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.<div>2. You can pass a -debug flag to opt to see the some debugging info.</div>
<div><br></div><div>3. I tried this </div><div> opt -mem2reg -loops -loopsimplify -loop-unroll -unroll-count=3 -debug loop.o -o tt.bc</div><div><br></div><div>and got this message.</div><div><div> Loop Size = 14</div><div>
Can't unroll; loop not terminated by a conditional branch.</div></div><div><br></div><div>4. Looking at LoopUnroll.cpp showed a comment that said -loop-rotate helps, so I used the following</div><div><br></div><div>
opt -mem2reg -loops -loopsimplify -loop-rotate -lcssa -loop-unroll -unroll-count=3 -debug loop.o -o tt.bc</div><div><br></div><div>Getting an unrolled version of the loop.</div><div><br></div><div>Arushi</div><div><br>
</div>
<div><br><br><div class="gmail_quote">On Tue, May 3, 2011 at 6:50 PM, Manish Gupta <span dir="ltr"><<a href="mailto:mgupta.iitr@gmail.com">mgupta.iitr@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
Even after all the sequence of commands below bit-code is not showing any effect of loop-unrolling<div><br></div><div><b>manish@manish:~/Test2$ llvm-gcc-4.2 -O2 -emit-llvm Hello.c -c -o Hello.bc</b></div><div><div><b>manish@manish:~/Test2$ opt-2.8 -loops Hello.bc -o Hello1.bc</b></div>
<div><b>manish@manish:~/Test2$ opt-2.8 -loopsimplify Hello1.bc -o Hello2.bc</b></div><div><b>manish@manish:~/Test2$ opt-2.8 -indvars Hello2.bc -o Hello3.bc</b></div><div><b>manish@manish:~/Test2$ opt-2.8 -loop-unroll Hello3.bc -o Hello4.bc</b></div>
<div><b>manish@manish:~/Test2$ llvm-dis-2.8 Hello4.bc</b></div></div><div><br></div><div>My <b>Hello.c</b> looks like: </div><div><div> for(i=0; i< 1000; i++)</div><div> {</div><div> c[i] = a[i] + b[i];</div>
<div>
}</div><div><br></div><div> printf("%d\n", c[999]);</div></div><div>-------------------------------------------------</div><div><br></div><div>and bit-code in <b>Hello4.bc</b></div><div><div>bb3: ; preds = %bb3, %bb3.preheader</div>
<div> %i.17 = phi i32 [ %5, %bb3 ], [ 0, %bb3.preheader ]</div><div> %scevgep11 = getelementptr [1000 x i32]* %b, i32 0, i32 %i.17</div><div> %scevgep10 = getelementptr [1000 x i32]* %a, i32 0, i32 %i.17</div><div> %scevgep = getelementptr [1000 x i32]* %c, i32 0, i32 %i.17</div>
<div> %2 = load i32* %scevgep10, align 4</div><div> %3 = load i32* %scevgep11, align 4</div><div> %4 = add nsw i32 %3, %2</div><div> store i32 %4, i32* %scevgep, align 4</div><div> %5 = add nsw i32 %i.17, 1</div><div>
%exitcond1 = icmp eq i32 %5, 1000</div><div> br i1 %exitcond1, label %bb5, label %bb3</div></div><div><br></div><div>-----------------------------------------</div><div><br><div class="gmail_quote">On Tue, May 3, 2011 at 12:32 PM, Arushi Aggarwal <span dir="ltr"><<a href="mailto:arushi987@gmail.com" target="_blank">arushi987@gmail.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Hi,<div><br></div><div>You might want to try running -loops -loop-simplify before loop unroll.</div><div><br></div><div>
>From loop simplify.cpp</div><div><span style="font-family:Fixed, monospace;font-size:14px"><pre style="font-family:Fixed, monospace;font-size:13px;border-top-width:1px;border-right-width:1px;border-bottom-width:1px;border-left-width:1px;border-top-style:solid;border-right-style:solid;border-bottom-style:solid;border-left-style:solid;border-top-color:rgb(204, 204, 204);border-right-color:rgb(204, 204, 204);border-bottom-color:rgb(204, 204, 204);border-left-color:rgb(204, 204, 204);background-color:rgb(245, 245, 245);margin-top:4px;margin-bottom:4px;margin-left:2px;margin-right:8px;padding-left:6px;padding-right:6px;padding-top:4px;padding-bottom:4px">
