Skip to content

[AArch64][CostModel] Add SVE bfloat arithmetic tests (NFC) #166951

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
merged 3 commits into from
Nov 11, 2025
Merged

[AArch64][CostModel] Add SVE bfloat arithmetic tests (NFC) #166951

merged 3 commits into from
Nov 11, 2025

Conversation

@MacDue
Copy link
Member

@MacDue MacDue commented Nov 7, 2025

This patch adds cost model tests for bfloat operations with +sve-b16b16. Currently, some of these costs are higher than they should be as the cost model is assuming bfloats need promotion, but some of these operations are natively supported with +sve-b16b16.

@MacDue MacDue requested review from c-rhodes and davemgreen November 7, 2025 15:10
@llvmbot llvmbot added the llvm:analysis Includes value tracking, cost tables and constant folding label Nov 7, 2025
@llvmbot
Copy link
Member

llvmbot commented Nov 7, 2025

@llvm/pr-subscribers-llvm-analysis

Author: Benjamin Maxwell (MacDue)

Changes

This patch adds cost model tests for bfloat operations with +sve-b16b16. Currently, some of these costs are higher than they should be as the cost model is assuming bfloats need promotion, but some of these operations are natively supported with +sve-b16b16.

Full diff: https://github.com/llvm/llvm-project/pull/166951.diff

1 Files Affected:

  • (modified) llvm/test/Analysis/CostModel/AArch64/sve-arith-fp.ll (+59-1)
diff --git a/llvm/test/Analysis/CostModel/AArch64/sve-arith-fp.ll b/llvm/test/Analysis/CostModel/AArch64/sve-arith-fp.ll
index ec848c2c08305..41e08fa903604 100644
--- a/llvm/test/Analysis/CostModel/AArch64/sve-arith-fp.ll
+++ b/llvm/test/Analysis/CostModel/AArch64/sve-arith-fp.ll
@@ -1,5 +1,5 @@
 ; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
-; RUN: opt < %s -enable-no-nans-fp-math -passes="print<cost-model>" -cost-kind=all 2>&1 -disable-output -mtriple=aarch64 -mattr=+fullfp16 -mattr=+sve | FileCheck %s
+; RUN: opt < %s -enable-no-nans-fp-math -passes="print<cost-model>" -cost-kind=all 2>&1 -disable-output -mtriple=aarch64 -mattr=+fullfp16 -mattr=+sve-b16b16 -mattr=+sve | FileCheck %s
 
 target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
 
@@ -8,6 +8,9 @@ define void @fadd() {
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:3 SizeLat:1 for: %V4F16 = fadd <vscale x 4 x half> poison, poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:3 SizeLat:1 for: %V8F16 = fadd <vscale x 8 x half> poison, poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:1 Lat:3 SizeLat:1 for: %V16F16 = fadd <vscale x 16 x half> poison, poison
+; CHECK-NEXT:  Cost Model: Found costs of RThru:11 CodeSize:1 Lat:3 SizeLat:1 for: %V4BF16 = fadd <vscale x 4 x bfloat> poison, poison
+; CHECK-NEXT:  Cost Model: Found costs of RThru:27 CodeSize:1 Lat:3 SizeLat:1 for: %V8BF16 = fadd <vscale x 8 x bfloat> poison, poison
+; CHECK-NEXT:  Cost Model: Found costs of RThru:54 CodeSize:1 Lat:3 SizeLat:1 for: %V16BF16 = fadd <vscale x 16 x bfloat> poison, poison
 ; CHECK-NEXT:  Cost Model: Found costs of Invalid for: %V1F32 = fadd <vscale x 1 x float> poison, poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:3 SizeLat:1 for: %V2F32 = fadd <vscale x 2 x float> poison, poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:3 SizeLat:1 for: %V4F32 = fadd <vscale x 4 x float> poison, poison
@@ -20,6 +23,10 @@ define void @fadd() {
   %V8F16 = fadd <vscale x 8 x half> poison, poison
   %V16F16 = fadd <vscale x 16 x half> poison, poison
 
