search for: inc15

Displaying 4 results from an estimated 4 matches for "inc15".

Did you mean: inc16
2012 Jul 30
0
[LLVMdev] IR optimization pass ideas for backend porting before ISel
...s in the "preheader" block(% for.body4.lr.ph) and a step-by-step instruction in the next block (%for.inc). 3) create a PHI instruction and a int-to-pointer instruction in the current block. 4) replace the old GEP with the new GEP. Next, I may reduce IR variables like %i.020, %sub2, %inc15 in bubbleSort-cast-2.ll and obtain better codes shown in bubbleSort-mcore-cast-2.s. The yellow codes are the original IR, and the red are modified ones: for.cond1.preheader: ; preds = %entry, %for.inc14 %indvars.iv = phi i32 [ %indvars.iv.next, %for.inc14 ], [ %sub,...
2011 Dec 14
0
[LLVMdev] Help with hazards
...ayidx = getelementptr inbounds [16000 x double]* @Y, i32 0, i32 %i.013 %0 = load double* %arrayidx, align 16, !tbaa !0 %add = fadd double %0, 1.000000e+00 %arrayidx5 = getelementptr inbounds [16000 x double]* @X, i32 0, i32 %i.013 store double %add, double* %arrayidx5, align 16, !tbaa !0 %inc15 = or i32 %i.013, 1 %arrayidx.1 = getelementptr inbounds [16000 x double]* @Y, i32 0, i32 %inc15 %1 = load double* %arrayidx.1, align 8, !tbaa !0 %add.1 = fadd double %1, 1.000000e+00 %arrayidx5.1 = getelementptr inbounds [16000 x double]* @X, i32 0, i32 %inc15 store double %add.1, double*...
2018 Jul 06
2
Verify that we only get loop metadata on latches
...ed_addr #0 { entry: br label %do.body, !llvm.loop !2 do.body: ; preds = %do.body, %entry ... br i1 %cmp, label %do.body, label %do.body6, !llvm.loop !8 do.body6: ; preds = %do.body, %do.body6 %i.1 = phi i32 [ %inc15, %do.body6 ], [ 0, %do.body ] ... br i1 %cmp17, label %do.body6, label %do.end18, !llvm.loop !8 do.end18: ; preds = %do.body6 ret i32 %add14 } ``` As seen above, when do.end is eliminated the metadata on the backward branch in do.body is changed from...
2017 Jan 24
3
[InstCombine] rL292492 affected LoopVectorizer and caused 17.30%/11.37% perf regressions on Cortex-A53/Cortex-A15 LNT machines
On Tue, Jan 24, 2017 at 1:20 PM, Sanjay Patel via llvm-dev < llvm-dev at lists.llvm.org> wrote: > > I started looking at the log files that you attached, and I'm confused. > The code that is supposedly causing the perf regression is created by the > loop vectorizer, right? Except the bad code is not in the "vector.body", so > is there something peculiar about