[BasicTTIImpl] Add cost entries for ldexp, [l]lround #146373
Open
+84
−75
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
The ldexp intrinsic is incorrectly costed as 1, due to a missing entry in BasicTTIImpl::getTypeBasedIntrinsicCost: fix this. While at it, fix missing entries for [l]lround, which is costed as 1 anyway, making the change non-functional.
llvmbot
added
backend:AMDGPU
llvm:analysis
|
@llvm/pr-subscribers-backend-amdgpu @llvm/pr-subscribers-llvm-analysis Author: Ramkumar Ramachandra (artagnon) ChangesThe ldexp intrinsic is incorrectly costed as 1, due to a missing entry in BasicTTIImpl::getTypeBasedIntrinsicCost: fix this. While at it, fix missing entries for [l]lround, which is costed as 1 anyway, making the change non-functional. Patch is 35.23 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/146373.diff 3 Files Affected:
diff --git a/llvm/include/llvm/CodeGen/BasicTTIImpl.h b/llvm/include/llvm/CodeGen/BasicTTIImpl.h
index bf958e100f2ac..2b9be43eadb7a 100644
--- a/llvm/include/llvm/CodeGen/BasicTTIImpl.h
+++ b/llvm/include/llvm/CodeGen/BasicTTIImpl.h
@@ -2243,6 +2243,9 @@ class BasicTTIImplBase : public TargetTransformInfoImplCRTPBase<T> {
case Intrinsic::log2:
ISD = ISD::FLOG2;
break;
+ case Intrinsic::ldexp:
+ ISD = ISD::FLDEXP;
+ break;
case Intrinsic::fabs:
ISD = ISD::FABS;
break;
@@ -2297,6 +2300,12 @@ class BasicTTIImplBase : public TargetTransformInfoImplCRTPBase<T> {
case Intrinsic::roundeven:
ISD = ISD::FROUNDEVEN;
break;
+ case Intrinsic::lround:
+ ISD = ISD::LROUND;
+ break;
+ case Intrinsic::llround:
+ ISD = ISD::LLROUND;
+ break;
case Intrinsic::pow:
ISD = ISD::FPOW;
break;
diff --git a/llvm/test/Analysis/CostModel/AMDGPU/ldexp.ll b/llvm/test/Analysis/CostModel/AMDGPU/ldexp.ll
index 2b1b5906ad017..260cf696270cd 100644
--- a/llvm/test/Analysis/CostModel/AMDGPU/ldexp.ll
+++ b/llvm/test/Analysis/CostModel/AMDGPU/ldexp.ll
@@ -13,45 +13,45 @@ define void @ldexp_f16_i32() {
; GFX7-LABEL: 'ldexp_f16_i32'
; GFX7-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f16 = call half @llvm.ldexp.f16.i32(half undef, i32 undef)
; GFX7-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v2f16 = call <2 x half> @llvm.ldexp.v2f16.v2i32(<2 x half> undef, <2 x i32> undef)
-; GFX7-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %v3f16 = call <3 x half> @llvm.ldexp.v3f16.v3i32(<3 x half> undef, <3 x i32> undef)
+; GFX7-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v3f16 = call <3 x half> @llvm.ldexp.v3f16.v3i32(<3 x half> undef, <3 x i32> undef)
; GFX7-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v4f16 = call <4 x half> @llvm.ldexp.v4f16.v4i32(<4 x half> undef, <4 x i32> undef)
-; GFX7-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %v5f16 = call <5 x half> @llvm.ldexp.v5f16.v5i32(<5 x half> undef, <5 x i32> undef)
+; GFX7-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v5f16 = call <5 x half> @llvm.ldexp.v5f16.v5i32(<5 x half> undef, <5 x i32> undef)
; GFX7-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v8f16 = call <8 x half> @llvm.ldexp.v8f16.v8i32(<8 x half> undef, <8 x i32> undef)
; GFX7-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %v16f16 = call <16 x half> @llvm.ldexp.v16f16.v16i32(<16 x half> undef, <16 x i32> undef)
-; GFX7-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %v17f16 = call <17 x half> @llvm.ldexp.v17f16.v17i32(<17 x half> undef, <17 x i32> undef)
+; GFX7-NEXT: Cost Model: Found an estimated cost of 64 for instruction: %v17f16 = call <17 x half> @llvm.ldexp.v17f16.v17i32(<17 x half> undef, <17 x i32> undef)
; GFX7-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void
;
; GFX8PLUS-LABEL: 'ldexp_f16_i32'
-; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f16 = call half @llvm.ldexp.f16.i32(half undef, i32 undef)
-; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v2f16 = call <2 x half> @llvm.ldexp.v2f16.v2i32(<2 x half> undef, <2 x i32> undef)
-; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %v3f16 = call <3 x half> @llvm.ldexp.v3f16.v3i32(<3 x half> undef, <3 x i32> undef)
-; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %v4f16 = call <4 x half> @llvm.ldexp.v4f16.v4i32(<4 x half> undef, <4 x i32> undef)
-; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %v5f16 = call <5 x half> @llvm.ldexp.v5f16.v5i32(<5 x half> undef, <5 x i32> undef)
-; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %v8f16 = call <8 x half> @llvm.ldexp.v8f16.v8i32(<8 x half> undef, <8 x i32> undef)
-; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %v16f16 = call <16 x half> @llvm.ldexp.v16f16.v16i32(<16 x half> undef, <16 x i32> undef)
-; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 33 for instruction: %v17f16 = call <17 x half> @llvm.ldexp.v17f16.v17i32(<17 x half> undef, <17 x i32> undef)
+; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %f16 = call half @llvm.ldexp.f16.i32(half undef, i32 undef)
+; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %v2f16 = call <2 x half> @llvm.ldexp.v2f16.v2i32(<2 x half> undef, <2 x i32> undef)
+; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v3f16 = call <3 x half> @llvm.ldexp.v3f16.v3i32(<3 x half> undef, <3 x i32> undef)
+; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %v4f16 = call <4 x half> @llvm.ldexp.v4f16.v4i32(<4 x half> undef, <4 x i32> undef)
+; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %v5f16 = call <5 x half> @llvm.ldexp.v5f16.v5i32(<5 x half> undef, <5 x i32> undef)
+; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 23 for instruction: %v8f16 = call <8 x half> @llvm.ldexp.v8f16.v8i32(<8 x half> undef, <8 x i32> undef)
+; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 47 for instruction: %v16f16 = call <16 x half> @llvm.ldexp.v16f16.v16i32(<16 x half> undef, <16 x i32> undef)
+; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 50 for instruction: %v17f16 = call <17 x half> @llvm.ldexp.v17f16.v17i32(<17 x half> undef, <17 x i32> undef)
; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void
;
; GFX7-SIZE-LABEL: 'ldexp_f16_i32'
; GFX7-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f16 = call half @llvm.ldexp.f16.i32(half undef, i32 undef)
; GFX7-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v2f16 = call <2 x half> @llvm.ldexp.v2f16.v2i32(<2 x half> undef, <2 x i32> undef)
-; GFX7-SIZE-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %v3f16 = call <3 x half> @llvm.ldexp.v3f16.v3i32(<3 x half> undef, <3 x i32> undef)
+; GFX7-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v3f16 = call <3 x half> @llvm.ldexp.v3f16.v3i32(<3 x half> undef, <3 x i32> undef)
; GFX7-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v4f16 = call <4 x half> @llvm.ldexp.v4f16.v4i32(<4 x half> undef, <4 x i32> undef)
-; GFX7-SIZE-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %v5f16 = call <5 x half> @llvm.ldexp.v5f16.v5i32(<5 x half> undef, <5 x i32> undef)
+; GFX7-SIZE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v5f16 = call <5 x half> @llvm.ldexp.v5f16.v5i32(<5 x half> undef, <5 x i32> undef)
; GFX7-SIZE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v8f16 = call <8 x half> @llvm.ldexp.v8f16.v8i32(<8 x half> undef, <8 x i32> undef)
; GFX7-SIZE-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %v16f16 = call <16 x half> @llvm.ldexp.v16f16.v16i32(<16 x half> undef, <16 x i32> undef)
-; GFX7-SIZE-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %v17f16 = call <17 x half> @llvm.ldexp.v17f16.v17i32(<17 x half> undef, <17 x i32> undef)
+; GFX7-SIZE-NEXT: Cost Model: Found an estimated cost of 64 for instruction: %v17f16 = call <17 x half> @llvm.ldexp.v17f16.v17i32(<17 x half> undef, <17 x i32> undef)
; GFX7-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void
;
; GFX8PLUS-SIZE-LABEL: 'ldexp_f16_i32'
-; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f16 = call half @llvm.ldexp.f16.i32(half undef, i32 undef)
-; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v2f16 = call <2 x half> @llvm.ldexp.v2f16.v2i32(<2 x half> undef, <2 x i32> undef)
-; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %v3f16 = call <3 x half> @llvm.ldexp.v3f16.v3i32(<3 x half> undef, <3 x i32> undef)
-; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %v4f16 = call <4 x half> @llvm.ldexp.v4f16.v4i32(<4 x half> undef, <4 x i32> undef)
-; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %v5f16 = call <5 x half> @llvm.ldexp.v5f16.v5i32(<5 x half> undef, <5 x i32> undef)
-; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %v8f16 = call <8 x half> @llvm.ldexp.v8f16.v8i32(<8 x half> undef, <8 x i32> undef)
-; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %v16f16 = call <16 x half> @llvm.ldexp.v16f16.v16i32(<16 x half> undef, <16 x i32> undef)
-; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 33 for instruction: %v17f16 = call <17 x half> @llvm.ldexp.v17f16.v17i32(<17 x half> undef, <17 x i32> undef)
+; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %f16 = call half @llvm.ldexp.f16.i32(half undef, i32 undef)
+; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %v2f16 = call <2 x half> @llvm.ldexp.v2f16.v2i32(<2 x half> undef, <2 x i32> undef)
+; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v3f16 = call <3 x half> @llvm.ldexp.v3f16.v3i32(<3 x half> undef, <3 x i32> undef)
+; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %v4f16 = call <4 x half> @llvm.ldexp.v4f16.v4i32(<4 x half> undef, <4 x i32> undef)
+; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %v5f16 = call <5 x half> @llvm.ldexp.v5f16.v5i32(<5 x half> undef, <5 x i32> undef)
+; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 23 for instruction: %v8f16 = call <8 x half> @llvm.ldexp.v8f16.v8i32(<8 x half> undef, <8 x i32> undef)
+; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 47 for instruction: %v16f16 = call <16 x half> @llvm.ldexp.v16f16.v16i32(<16 x half> undef, <16 x i32> undef)
+; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 50 for instruction: %v17f16 = call <17 x half> @llvm.ldexp.v17f16.v17i32(<17 x half> undef, <17 x i32> undef)
; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void
;
%f16 = call half @llvm.ldexp.f16.i32(half undef, i32 undef)
@@ -69,45 +69,45 @@ define void @ldexp_f16_i16() {
; GFX7-LABEL: 'ldexp_f16_i16'
; GFX7-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f16 = call half @llvm.ldexp.f16.i16(half undef, i16 undef)
; GFX7-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v2f16 = call <2 x half> @llvm.ldexp.v2f16.v2i16(<2 x half> undef, <2 x i16> undef)
-; GFX7-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %v3f16 = call <3 x half> @llvm.ldexp.v3f16.v3i16(<3 x half> undef, <3 x i16> undef)
+; GFX7-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v3f16 = call <3 x half> @llvm.ldexp.v3f16.v3i16(<3 x half> undef, <3 x i16> undef)
; GFX7-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v4f16 = call <4 x half> @llvm.ldexp.v4f16.v4i16(<4 x half> undef, <4 x i16> undef)
-; GFX7-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %v5f16 = call <5 x half> @llvm.ldexp.v5f16.v5i16(<5 x half> undef, <5 x i16> undef)
+; GFX7-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v5f16 = call <5 x half> @llvm.ldexp.v5f16.v5i16(<5 x half> undef, <5 x i16> undef)
; GFX7-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v8f16 = call <8 x half> @llvm.ldexp.v8f16.v8i16(<8 x half> undef, <8 x i16> undef)
; GFX7-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %v16f16 = call <16 x half> @llvm.ldexp.v16f16.v16i16(<16 x half> undef, <16 x i16> undef)
-; GFX7-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %v17f16 = call <17 x half> @llvm.ldexp.v17f16.v17i16(<17 x half> undef, <17 x i16> undef)
+; GFX7-NEXT: Cost Model: Found an estimated cost of 64 for instruction: %v17f16 = call <17 x half> @llvm.ldexp.v17f16.v17i16(<17 x half> undef, <17 x i16> undef)
; GFX7-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void
;
; GFX8PLUS-LABEL: 'ldexp_f16_i16'
-; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f16 = call half @llvm.ldexp.f16.i16(half undef, i16 undef)
-; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v2f16 = call <2 x half> @llvm.ldexp.v2f16.v2i16(<2 x half> undef, <2 x i16> undef)
-; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %v3f16 = call <3 x half> @llvm.ldexp.v3f16.v3i16(<3 x half> undef, <3 x i16> undef)
-; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %v4f16 = call <4 x half> @llvm.ldexp.v4f16.v4i16(<4 x half> undef, <4 x i16> undef)
-; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %v5f16 = call <5 x half> @llvm.ldexp.v5f16.v5i16(<5 x half> undef, <5 x i16> undef)
-; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %v8f16 = call <8 x half> @llvm.ldexp.v8f16.v8i16(<8 x half> undef, <8 x i16> undef)
-; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %v16f16 = call <16 x half> @llvm.ldexp.v16f16.v16i16(<16 x half> undef, <16 x i16> undef)
-; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 33 for instruction: %v17f16 = call <17 x half> @llvm.ldexp.v17f16.v17i16(<17 x half> undef, <17 x i16> undef)
+; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %f16 = call half @llvm.ldexp.f16.i16(half undef, i16 undef)
+; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %v2f16 = call <2 x half> @llvm.ldexp.v2f16.v2i16(<2 x half> undef, <2 x i16> undef)
+; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v3f16 = call <3 x half> @llvm.ldexp.v3f16.v3i16(<3 x half> undef, <3 x i16> undef)
+; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %v4f16 = call <4 x half> @llvm.ldexp.v4f16.v4i16(<4 x half> undef, <4 x i16> undef)
+; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %v5f16 = call <5 x half> @llvm.ldexp.v5f16.v5i16(<5 x half> undef, <5 x i16> undef)
+; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 23 for instruction: %v8f16 = call <8 x half> @llvm.ldexp.v8f16.v8i16(<8 x half> undef, <8 x i16> undef)
+; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 47 for instruction: %v16f16 = call <16 x half> @llvm.ldexp.v16f16.v16i16(<16 x half> undef, <16 x i16> undef)
+; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 50 for instruction: %v17f16 = call <17 x half> @llvm.ldexp.v17f16.v17i16(<17 x half> undef, <17 x i16> undef)
; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void
;
; GFX7-SIZE-LABEL: 'ldexp_f16_i16'
; GFX7-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f16 = call half @llvm.ldexp.f16.i16(half undef, i16 undef)
; GFX7-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v2f16 = call <2 x half> @llvm.ldexp.v2f16.v2i16(<2 x half> undef, <2 x i16> undef)
-; GFX7-SIZE-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %v3f16 = call <3 x half> @llvm.ldexp.v3f16.v3i16(<3 x half> undef, <3 x i16> undef)
+; GFX7-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v3f16 = call <3 x half> @llvm.ldexp.v3f16.v3i16(<3 x half> undef, <3 x i16> undef)
; GFX7-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v4f16 = call <4 x half> @llvm.ldexp.v4f16.v4i16(<4 x half> undef, <4 x i16> undef)
-; GFX7-SIZE-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %v5f16 = call <5 x half> @llvm.ldexp.v5f16.v5i16(<5 x half> undef, <5 x i16> undef)
+; GFX7-SIZE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v5f16 = call <5 x half> @llvm.ldexp.v5f16.v5i16(<5 x half> undef, <5 x i16> undef)
; GFX7-SIZE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v8f16 = call <8 x half> @llvm.ldexp.v8f16.v8i16(<8 x half> undef, <8 x i16> undef)
; GFX7-SIZE-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %v16f16 = call <16 x half> @llvm.ldexp.v16f16.v16i16(<16 x half> undef, <16 x i16> undef)
-; GFX7-SIZE-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %v17f16 = call <17 x half> @llvm.ldexp.v17f16.v17i16(<17 x half> undef, <17 x i16> undef)
+; GFX7-SIZE-NEXT: Cost Model: Found an estimated cost of 64 for instruction: %v17f16 = call <17 x half> @llvm.ldexp.v17f16.v17i16(<17 x half> undef, <17 x i16> undef)
; GFX7-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void
;
; GFX8PLUS-SIZE-LABEL: 'ldexp_f16_i16'
-; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f16 = call half @llvm.ldexp.f16.i16(half undef, i16 undef)
-; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v2f16 = call <2 x half> @llvm.ldexp.v2f16.v2i16(<2 x half> undef, <2 x i16> undef)
-; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %v3f16 = call <3 x half> @llvm.ldexp.v3f16.v3i16(<3 x half> undef, <3 x i16> undef)
-; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %v4f16 = call <4 x half> @llvm.ldexp.v4f16.v4i16(<4 x half> undef, <4 x i16> undef)
-; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %v5f16 = call <5 x half> @llvm.ldexp.v5f16.v5i16(<5 x half> undef, <5 x i16> undef)
-; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %v8f16 = call <8 x half> @llvm.ldexp.v8f16.v8i16(<8 x half> undef, <8 x i16> undef)
-; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %v16f16 = call <16 x half> @llvm.ldexp.v16f16.v16i16(<16 x half> undef, <16 x i16> undef)
-; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 33 for instruction: %v17f16 = call <17 x half> @llvm.ldexp.v17f16.v17i16(<17 x half> undef, <17 x i16> undef)
+; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %f16 = call half @llvm.ldexp.f16.i16(half undef, i16 undef)
+; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %v2f16 = call <2 x half> @llvm.ldexp.v2f16.v2i16(<2 x half> undef, <2 x i16> undef)
+; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v3f16 = call <3 x half> @llvm.ldexp.v3f16.v3i16(<3 x half> undef, <3 x i16> undef)
+; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %v4f16 = call <4 x half> @llvm.ldexp.v4f16.v4i16(<4 x half> undef, <4 x i16> undef)
+; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %v5f16 = call <5 x half> @llvm.ldexp.v5f16.v5i16(<5 x half> undef, <5 x i16> undef)
+; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 23 for instruction: %v8f16 = call <8 x half> @llvm.ldexp.v8f16.v8i16(<8 x half> undef, <8 x i16> undef)
+; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 47 for instruction: %v16f16 = call <16 x half> @llvm.ldexp.v16f16.v16i16(<16 x half> undef, <16 x i16> undef)
+; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 50 for instruction: %v17f16 = call <17 x half> @llvm.ldexp.v17f16.v17i16(<17 x half> undef, <17 x i16> undef)
; GFX8PLUS-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void
;
%f16 = call half @llvm.ldexp.f16.i16(half undef, i16 undef)
@@ -125,12 +125,12 @@ define void @ldexp_bf16() {
; GFX7-LABEL: 'ldexp_bf16'
; GFX7-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %bf16 = call bfloat @llvm.ldexp.bf16.i32(bfloat undef, i32 undef)
; GFX7-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v2bf16 = call <2 x bfloat> @llvm.ldexp.v2bf16.v2i32(<2 x bfloat> undef, <2 x i32> undef)
-; GFX7-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %v3bf16 = call <3 x bfloat> @llvm.ldexp.v3bf16.v3i32(<3 x bfloat> undef, <3 x i32> undef)
+; GFX7-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v3bf16 = call <3 x bfloat> @llvm.ldexp.v3bf16.v3i32(<3 x bfloat> unde...
[truncated]
|
arsenm
approved these changes
Jul 2, 2025
| ; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %v8f16 = call <8 x half> @llvm.ldexp.v8f16.v8i32(<8 x half> undef, <8 x i32> undef) | ||
| ; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %v16f16 = call <16 x half> @llvm.ldexp.v16f16.v16i32(<16 x half> undef, <16 x i32> undef) | ||
| ; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 33 for instruction: %v17f16 = call <17 x half> @llvm.ldexp.v17f16.v17i32(<17 x half> undef, <17 x i32> undef) | ||
| ; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %f16 = call half @llvm.ldexp.f16.i32(half undef, i32 undef) |
There was a problem hiding this comment.
This should stay 1, it's directly legal
| ; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %v16f16 = call <16 x half> @llvm.ldexp.v16f16.v16i32(<16 x half> undef, <16 x i32> undef) | ||
| ; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 33 for instruction: %v17f16 = call <17 x half> @llvm.ldexp.v17f16.v17i32(<17 x half> undef, <17 x i32> undef) | ||
| ; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %f16 = call half @llvm.ldexp.f16.i32(half undef, i32 undef) | ||
| ; GFX8PLUS-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %v2f16 = call <2 x half> @llvm.ldexp.v2f16.v2i32(<2 x half> undef, <2 x i32> undef) |
There was a problem hiding this comment.
This continues to unnecessarily scale up
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
The ldexp intrinsic is incorrectly costed as 1, due to a missing entry in BasicTTIImpl::getTypeBasedIntrinsicCost: fix this. While at it, fix missing entries for [l]lround, which is costed as 1 anyway, making the change non-functional.