+  %V4BF16 = fadd <vscale x 4 x bfloat> poison, poison
+  %V8BF16 = fadd <vscale x 8 x bfloat> poison, poison
+  %V16BF16 = fadd <vscale x 16 x bfloat> poison, poison
+
   %V1F32 = fadd <vscale x 1 x float> poison, poison
   %V2F32 = fadd <vscale x 2 x float> poison, poison
   %V4F32 = fadd <vscale x 4 x float> poison, poison
@@ -36,6 +43,9 @@ define void @fsub() {
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:3 SizeLat:1 for: %V4F16 = fsub <vscale x 4 x half> poison, poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:3 SizeLat:1 for: %V8F16 = fsub <vscale x 8 x half> poison, poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:1 Lat:3 SizeLat:1 for: %V16F16 = fsub <vscale x 16 x half> poison, poison
+; CHECK-NEXT:  Cost Model: Found costs of RThru:11 CodeSize:1 Lat:3 SizeLat:1 for: %V4BF16 = fsub <vscale x 4 x bfloat> poison, poison
+; CHECK-NEXT:  Cost Model: Found costs of RThru:27 CodeSize:1 Lat:3 SizeLat:1 for: %V8BF16 = fsub <vscale x 8 x bfloat> poison, poison
+; CHECK-NEXT:  Cost Model: Found costs of RThru:54 CodeSize:1 Lat:3 SizeLat:1 for: %V16BF16 = fsub <vscale x 16 x bfloat> poison, poison
 ; CHECK-NEXT:  Cost Model: Found costs of Invalid for: %V1F32 = fsub <vscale x 1 x float> poison, poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:3 SizeLat:1 for: %V2F32 = fsub <vscale x 2 x float> poison, poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:3 SizeLat:1 for: %V4F32 = fsub <vscale x 4 x float> poison, poison
@@ -48,6 +58,10 @@ define void @fsub() {
   %V8F16 = fsub <vscale x 8 x half> poison, poison
   %V16F16 = fsub <vscale x 16 x half> poison, poison
 
+  %V4BF16 = fsub <vscale x 4 x bfloat> poison, poison
+  %V8BF16 = fsub <vscale x 8 x bfloat> poison, poison
+  %V16BF16 = fsub <vscale x 16 x bfloat> poison, poison
+
   %V1F32 = fsub <vscale x 1 x float> poison, poison
   %V2F32 = fsub <vscale x 2 x float> poison, poison
   %V4F32 = fsub <vscale x 4 x float> poison, poison
@@ -65,6 +79,10 @@ define void @fneg() {
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:3 SizeLat:1 for: %V4F16 = fneg <vscale x 4 x half> poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:3 SizeLat:1 for: %V8F16 = fneg <vscale x 8 x half> poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:1 Lat:3 SizeLat:1 for: %V16F16 = fneg <vscale x 16 x half> poison
+; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:3 SizeLat:1 for: %V2BF16 = fneg <vscale x 2 x bfloat> poison
+; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:3 SizeLat:1 for: %V4BF16 = fneg <vscale x 4 x bfloat> poison
+; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:3 SizeLat:1 for: %V8BF16 = fneg <vscale x 8 x bfloat> poison
+; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:1 Lat:3 SizeLat:1 for: %V16BF16 = fneg <vscale x 16 x bfloat> poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:3 SizeLat:1 for: %V2F32 = fneg <vscale x 2 x float> poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:3 SizeLat:1 for: %V4F32 = fneg <vscale x 4 x float> poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:1 Lat:3 SizeLat:1 for: %V8F32 = fneg <vscale x 8 x float> poison
@@ -77,6 +95,11 @@ define void @fneg() {
   %V8F16 = fneg <vscale x 8 x half> poison
   %V16F16 = fneg <vscale x 16 x half> poison
 
