AMDGPU: Reduce AddedComplexity on canonicalize pattern

[arsenm]

Pick the minimum complexity required for tests to pass instead of
a giant debug value of 1000.

[arsenm]

• AMDGPU: Reduce AddedComplexity on canonicalize pattern #119796 👈 (View in Graphite)
• AMDGPU: Remove large, negative AddedComplexity from minimum/maximum patterns #119795
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

[llvmbot]

@llvm/pr-subscribers-backend-amdgpu

Author: Matt Arsenault (arsenm)

Changes

Pick the minimum complexity required for tests to pass instead of
a giant debug value of 1000.

---

Full diff: https://github.com/llvm/llvm-project/pull/119796.diff

1 Files Affected:

• (modified) llvm/lib/Target/AMDGPU/SIInstructions.td (+1-1)

diff --git a/llvm/lib/Target/AMDGPU/SIInstructions.td b/llvm/lib/Target/AMDGPU/SIInstructions.td
index bc25d75131cc35..9bd04b939c8a55 100644
--- a/llvm/lib/Target/AMDGPU/SIInstructions.td
+++ b/llvm/lib/Target/AMDGPU/SIInstructions.td
@@ -3055,7 +3055,7 @@ def : GCNPat<
    (V_BFREV_B32_e64 (i32 (EXTRACT_SUBREG VReg_64:$a, sub0))), sub1)>;
 
 // If fcanonicalize's operand is implicitly canonicalized, we only need a copy.
-let AddedComplexity = 1000 in {
+let AddedComplexity = 8 in {
 foreach vt = [f16, v2f16, f32, v2f32, f64] in {
   def : GCNPat<
     (fcanonicalize (vt is_canonicalized:$src)),

[arsenm]

ping