[LLVMdev] Loop-Unroll optimization

[Arushi Aggarwal]

1. You should run the passes in the same opt command, for passes like loops
which is an analysis pass provides results to the following passes.
2. You can pass a -debug flag to opt to see the some debugging info.

3. I tried this
 opt -mem2reg -loops  -loopsimplify -loop-unroll -unroll-count=3 -debug
loop.o -o tt.bc

and got this message.
 Loop Size = 14
  Can't unroll; loop not terminated by a conditional branch.

4. Looking at LoopUnroll.cpp showed a comment that said -loop-rotate helps,
so I used the following

 opt -mem2reg -loops  -loopsimplify -loop-rotate -lcssa -loop-unroll
-unroll-count=3 -debug loop.o -o tt.bc

Getting an unrolled version of the loop.

Arushi



On Tue, May 3, 2011 at 6:50 PM, Manish Gupta <mgupta.iitr at gmail.com> wrote:

> Even after all the sequence of commands below bit-code is not showing any
> effect of loop-unrolling
>
> *manish at manish:~/Test2$ llvm-gcc-4.2 -O2 -emit-llvm Hello.c -c -o Hello.bc
> *
> *manish at manish:~/Test2$ opt-2.8 -loops Hello.bc -o Hello1.bc*
> *manish at manish:~/Test2$ opt-2.8 -loopsimplify Hello1.bc -o Hello2.bc*
> *manish at manish:~/Test2$ opt-2.8 -indvars Hello2.bc -o Hello3.bc*
> *manish at manish:~/Test2$ opt-2.8 -loop-unroll Hello3.bc -o Hello4.bc*
> *manish at manish:~/Test2$ llvm-dis-2.8 Hello4.bc*
>
> My *Hello.c* looks like:
>   for(i=0; i< 1000; i++)
>     {
>       c[i] = a[i] + b[i];
>      }
>
>   printf("%d\n", c[999]);
> -------------------------------------------------
>
> and bit-code in *Hello4.bc*
> bb3:                                              ; preds = %bb3,
> %bb3.preheader
>   %i.17 = phi i32 [ %5, %bb3 ], [ 0, %bb3.preheader ]
>   %scevgep11 = getelementptr [1000 x i32]* %b, i32 0, i32 %i.17
>   %scevgep10 = getelementptr [1000 x i32]* %a, i32 0, i32 %i.17
>   %scevgep = getelementptr [1000 x i32]* %c, i32 0, i32 %i.17
>   %2 = load i32* %scevgep10, align 4
>   %3 = load i32* %scevgep11, align 4
>   %4 = add nsw i32 %3, %2
>   store i32 %4, i32* %scevgep, align 4
>   %5 = add nsw i32 %i.17, 1
>   %exitcond1 = icmp eq i32 %5, 1000
>   br i1 %exitcond1, label %bb5, label %bb3
>
> -----------------------------------------
>
> On Tue, May 3, 2011 at 12:32 PM, Arushi Aggarwal <arushi987 at gmail.com>wrote:
>
>> Hi,
>>
>> You might want to try running -loops -loop-simplify before loop unroll.
>>
>> From loop simplify.cpp
>>
>> This pass performs several transformations to transform natural loops into a00011 // simpler form, which makes subsequent analyses and transformations simpler and00012 // more effective.
>>
>>
>> Arushi
>>
>> On Tue, May 3, 2011 at 2:17 PM, Manish Gupta <mgupta.iitr at gmail.com>wrote:
>>
>>> You mean like
>>>
>>> *llvm-gcc-4.2 -O2 -emit-llvm Hello.c -c -o Hello.bc*
>>>
>>> But still i am not able to observe any effect on bit code by running
>>>
>>> *opt-2.8 -loop-unroll Hello.bc -o Hello_unroll.bc*
>>>
>>> On Tue, May 3, 2011 at 3:58 AM, Zakk <zakk0610 at gmail.com> wrote:
>>>
>>>> Hi, you need to run some optimization passes first. (like -O2)
>>>>
>>>> 2011/5/3 Manish Gupta <mgupta.iitr at gmail.com>
>>>>
>>>>> I just want to try loop-unroll and see corresponding changes in the
>>>>> bitcode file. For that any loop will do. Have you been able to test llvm
>>>>> loop-unroll successfully?
>>>>>
>>>>>
>>>>> On Mon, May 2, 2011 at 10:04 PM, Yuan Pengfei <coolypf at qq.com> wrote:
>>>>>
>>>>>> Hi,
>>>>>>
>>>>>> > The loop that I am trying it on is:
>>>>>> > for(i=0; i< 1000; i++)
>>>>>> > {
>>>>>> > c[i] = a[i] + b[i];
>>>>>> > }
>>>>>>
>>>>>> I can't find any benefit unrolling this loop.
>>>>>>
>>>>>> ------------------
>>>>>> Yuan Pengfei
>>>>>> Peking Unversity, China
>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> LLVM Developers mailing list
>>>>> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
>>>>> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>>>>>
>>>>>
>>>>
>>>>
>>>> --
>>>> Best regards,
>>>> Kuan-Hsu
>>>>
>>>>
>>>>
>>>
>>> _______________________________________________
>>> LLVM Developers mailing list
>>> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
>>> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>>>
>>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20110503/7e0b4a9b/attachment.html>