+  %V2BF16 = fneg <vscale x 2 x bfloat> poison
+  %V4BF16 = fneg <vscale x 4 x bfloat> poison
+  %V8BF16 = fneg <vscale x 8 x bfloat> poison
+  %V16BF16 = fneg <vscale x 16 x bfloat> poison
+
   %V2F32 = fneg <vscale x 2 x float> poison
   %V4F32 = fneg <vscale x 4 x float> poison
   %V8F32 = fneg <vscale x 8 x float> poison
@@ -92,6 +115,9 @@ define void @fmul() {
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:1 Lat:3 SizeLat:1 for: %V4F16 = fmul <vscale x 4 x half> poison, poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:1 Lat:3 SizeLat:1 for: %V8F16 = fmul <vscale x 8 x half> poison, poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:4 CodeSize:1 Lat:3 SizeLat:1 for: %V16F16 = fmul <vscale x 16 x half> poison, poison
+; CHECK-NEXT:  Cost Model: Found costs of RThru:12 CodeSize:1 Lat:3 SizeLat:1 for: %V4BF16 = fmul <vscale x 4 x bfloat> poison, poison
+; CHECK-NEXT:  Cost Model: Found costs of RThru:29 CodeSize:1 Lat:3 SizeLat:1 for: %V8BF16 = fmul <vscale x 8 x bfloat> poison, poison
+; CHECK-NEXT:  Cost Model: Found costs of RThru:58 CodeSize:1 Lat:3 SizeLat:1 for: %V16BF16 = fmul <vscale x 16 x bfloat> poison, poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:1 Lat:3 SizeLat:1 for: %V2F32 = fmul <vscale x 2 x float> poison, poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:1 Lat:3 SizeLat:1 for: %V4F32 = fmul <vscale x 4 x float> poison, poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:4 CodeSize:1 Lat:3 SizeLat:1 for: %V8F32 = fmul <vscale x 8 x float> poison, poison
@@ -103,6 +129,10 @@ define void @fmul() {
   %V8F16 = fmul <vscale x 8 x half> poison, poison
   %V16F16 = fmul <vscale x 16 x half> poison, poison
 
+  %V4BF16 = fmul <vscale x 4 x bfloat> poison, poison
+  %V8BF16 = fmul <vscale x 8 x bfloat> poison, poison
+  %V16BF16 = fmul <vscale x 16 x bfloat> poison, poison
+
   %V2F32 = fmul <vscale x 2 x float> poison, poison
   %V4F32 = fmul <vscale x 4 x float> poison, poison
   %V8F32 = fmul <vscale x 8 x float> poison, poison
@@ -118,6 +148,9 @@ define void @fdiv() {
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:4 Lat:4 SizeLat:4 for: %V4F16 = fdiv <vscale x 4 x half> poison, poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:4 Lat:4 SizeLat:4 for: %V8F16 = fdiv <vscale x 8 x half> poison, poison
 ; CHECK-NEXT:  Cost Model: Found costs of 4 for: %V16F16 = fdiv <vscale x 16 x half> poison, poison
+; CHECK-NEXT:  Cost Model: Found costs of RThru:12 CodeSize:4 Lat:4 SizeLat:4 for: %V4BF16 = fdiv <vscale x 4 x bfloat> poison, poison
+; CHECK-NEXT:  Cost Model: Found costs of RThru:29 CodeSize:4 Lat:4 SizeLat:4 for: %V8BF16 = fdiv <vscale x 8 x bfloat> poison, poison
+; CHECK-NEXT:  Cost Model: Found costs of RThru:58 CodeSize:4 Lat:4 SizeLat:4 for: %V16BF16 = fdiv <vscale x 16 x bfloat> poison, poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:4 Lat:4 SizeLat:4 for: %V2F32 = fdiv <vscale x 2 x float> poison, poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:4 Lat:4 SizeLat:4 for: %V4F32 = fdiv <vscale x 4 x float> poison, poison
 ; CHECK-NEXT:  Cost Model: Found costs of 4 for: %V8F32 = fdiv <vscale x 8 x float> poison, poison
@@ -129,6 +162,10 @@ define void @fdiv() {
   %V8F16 = fdiv <vscale x 8 x half> poison, poison
   %V16F16 = fdiv <vscale x 16 x half> poison, poison
 
+  %V4BF16 = fdiv <vscale x 4 x bfloat> poison, poison
+  %V8BF16 = fdiv <vscale x 8 x bfloat> poison, poison
+  %V16BF16 = fdiv <vscale x 16 x bfloat> poison, poison
+
   %V2F32 = fdiv <vscale x 2 x float> poison, poison
   %V4F32 = fdiv <vscale x 4 x float> poison, poison
   %V8F32 = fdiv <vscale x 8 x float> poison, poison
@@ -144,6 +181,9 @@ define void @frem() {
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:Invalid CodeSize:4 Lat:4 SizeLat:4 for: %V4F16 = frem <vscale x 4 x half> poison, poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:Invalid CodeSize:4 Lat:4 SizeLat:4 for: %V8F16 = frem <vscale x 8 x half> poison, poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:Invalid CodeSize:4 Lat:4 SizeLat:4 for: %V16F16 = frem <vscale x 16 x half> poison, poison
+; CHECK-NEXT:  Cost Model: Found costs of RThru:Invalid CodeSize:4 Lat:4 SizeLat:4 for: %V4BF16 = frem <vscale x 4 x bfloat> poison, poison
+; CHECK-NEXT:  Cost Model: Found costs of RThru:Invalid CodeSize:4 Lat:4 SizeLat:4 for: %V8BF16 = frem <vscale x 8 x bfloat> poison, poison
+; CHECK-NEXT:  Cost Model: Found costs of RThru:Invalid CodeSize:4 Lat:4 SizeLat:4 for: %V16BF16 = frem <vscale x 16 x bfloat> poison, poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:Invalid CodeSize:4 Lat:4 SizeLat:4 for: %V2F32 = frem <vscale x 2 x float> poison, poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:Invalid CodeSize:4 Lat:4 SizeLat:4 for: %V4F32 = frem <vscale x 4 x float> poison, poison
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:Invalid CodeSize:4 Lat:4 SizeLat:4 for: %V8F32 = frem <vscale x 8 x float> poison, poison
@@ -155,6 +195,10 @@ define void @frem() {
   %V8F16 = frem <vscale x 8 x half> poison, poison
   %V16F16 = frem <vscale x 16 x half> poison, poison
 
+  %V4BF16 = frem <vscale x 4 x bfloat> poison, poison
+  %V8BF16 = frem <vscale x 8 x bfloat> poison, poison
+  %V16BF16 = frem <vscale x 16 x bfloat> poison, poison
+
   %V2F32 = frem <vscale x 2 x float> poison, poison
   %V4F32 = frem <vscale x 4 x float> poison, poison
   %V8F32 = frem <vscale x 8 x float> poison, poison
@@ -170,6 +214,9 @@ define void @fma() {
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:1 Lat:3 SizeLat:1 for: %V4F16 = call <vscale x 4 x half> @llvm.fma.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x half> poison, <vscale x 4 x half> poison)
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:1 Lat:3 SizeLat:1 for: %V8F16 = call <vscale x 8 x half> @llvm.fma.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x half> poison, <vscale x 8 x half> poison)
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:4 CodeSize:1 Lat:3 SizeLat:1 for: %V16F16 = call <vscale x 16 x half> @llvm.fma.nxv16f16(<vscale x 16 x half> poison, <vscale x 16 x half> poison, <vscale x 16 x half> poison)
+; CHECK-NEXT:  Cost Model: Found costs of 2 for: %V4BF16 = call <vscale x 4 x bfloat> @llvm.fma.nxv4bf16(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found costs of 2 for: %V8BF16 = call <vscale x 8 x bfloat> @llvm.fma.nxv8bf16(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found costs of 4 for: %V16BF16 = call <vscale x 16 x bfloat> @llvm.fma.nxv16bf16(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> poison)
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:1 Lat:3 SizeLat:1 for: %V2F32 = call <vscale x 2 x float> @llvm.fma.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x float> poison, <vscale x 2 x float> poison)
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:1 Lat:3 SizeLat:1 for: %V4F32 = call <vscale x 4 x float> @llvm.fma.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x float> poison, <vscale x 4 x float> poison)
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:4 CodeSize:1 Lat:3 SizeLat:1 for: %V8F32 = call <vscale x 8 x float> @llvm.fma.nxv8f32(<vscale x 8 x float> poison, <vscale x 8 x float> poison, <vscale x 8 x float> poison)
@@ -181,6 +228,10 @@ define void @fma() {
   %V8F16 = call <vscale x 8 x half> @llvm.fma.v8f16(<vscale x 8 x half> poison, <vscale x 8 x half> poison, <vscale x 8 x half> poison)
   %V16F16 = call <vscale x 16 x half> @llvm.fma.v16f16(<vscale x 16 x half> poison, <vscale x 16 x half> poison, <vscale x 16 x half> poison)
 
