<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<style type="text/css" style="display:none;"><!-- P {margin-top:0;margin-bottom:0;} --></style>
</head>
<body dir="ltr">
<div id="divtagdefaultwrapper" dir="ltr" style="font-size: 12pt; color: rgb(0, 0, 0); font-family: Calibri, Helvetica, sans-serif, EmojiFont, "Apple Color Emoji", "Segoe UI Emoji", NotoColorEmoji, "Segoe UI Symbol", "Android Emoji", EmojiSymbols;">
<p style="margin-top:0; margin-bottom:0">Hi,</p>
<p style="margin-top:0; margin-bottom:0"><br>
</p>
<p style="margin-top:0; margin-bottom:0">I have a compiler project using an LLVM backend (compiled for 64 bit windows), and since it has started to get to production level source code sizes the compile times have gotten somewhat extreme (around 30 mins each
time). </p>
<p style="margin-top:0; margin-bottom:0"><br>
</p>
<p style="margin-top:0; margin-bottom:0">Not sure what is the best metric for measuring whether 30 mins is reasonable. The .ll output file is ~1.5 million lines.</p>
<p style="margin-top:0; margin-bottom:0">A snip of the top of the stats output is below:</p>
<p style="margin-top:0; margin-bottom:0"><br>
</p>
<p style="margin-top:0; margin-bottom:0"></p>
<div> 2495586 asm-printer - Number of machine instrs printed</div>
<div> 2 assembler - Number of assembler layout and relaxation steps</div>
<div> 32434 assembler - Number of emitted assembler fragments - align</div>
<div> 106972 assembler - Number of emitted assembler fragments - data</div>
<div> 11179 assembler - Number of emitted assembler fragments - fill</div>
<div> 86082 assembler - Number of emitted assembler fragments - relaxable</div>
<div> 236667 assembler - Number of emitted assembler fragments - total</div>
<div>22267993 assembler - Number of emitted object file bytes</div>
<div> 1083677 assembler - Number of evaluated fixups</div>
<div> 648094 assembler - Number of fragment layouts</div>
<div> 41654 assembler - Number of relaxed instructions</div>
<div> 2345 branchfolding - Number of block tails merged</div>
<div> 687 branchfolding - Number of branches optimized</div>
<div> 1047 branchfolding - Number of dead blocks removed</div>
<div> 281 branchfolding - Number of times common instructions are hoisted</div>
<div> 8002 codegen-dce - Number of dead instructions deleted</div>
<div> 860 codegenprepare - Number of GEPs converted to casts</div>
<div> 8907 codegenprepare - Number of blocks eliminated</div>
<div> 591 codegenprepare - Number of memory instructions whose address computations were sunk</div>
<div> 801 codegenprepare - Number of uses of Cast expressions replaced with uses of sunken Casts</div>
<div> 203 codegenprepare - Number of uses of Cmp expressions replaced with uses of sunken Cmps</div>
<div> 315089 dagcombine - Number of dag nodes combined</div>
<div> 135493 isel - Number of blocks selected using DAG</div>
<div> 695 isel - Number of entry blocks encountered</div>
<div> 9853322 isel - Number of times dag isel has to try another path</div>
<div><br>
</div>
<div>The top of the pass execution table is:</div>
<div><br>
</div>
<div>
<div> Total Execution Time: 2061.1788 seconds (2068.1383 wall clock)</div>
<div><br>
</div>
<div> ---User Time--- --System Time-- --User+System-- ---Wall Time--- --- Name ---</div>
<div> 682.0364 ( 33.1%) 0.0468 ( 2.6%) 682.0832 ( 33.1%) 684.0131 ( 33.1%) Machine Instruction Scheduler</div>
<div> 309.1940 ( 15.0%) 1.0920 ( 60.9%) 310.2860 ( 15.1%) 311.8238 ( 15.1%) X86 DAG->DAG Instruction Selection</div>
<div> 295.4971 ( 14.3%) 0.2964 ( 16.5%) 295.7935 ( 14.4%) 296.6090 ( 14.3%) Greedy Register Allocator</div>
<div> 231.7395 ( 11.3%) 0.0156 ( 0.9%) 231.7551 ( 11.2%) 232.4293 ( 11.2%) Simple Register Coalescing</div>
<div> 196.9981 ( 9.6%) 0.0000 ( 0.0%) 196.9981 ( 9.6%) 197.6663 ( 9.6%) Live Variable Analysis</div>
<div> 49.5459 ( 2.4%) 0.0156 ( 0.9%) 49.5615 ( 2.4%) 49.7798 ( 2.4%) Live Interval Analysis</div>
<div> 23.6342 ( 1.1%) 0.0000 ( 0.0%) 23.6342 ( 1.1%) 23.8153 ( 1.2%) Machine Copy Propagation Pass</div>
<div> 21.9181 ( 1.1%) 0.0000 ( 0.0%) 21.9181 ( 1.1%) 21.8963 ( 1.1%) Machine Loop Invariant Code Motion</div>
</div>
<div><br>
</div>
<div><br>
</div>
<p></p>
<p style="margin-top:0; margin-bottom:0">I'm looking to find out whether this is something we can affect significantly via changing our IR code gen phase, altering llvm (or opt) parameters or something else. </p>
<p style="margin-top:0; margin-bottom:0"><br>
</p>
<p style="margin-top:0; margin-bottom:0">Is the compile time likely to be exponential with source file size or something like that? (the compile above is currently combining a number of modules that we could perhaps separate).</p>
<p style="margin-top:0; margin-bottom:0"><br>
</p>
<p style="margin-top:0; margin-bottom:0">Also, we are on a pretty old version now (3.7) as that was what was current when the project started in 2016, but would rather not upgrade (right now) if it is not likely to affect the performance here. Would be interested
in opinion in whether this is likely to have an effect. Was planning on dumping an IR file and running it through 6.0 independently when I have time to build that for windows version. </p>
<p style="margin-top:0; margin-bottom:0"><br>
</p>
<p style="margin-top:0; margin-bottom:0">Any advice appreciated.</p>
<p style="margin-top:0; margin-bottom:0"><br>
</p>
<p style="margin-top:0; margin-bottom:0"><br>
</p>
<p style="margin-top:0; margin-bottom:0">Regards,</p>
<p style="margin-top:0; margin-bottom:0">Gordon</p>
<p style="margin-top:0; margin-bottom:0"><br>
</p>
<p style="margin-top:0; margin-bottom:0"><br>
</p>
</div>
</body>
</html>