6502/llvm-6502
1
0
Fork 0
mirror of https://github.com/c64scene-ar/llvm-6502.git synced 2024-07-05 13:29:22 +00:00
Code Issues Projects Releases Wiki Activity

R600/SI: Fix bug in TTI loop unrolling preferences

Browse Source 
We should be setting UnrollingPreferences::MaxCount to MAX_UINT instead
of UnrollingPreferences::Count.

Count is a 'forced unrolling factor', while MaxCount sets an upper
limit to the unrolling factor.

Setting Count to MAX_UINT was causing the loop in the testcase to be
unrolled 15 times, when it only had a maximum of 4 iterations.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@228303 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Tom Stellard 2015-02-05 15:32:18 +00:00
parent 041211cd79
commit c7198528eb
2 changed files with 59 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
lib/Target/R600/AMDGPUTargetTransformInfo.cpp
View File

@ -30,7 +30,7 @@ using namespace llvm;
void AMDGPUTTIImpl::getUnrollingPreferences(Loop *L,
TTI::UnrollingPreferences &UP) {
UP.Threshold = 300; // Twice the default.
UP.Count = UINT_MAX;
UP.MaxCount = UINT_MAX;
UP.Partial = true;
// TODO: Do we want runtime unrolling?

58
test/CodeGen/R600/tti-unroll-prefs.ll Normal file
View File

@ -0,0 +1,58 @@
; RUN: opt -loop-unroll -S -mtriple=amdgcn-- -mcpu=SI %s | FileCheck %s
; This IR comes from this OpenCL C code:
;
; if (b + 4 > a) {
; for (int i = 0; i < 4; i++, b++) {
; if (b + 1 <= a)
; *(dst + c + b) = 0;
; else
; break;
; }
; }
;
; This test is meant to check that this loop isn't unrolled into more than
; four iterations. The loop unrolling preferences we currently use cause this
; loop to not be unrolled at all, but that may change in the future.
; CHECK-LABEL: @test
; CHECK: store i8 0, i8 addrspace(1)*
; CHECK-NOT: store i8 0, i8 addrspace(1)*
; CHECK: ret void
define void @test(i8 addrspace(1)* nocapture %dst, i32 %a, i32 %b, i32 %c) {
entry:
%add = add nsw i32 %b, 4
%cmp = icmp sgt i32 %add, %a
br i1 %cmp, label %for.cond.preheader, label %if.end7
for.cond.preheader: ; preds = %entry
%cmp313 = icmp slt i32 %b, %a
br i1 %cmp313, label %if.then4.lr.ph, label %if.end7.loopexit
if.then4.lr.ph: ; preds = %for.cond.preheader
%0 = sext i32 %c to i64
br label %if.then4
if.then4: ; preds = %if.then4.lr.ph, %if.then4
%i.015 = phi i32 [ 0, %if.then4.lr.ph ], [ %inc, %if.then4 ]
%b.addr.014 = phi i32 [ %b, %if.then4.lr.ph ], [ %add2, %if.then4 ]
%add2 = add nsw i32 %b.addr.014, 1
%1 = sext i32 %b.addr.014 to i64
%add.ptr.sum = add nsw i64 %1, %0
%add.ptr5 = getelementptr inbounds i8 addrspace(1)* %dst, i64 %add.ptr.sum
store i8 0, i8 addrspace(1)* %add.ptr5, align 1
%inc = add nsw i32 %i.015, 1
%cmp1 = icmp slt i32 %inc, 4
%cmp3 = icmp slt i32 %add2, %a
%or.cond = and i1 %cmp3, %cmp1
br i1 %or.cond, label %if.then4, label %for.cond.if.end7.loopexit_crit_edge
for.cond.if.end7.loopexit_crit_edge: ; preds = %if.then4
br label %if.end7.loopexit
if.end7.loopexit: ; preds = %for.cond.if.end7.loopexit_crit_edge, %for.cond.preheader
br label %if.end7
if.end7: ; preds = %if.end7.loopexit, %entry
ret void
}