+  %V4BF16 = call <vscale x 4 x bfloat> @llvm.fma.v4BF16(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> poison)
+  %V8BF16 = call <vscale x 8 x bfloat> @llvm.fma.v8BF16(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> poison)
+  %V16BF16 = call <vscale x 16 x bfloat> @llvm.fma.v16BF16(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> poison)
+
   %V2F32 = call <vscale x 2 x float> @llvm.fma.v2f32(<vscale x 2 x float> poison, <vscale x 2 x float> poison, <vscale x 2 x float> poison)
   %V4F32 = call <vscale x 4 x float> @llvm.fma.v4f32(<vscale x 4 x float> poison, <vscale x 4 x float> poison, <vscale x 4 x float> poison)
   %V8F32 = call <vscale x 8 x float> @llvm.fma.v8f32(<vscale x 8 x float> poison, <vscale x 8 x float> poison, <vscale x 8 x float> poison)
@@ -196,6 +247,9 @@ define void @fmuladd() {
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:1 Lat:3 SizeLat:1 for: %V4F16 = call <vscale x 4 x half> @llvm.fmuladd.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x half> poison, <vscale x 4 x half> poison)
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:1 Lat:3 SizeLat:1 for: %V8F16 = call <vscale x 8 x half> @llvm.fmuladd.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x half> poison, <vscale x 8 x half> poison)
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:4 CodeSize:1 Lat:3 SizeLat:1 for: %V16F16 = call <vscale x 16 x half> @llvm.fmuladd.nxv16f16(<vscale x 16 x half> poison, <vscale x 16 x half> poison, <vscale x 16 x half> poison)
+; CHECK-NEXT:  Cost Model: Found costs of 2 for: %V4BF16 = call <vscale x 4 x bfloat> @llvm.fmuladd.nxv4bf16(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found costs of 2 for: %V8BF16 = call <vscale x 8 x bfloat> @llvm.fmuladd.nxv8bf16(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found costs of 4 for: %V16BF16 = call <vscale x 16 x bfloat> @llvm.fmuladd.nxv16bf16(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> poison)
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:1 Lat:3 SizeLat:1 for: %V2F32 = call <vscale x 2 x float> @llvm.fmuladd.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x float> poison, <vscale x 2 x float> poison)
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:1 Lat:3 SizeLat:1 for: %V4F32 = call <vscale x 4 x float> @llvm.fmuladd.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x float> poison, <vscale x 4 x float> poison)
 ; CHECK-NEXT:  Cost Model: Found costs of RThru:4 CodeSize:1 Lat:3 SizeLat:1 for: %V8F32 = call <vscale x 8 x float> @llvm.fmuladd.nxv8f32(<vscale x 8 x float> poison, <vscale x 8 x float> poison, <vscale x 8 x float> poison)
@@ -207,6 +261,10 @@ define void @fmuladd() {
   %V8F16 = call <vscale x 8 x half> @llvm.fmuladd.v8f16(<vscale x 8 x half> poison, <vscale x 8 x half> poison, <vscale x 8 x half> poison)
   %V16F16 = call <vscale x 16 x half> @llvm.fmuladd.v16f16(<vscale x 16 x half> poison, <vscale x 16 x half> poison, <vscale x 16 x half> poison)
 
+  %V4BF16 = call <vscale x 4 x bfloat> @llvm.fmuladd.v4BF16(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> poison)
+  %V8BF16 = call <vscale x 8 x bfloat> @llvm.fmuladd.v8BF16(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> poison)
+  %V16BF16 = call <vscale x 16 x bfloat> @llvm.fmuladd.v16BF16(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> poison)
+
   %V2F32 = call <vscale x 2 x float> @llvm.fmuladd.v2f32(<vscale x 2 x float> poison, <vscale x 2 x float> poison, <vscale x 2 x float> poison)
   %V4F32 = call <vscale x 4 x float> @llvm.fmuladd.v4f32(<vscale x 4 x float> poison, <vscale x 4 x float> poison, <vscale x 4 x float> poison)
   %V8F32 = call <vscale x 8 x float> @llvm.fmuladd.v8f32(<vscale x 8 x float> poison, <vscale x 8 x float> poison, <vscale x 8 x float> poison)

c-rhodes
c-rhodes approved these changes Nov 7, 2025
View reviewed changes
Copy link
Collaborator

@c-rhodes c-rhodes left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

couple of minor comments but otherwise LGTM cheers

davemgreen
davemgreen reviewed Nov 7, 2025
View reviewed changes
Copy link
Collaborator

@davemgreen davemgreen left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

When adding the fp16 tests for fp/neon (which is in a similar situation to bf16 for sve), we separate the fp16 tests into their own function and then add two runs lines with and without +fullfp16. Can we do the same here so we can test both costs?

@MacDue
Copy link
Member Author

MacDue commented Nov 10, 2025

When adding the fp16 tests for fp/neon (which is in a similar situation to bf16 for sve), we separate the fp16 tests into their own function and then add two runs lines with and without +fullfp16. Can we do the same here so we can test both costs?

Done 👍

davemgreen
davemgreen approved these changes Nov 10, 2025
View reviewed changes
Copy link
Collaborator

@davemgreen davemgreen left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks - this has worked quite well for FP16, hopefully it can work well here too.

LGTM

@MacDue MacDue merged commit efc0ab0 into llvm:main Nov 11, 2025
10 checks passed
@MacDue MacDue deleted the tests_base branch November 11, 2025 10:43
@llvm-ci
Copy link
Collaborator

llvm-ci commented Nov 11, 2025

LLVM Buildbot has detected a new failure on builder ppc64le-flang-rhel-clang running on ppc64le-flang-rhel-test while building llvm at step 6 "test-build-unified-tree-check-flang".

Full details are available at: https://lab.llvm.org/buildbot/#/builders/157/builds/42112

Here is the relevant piece of the build log for the reference 
Step 6 (test-build-unified-tree-check-flang) failure: 1200 seconds without output running [b'ninja', b'check-flang'], attempting to kill
...
PASS: Flang :: Transforms/DoConcurrent/use_loop_bounds_in_body.f90 (3974 of 3984)
PASS: Flang :: Transforms/external-name-interop-symref-array.fir (3975 of 3984)
PASS: Flang :: Driver/fopenmp.f90 (3976 of 3984)
PASS: Flang :: Semantics/modfile73.f90 (3977 of 3984)
PASS: Flang :: Transforms/dlti-dependency.fir (3978 of 3984)
PASS: Flang :: Transforms/constant-argument-globalisation.fir (3979 of 3984)
PASS: Flang :: Transforms/debug-dwarf-version.fir (3980 of 3984)
PASS: Flang :: Driver/omp-driver-offload.f90 (3981 of 3984)
PASS: Flang :: Driver/linker-options.f90 (3982 of 3984)
PASS: Flang :: Intrinsics/math-codegen.fir (3983 of 3984)
command timed out: 1200 seconds without output running [b'ninja', b'check-flang'], attempting to kill
process killed by signal 9
program finished with exit code -1
elapsedTime=2335.107099

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@c-rhodes c-rhodes c-rhodes approved these changes

@davemgreen davemgreen davemgreen approved these changes

Assignees

No one assigned

Labels

llvm:analysis Includes value tracking, cost tables and constant folding

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

5 participants

@MacDue @llvmbot @llvm-ci @c-rhodes @davemgreen