[llvm] [AArch64] Fix scheduling information for STNT1H (PR #164780)
via llvm-commits
llvm-commits at lists.llvm.org
Thu Oct 23 02:16:55 PDT 2025
llvmbot wrote:
<!--LLVM PR SUMMARY COMMENT-->
@llvm/pr-subscribers-backend-aarch64
Author: Simon Wallis (simonwallis2)
<details>
<summary>Changes</summary>
Fix 3 cases in the scheduler tables to match the current SWOG, in section 3.29 SVE Store.
---
Full diff: https://github.com/llvm/llvm-project/pull/164780.diff
2 Files Affected:
- (modified) llvm/lib/Target/AArch64/AArch64SchedNeoverseV2.td (+3-3)
- (modified) llvm/test/tools/llvm-mca/AArch64/Neoverse/V2-sve-instructions.s (+17-17)
``````````diff
diff --git a/llvm/lib/Target/AArch64/AArch64SchedNeoverseV2.td b/llvm/lib/Target/AArch64/AArch64SchedNeoverseV2.td
index bdde8e388cccc..2387f176f3051 100644
--- a/llvm/lib/Target/AArch64/AArch64SchedNeoverseV2.td
+++ b/llvm/lib/Target/AArch64/AArch64SchedNeoverseV2.td
@@ -2762,11 +2762,11 @@ def : InstRW<[V2Write_11c_18L01_18V01], (instregex "^ST4[BHWD]_IMM$")>;
def : InstRW<[V2Write_11c_18L01_18S_18V01], (instregex "^ST4[BHWD]$")>;
// Non temporal store, scalar + imm
-def : InstRW<[V2Write_2c_1L01_1V], (instregex "^STNT1[BHWD]_ZRI$")>;
+def : InstRW<[V2Write_2c_1L01_1V01], (instregex "^STNT1[BHWD]_ZRI$")>;
// Non temporal store, scalar + scalar
-def : InstRW<[V2Write_2c_1L01_1S_1V], (instrs STNT1H_ZRR)>;
-def : InstRW<[V2Write_2c_1L01_1V], (instregex "^STNT1[BWD]_ZRR$")>;
+def : InstRW<[V2Write_2c_1L01_1S_1V01], (instrs STNT1H_ZRR)>;
+def : InstRW<[V2Write_2c_1L01_1V01], (instregex "^STNT1[BWD]_ZRR$")>;
// Scatter non temporal store, vector + scalar 32-bit element size
def : InstRW<[V2Write_4c_4L01_4V01], (instregex "^STNT1[BHW]_ZZR_S")>;
diff --git a/llvm/test/tools/llvm-mca/AArch64/Neoverse/V2-sve-instructions.s b/llvm/test/tools/llvm-mca/AArch64/Neoverse/V2-sve-instructions.s
index 49af4df3d8ff5..c20409e783013 100644
--- a/llvm/test/tools/llvm-mca/AArch64/Neoverse/V2-sve-instructions.s
+++ b/llvm/test/tools/llvm-mca/AArch64/Neoverse/V2-sve-instructions.s
@@ -6864,7 +6864,7 @@ zip2 z31.s, z31.s, z31.s
# CHECK: Resource pressure per iteration:
# CHECK-NEXT: [0.0] [0.1] [1.0] [1.1] [2.0] [2.1] [2.2] [3] [4.0] [4.1] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14]
-# CHECK-NEXT: - - - - - - - 245.00 651.00 651.00 570.50 272.50 83.75 83.75 81.75 81.75 1536.75 1281.75 794.25 748.25
+# CHECK-NEXT: - - - - - - - 245.00 651.00 651.00 570.50 272.50 83.75 83.75 81.75 81.75 1540.75 1285.75 790.25 744.25
# CHECK: Resource pressure by instruction:
# CHECK-NEXT: [0.0] [0.1] [1.0] [1.1] [2.0] [2.1] [2.2] [3] [4.0] [4.1] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] Instructions:
@@ -9617,39 +9617,39 @@ zip2 z31.s, z31.s, z31.s
# CHECK-NEXT: - - - - - - - - 9.00 9.00 - - - - - - 9.00 9.00 - - st4w { z21.s - z24.s }, p5, [x10, #20, mul vl]
# CHECK-NEXT: - - - - - - - - 9.00 9.00 - - - - - - 9.00 9.00 - - st4w { z23.s - z26.s }, p3, [x13, #-32, mul vl]
# CHECK-NEXT: - - - - - - - - 9.00 9.00 - - 4.50 4.50 4.50 4.50 9.00 9.00 - - st4w { z5.s - z8.s }, p3, [x17, x16, lsl #2]
-# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.25 0.25 0.25 0.25 stnt1b { z0.b }, p0, [x0, x0]
-# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.25 0.25 0.25 0.25 stnt1b { z0.b }, p0, [x0]
+# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.50 0.50 - - stnt1b { z0.b }, p0, [x0, x0]
+# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.50 0.50 - - stnt1b { z0.b }, p0, [x0]
# CHECK-NEXT: - - - - - - - - 1.00 1.00 - - - - - - 1.00 1.00 - - stnt1b { z0.d }, p0, [z1.d]
# CHECK-NEXT: - - - - - - - - 2.00 2.00 - - - - - - 2.00 2.00 - - stnt1b { z0.s }, p0, [z1.s]
-# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.25 0.25 0.25 0.25 stnt1b { z21.b }, p5, [x10, #7, mul vl]
-# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.25 0.25 0.25 0.25 stnt1b { z23.b }, p3, [x13, #-8, mul vl]
+# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.50 0.50 - - stnt1b { z21.b }, p5, [x10, #7, mul vl]
+# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.50 0.50 - - stnt1b { z23.b }, p3, [x13, #-8, mul vl]
# CHECK-NEXT: - - - - - - - - 1.00 1.00 - - - - - - 1.00 1.00 - - stnt1b { z31.d }, p7, [z31.d, x0]
# CHECK-NEXT: - - - - - - - - 1.00 1.00 - - - - - - 1.00 1.00 - - stnt1b { z31.d }, p7, [z31.d]
# CHECK-NEXT: - - - - - - - - 2.00 2.00 - - - - - - 2.00 2.00 - - stnt1b { z31.s }, p7, [z31.s, x0]
# CHECK-NEXT: - - - - - - - - 2.00 2.00 - - - - - - 2.00 2.00 - - stnt1b { z31.s }, p7, [z31.s]
-# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.25 0.25 0.25 0.25 stnt1d { z0.d }, p0, [x0, x0, lsl #3]
-# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.25 0.25 0.25 0.25 stnt1d { z0.d }, p0, [x0]
+# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.50 0.50 - - stnt1d { z0.d }, p0, [x0, x0, lsl #3]
+# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.50 0.50 - - stnt1d { z0.d }, p0, [x0]
# CHECK-NEXT: - - - - - - - - 1.00 1.00 - - - - - - 1.00 1.00 - - stnt1d { z0.d }, p0, [z1.d]
-# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.25 0.25 0.25 0.25 stnt1d { z21.d }, p5, [x10, #7, mul vl]
-# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.25 0.25 0.25 0.25 stnt1d { z23.d }, p3, [x13, #-8, mul vl]
+# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.50 0.50 - - stnt1d { z21.d }, p5, [x10, #7, mul vl]
+# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.50 0.50 - - stnt1d { z23.d }, p3, [x13, #-8, mul vl]
# CHECK-NEXT: - - - - - - - - 1.00 1.00 - - - - - - 1.00 1.00 - - stnt1d { z31.d }, p7, [z31.d, x0]
# CHECK-NEXT: - - - - - - - - 1.00 1.00 - - - - - - 1.00 1.00 - - stnt1d { z31.d }, p7, [z31.d]
# CHECK-NEXT: - - - - - - - - 1.00 1.00 - - - - - - 1.00 1.00 - - stnt1h { z0.d }, p0, [z1.d]
-# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 stnt1h { z0.h }, p0, [x0, x0, lsl #1]
-# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.25 0.25 0.25 0.25 stnt1h { z0.h }, p0, [x0]
+# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - 0.25 0.25 0.25 0.25 0.50 0.50 - - stnt1h { z0.h }, p0, [x0, x0, lsl #1]
+# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.50 0.50 - - stnt1h { z0.h }, p0, [x0]
# CHECK-NEXT: - - - - - - - - 2.00 2.00 - - - - - - 2.00 2.00 - - stnt1h { z0.s }, p0, [z1.s]
-# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.25 0.25 0.25 0.25 stnt1h { z21.h }, p5, [x10, #7, mul vl]
-# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.25 0.25 0.25 0.25 stnt1h { z23.h }, p3, [x13, #-8, mul vl]
+# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.50 0.50 - - stnt1h { z21.h }, p5, [x10, #7, mul vl]
+# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.50 0.50 - - stnt1h { z23.h }, p3, [x13, #-8, mul vl]
# CHECK-NEXT: - - - - - - - - 1.00 1.00 - - - - - - 1.00 1.00 - - stnt1h { z31.d }, p7, [z31.d, x0]
# CHECK-NEXT: - - - - - - - - 1.00 1.00 - - - - - - 1.00 1.00 - - stnt1h { z31.d }, p7, [z31.d]
# CHECK-NEXT: - - - - - - - - 2.00 2.00 - - - - - - 2.00 2.00 - - stnt1h { z31.s }, p7, [z31.s, x0]
# CHECK-NEXT: - - - - - - - - 2.00 2.00 - - - - - - 2.00 2.00 - - stnt1h { z31.s }, p7, [z31.s]
# CHECK-NEXT: - - - - - - - - 1.00 1.00 - - - - - - 1.00 1.00 - - stnt1w { z0.d }, p0, [z1.d]
-# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.25 0.25 0.25 0.25 stnt1w { z0.s }, p0, [x0, x0, lsl #2]
-# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.25 0.25 0.25 0.25 stnt1w { z0.s }, p0, [x0]
+# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.50 0.50 - - stnt1w { z0.s }, p0, [x0, x0, lsl #2]
+# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.50 0.50 - - stnt1w { z0.s }, p0, [x0]
# CHECK-NEXT: - - - - - - - - 2.00 2.00 - - - - - - 2.00 2.00 - - stnt1w { z0.s }, p0, [z1.s]
-# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.25 0.25 0.25 0.25 stnt1w { z21.s }, p5, [x10, #7, mul vl]
-# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.25 0.25 0.25 0.25 stnt1w { z23.s }, p3, [x13, #-8, mul vl]
+# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.50 0.50 - - stnt1w { z21.s }, p5, [x10, #7, mul vl]
+# CHECK-NEXT: - - - - - - - - 0.50 0.50 - - - - - - 0.50 0.50 - - stnt1w { z23.s }, p3, [x13, #-8, mul vl]
# CHECK-NEXT: - - - - - - - - 1.00 1.00 - - - - - - 1.00 1.00 - - stnt1w { z31.d }, p7, [z31.d, x0]
# CHECK-NEXT: - - - - - - - - 1.00 1.00 - - - - - - 1.00 1.00 - - stnt1w { z31.d }, p7, [z31.d]
# CHECK-NEXT: - - - - - - - - 2.00 2.00 - - - - - - 2.00 2.00 - - stnt1w { z31.s }, p7, [z31.s, x0]
``````````
</details>
https://github.com/llvm/llvm-project/pull/164780
More information about the llvm-commits
mailing list