; NOVLOPT-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; NOVLOPT-NEXT:    vadd.vv v8, v10, v8
+; NOVLOPT-NEXT:    ret
+;
+; VLOPT-LABEL: vasubu_vx:
+; VLOPT:       # %bb.0:
+; VLOPT-NEXT:    csrwi vxrm, 0
+; VLOPT-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; VLOPT-NEXT:    vasubu.vx v10, v8, a0
+; VLOPT-NEXT:    vadd.vv v8, v10, v8
+; VLOPT-NEXT:    ret
+  %1 = call <vscale x 4 x i32> @llvm.riscv.vasubu.nxv4i32.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i32> %a, i32 %b, iXLen 0, iXLen -1)
+  %2 = call <vscale x 4 x i32> @llvm.riscv.vadd.nxv4i32.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i32> %1, <vscale x 4 x i32> %a, iXLen %vl)
+  ret <vscale x 4 x i32> %2
+}

preames

LGTM

…isSupportedInstr

llvm-ci · 2025-01-10T05:34:56Z

LLVM Buildbot has detected a new failure on builder llvm-clang-x86_64-expensive-checks-debian running on gribozavr4 while building llvm at step 6 "test-build-unified-tree-check-all".

Full details are available at: https://lab.llvm.org/buildbot/#/builders/16/builds/11732

Here is the relevant piece of the build log for the reference

Step 6 (test-build-unified-tree-check-all) failure: test (failure)
******************** TEST 'LLVM :: tools/llvm-gsymutil/ARM_AArch64/macho-merged-funcs-dwarf.yaml' FAILED ********************
Exit Code: 1

Command Output (stdout):
--
Input file: /b/1/llvm-clang-x86_64-expensive-checks-debian/build/test/tools/llvm-gsymutil/ARM_AArch64/Output/macho-merged-funcs-dwarf.yaml.tmp.dSYM
Output file (aarch64): /b/1/llvm-clang-x86_64-expensive-checks-debian/build/test/tools/llvm-gsymutil/ARM_AArch64/Output/macho-merged-funcs-dwarf.yaml.tmp.default.gSYM
Loaded 3 functions from DWARF.
Loaded 3 functions from symbol table.
warning: same address range contains different debug info. Removing:
[0x0000000000000248 - 0x0000000000000270): Name=0x00000047
addr=0x0000000000000248, file=  3, line=  5
addr=0x0000000000000254, file=  3, line=  7
addr=0x0000000000000258, file=  3, line=  9
addr=0x000000000000025c, file=  3, line=  8
addr=0x0000000000000260, file=  3, line= 11
addr=0x0000000000000264, file=  3, line= 10
addr=0x0000000000000268, file=  3, line=  6


In favor of this one:
[0x0000000000000248 - 0x0000000000000270): Name=0x00000030
addr=0x0000000000000248, file=  2, line=  5
addr=0x0000000000000254, file=  2, line=  7
addr=0x0000000000000258, file=  2, line=  9
addr=0x000000000000025c, file=  2, line=  8
addr=0x0000000000000260, file=  2, line= 11
addr=0x0000000000000264, file=  2, line= 10
addr=0x0000000000000268, file=  2, line=  6


warning: same address range contains different debug info. Removing:
[0x0000000000000248 - 0x0000000000000270): Name=0x00000030
addr=0x0000000000000248, file=  2, line=  5
addr=0x0000000000000254, file=  2, line=  7
addr=0x0000000000000258, file=  2, line=  9
addr=0x000000000000025c, file=  2, line=  8
addr=0x0000000000000260, file=  2, line= 11
addr=0x0000000000000264, file=  2, line= 10
addr=0x0000000000000268, file=  2, line=  6


In favor of this one:
[0x0000000000000248 - 0x0000000000000270): Name=0x00000001
addr=0x0000000000000248, file=  1, line=  5
addr=0x0000000000000254, file=  1, line=  7
addr=0x0000000000000258, file=  1, line=  9
addr=0x000000000000025c, file=  1, line=  8
addr=0x0000000000000260, file=  1, line= 11
addr=0x0000000000000264, file=  1, line= 10
...

…isSupportedInstr (llvm#122351)

michaelmaitland added the backend:RISC-V label Jan 9, 2025

michaelmaitland requested review from lukel97, preames and topperc January 9, 2025 19:39

preames approved these changes Jan 9, 2025

View reviewed changes

michaelmaitland changed the title ~~Add Vector Single-Width Averaging Add and Subtract to isSupportedInstr~~ [RISCV][VLOPT] Add Vector Single-Width Averaging Add and Subtract to isSupportedInstr Jan 10, 2025

[RISCV][VLOPT] Add Vector Single-Width Averaging Add and Subtract to …

06359f2

…isSupportedInstr

michaelmaitland force-pushed the vlopt-average-add-sub branch from 2d4a650 to 06359f2 Compare January 10, 2025 00:38

michaelmaitland merged commit 04e54cc into llvm:main Jan 10, 2025
4 of 6 checks passed

michaelmaitland deleted the vlopt-average-add-sub branch January 10, 2025 00:39

BaiXilin pushed a commit to BaiXilin/llvm-project that referenced this pull request Jan 12, 2025

[RISCV][VLOPT] Add Vector Single-Width Averaging Add and Subtract to …

021fc33

…isSupportedInstr (llvm#122351)

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[RISCV][VLOPT] Add Vector Single-Width Averaging Add and Subtract to isSupportedInstr #122351

[RISCV][VLOPT] Add Vector Single-Width Averaging Add and Subtract to isSupportedInstr #122351

Uh oh!

michaelmaitland commented Jan 9, 2025

Uh oh!

llvmbot commented Jan 9, 2025

Uh oh!

preames left a comment

Uh oh!

Uh oh!

llvm-ci commented Jan 10, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

[RISCV][VLOPT] Add Vector Single-Width Averaging Add and Subtract to isSupportedInstr #122351

[RISCV][VLOPT] Add Vector Single-Width Averaging Add and Subtract to isSupportedInstr #122351

Uh oh!

Conversation

michaelmaitland commented Jan 9, 2025

Uh oh!

llvmbot commented Jan 9, 2025

Uh oh!

preames left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

llvm-ci commented Jan 10, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants