Databases
StackOverflow2013
Postsdbo
POWhy does compiler inlining produce slower code than manual inlining?
 

Note that there are some explanatory texts on larger screens.

plurals
  1. POWhy does compiler inlining produce slower code than manual inlining?
    primarykey
    data
    text
    <h2>Background</h2> <p>The following critical loop of a piece of numerical software, written in C++, basically compares two objects by one of their members:</p> <pre><code>for(int j=n;--j&gt;0;) asd[j%16]=a.e&lt;b.e; </code></pre> <p><code>a</code> and <code>b</code> are of class <code>ASD</code>:</p> <pre><code>struct ASD { float e; ... }; </code></pre> <p>I was investigating the effect of putting this comparison in a lightweight member function:</p> <pre><code>bool test(const ASD&amp; y)const { return e&lt;y.e; } </code></pre> <p>and using it like this:</p> <pre><code>for(int j=n;--j&gt;0;) asd[j%16]=a.test(b); </code></pre> <p>The compiler is inlining this function, but the problem is, that the assembly code will be different and cause >10% of runtime overhead. I have to question:</p> <h2>Questions</h2> <ol> <li><p>Why is the compiler prodrucing different assembly code?</p></li> <li><p>Why is the produced assembly slower?</p></li> </ol> <p><strong>EDIT:</strong> The second question has been answered by implementing @KamyarSouri's suggestion (j%16). The assembly code now looks almost identical (see <a href="http://pastebin.com/diff.php?i=yqXedtPm" rel="noreferrer">http://pastebin.com/diff.php?i=yqXedtPm</a>). The only differences are the lines 18, 33, 48:</p> <pre><code>000646F9 movzx edx,dl </code></pre> <h2>Material</h2> <ul> <li>The test code: <a href="http://pastebin.com/03s3Kvry" rel="noreferrer">http://pastebin.com/03s3Kvry</a></li> <li>The assembly output on MSVC10 with /Ox /Ob2 /Ot /arch:SSE2: <ul> <li>Compiler inlined version: <a href="http://pastebin.com/yqXedtPm" rel="noreferrer">http://pastebin.com/yqXedtPm</a></li> <li>Manually inlined version: <a href="http://pastebin.com/pYSXL77f" rel="noreferrer">http://pastebin.com/pYSXL77f</a></li> <li>Difference <a href="http://pastebin.com/diff.php?i=yqXedtPm" rel="noreferrer">http://pastebin.com/diff.php?i=yqXedtPm</a></li> </ul></li> </ul> <p>This chart shows the FLOP/s (up to a scaling factor) for 50 testruns of my code.</p> <p><img src="https://i.stack.imgur.com/BlGeJ.png" alt="enter image description here"></p> <p>The gnuplot script to generate the plot: <a href="http://pastebin.com/8amNqya7" rel="noreferrer">http://pastebin.com/8amNqya7</a></p> <p>Compiler Options:</p> <p>/Zi /W3 /WX- /MP /Ox /Ob2 /Oi /Ot /Oy /GL /D "WIN32" /D "NDEBUG" /D "_CONSOLE" /D "_UNICODE" /D "UNICODE" /Gm- /EHsc /MT /GS- /Gy /arch:SSE2 /fp:precise /Zc:wchar_t /Zc:forScope /Gd /analyze-</p> <p>Linker Options: /INCREMENTAL:NO "kernel32.lib" "user32.lib" "gdi32.lib" "winspool.lib" "comdlg32.lib" "advapi32.lib" "shell32.lib" "ole32.lib" "oleaut32.lib" "uuid.lib" "odbc32.lib" "odbccp32.lib" /ALLOWISOLATION /MANIFESTUAC:"level='asInvoker' uiAccess='false'" /SUBSYSTEM:CONSOLE /OPT:REF /OPT:ICF /LTCG /TLBID:1 /DYNAMICBASE /NXCOMPAT /MACHINE:X86 /ERRORREPORT:QUEUE </p>
    singulars
    1. PO
      singulars
      1. PTAnswer
    1. This table or related slice is empty.
    1. PTQuestion
    1. USMysticial
    1. USJohannes Gerer
    plurals
    1. This table or related slice is empty.
    1. This table or related slice is empty.
    1. This table or related slice is empty.
    1. PO
      singulars
      1. PTAnswer
    2. PO
      singulars
      1. PTAnswer
    1. VO
      singulars
      1. POWhy does compiler inlining produce slower code than manual inlining?
      1. This table or related slice is empty.
      1. VTUpMod
    2. VO
      singulars
      1. POWhy does compiler inlining produce slower code than manual inlining?
      1. USsje397
      1. VTFavorite
    3. VO
      singulars
      1. POWhy does compiler inlining produce slower code than manual inlining?
      1. This table or related slice is empty.
      1. VTUpMod
    1. CONice question. It might be instructive to post the optimization and other relevant compiler settings used to produce this result.
      singulars
      1. POWhy does compiler inlining produce slower code than manual inlining?
      1. USTed Hopp
    2. CO@user578832: not an answer to your question but wouldn't *"loop unrolling"* be useful in your case? Wouldn't it help get more precise readings between the two methods you're comparing? (you may be spending less time in the *for* comparison and you may notice even bigger differences)
      singulars
      1. POWhy does compiler inlining produce slower code than manual inlining?
      1. USTacticalCoder
    3. COI haven't read the assembler, but won't the compiler pull the comparison out of the loop, because it's the same thing on each iteration?
      singulars
      1. POWhy does compiler inlining produce slower code than manual inlining?
      1. USOliver Charlesworth
 

Querying!

 
Guidance
A row detail

Detail views are divided into sections. All the information in the data section comes from columns in the selected row. The other sections display data from other, related rows.

Related data can be related in a to-one or a to-many fashion. Captions of data related in a to-many fashion link to a list view showing a filtered view of the table.

Try moving around until you find a non-empty to-many entry and click on the label to get to one. You can move back to the root by clicking on the database name in the header.

SQuiL has stopped working due to an internal error.

If you are curious you may find further information in the browser console, which is accessible through the devtools (F12).

Reload