<span style="color:rgb(128, 0, 0)">This pass performs several transformations to transform natural loops into a</span>
<a name="12fb845dd834bc25_12fb7598dc278aa2_l00011" style="color:rgb(37, 46, 120)"></a>00011 <span style="color:rgb(128, 0, 0)">// simpler form, which makes subsequent analyses and transformations simpler and</span>
<a name="12fb845dd834bc25_12fb7598dc278aa2_l00012" style="color:rgb(37, 46, 120)"></a>00012 <span style="color:rgb(128, 0, 0)">// more effective.</span></pre></span><div><br></div><font color="#888888">Arushi</font></div>
<div><div></div>
<div><div><br><div class="gmail_quote">On Tue, May 3, 2011 at 2:17 PM, Manish Gupta <span dir="ltr"><<a href="mailto:mgupta.iitr@gmail.com" target="_blank">mgupta.iitr@gmail.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">You mean like <div><br></div><div><b>llvm-gcc-4.2 -O2 -emit-llvm Hello.c -c -o Hello.bc</b></div><div><br></div><div>
But still i am not able to observe any effect on bit code by running </div>
<div><br></div><div><b>opt-2.8 -loop-unroll Hello.bc -o Hello_unroll.bc</b></div><div><div></div><div>
<div><br><div class="gmail_quote">On Tue, May 3, 2011 at 3:58 AM, Zakk <span dir="ltr"><<a href="mailto:zakk0610@gmail.com" target="_blank">zakk0610@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Hi, you need to run some optimization passes first. (like -O2)<br><br><div class="gmail_quote"><div><div></div><div>2011/5/3 Manish Gupta <span dir="ltr"><<a href="mailto:mgupta.iitr@gmail.com" target="_blank">mgupta.iitr@gmail.com</a>></span><br>
</div></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div><div></div><div>
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? <div><div></div><div><br><br><div class="gmail_quote">
On Mon, May 2, 2011 at 10:04 PM, Yuan Pengfei <span dir="ltr"><<a href="mailto:coolypf@qq.com" target="_blank">coolypf@qq.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Hi,<br>
<div><br>
> The loop that I am trying it on is:<br>
> for(i=0; i< 1000; i++)<br>
> {<br>
> c[i] = a[i] + b[i];<br>
> }<br>
<br>
</div>I can't find any benefit unrolling this loop.<br>
<br>
------------------<br>
<font color="#888888">Yuan Pengfei<br>
Peking Unversity, China</font></blockquote></div><br>
</div></div><br></div></div>_______________________________________________<br>
LLVM Developers mailing list<br>
<a href="mailto:LLVMdev@cs.uiuc.edu" target="_blank">LLVMdev@cs.uiuc.edu</a> <a href="http://llvm.cs.uiuc.edu" target="_blank">http://llvm.cs.uiuc.edu</a><br>
<a href="http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev" target="_blank">http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev</a><br>
<br></blockquote></div><font color="#888888"><br><br clear="all"><br>-- <br>Best regards,<br>Kuan-Hsu<br><br><br>
</font></blockquote></div><br></div>
</div></div><br>_______________________________________________<br>
LLVM Developers mailing list<br>
<a href="mailto:LLVMdev@cs.uiuc.edu" target="_blank">LLVMdev@cs.uiuc.edu</a> <a href="http://llvm.cs.uiuc.edu" target="_blank">http://llvm.cs.uiuc.edu</a><br>
<a href="http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev" target="_blank">http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev</a><br>
<br></blockquote></div><br></div>
</div></div></blockquote></div><br></div>
</blockquote></div><br></div>