<div dir="ltr"><div dir="ltr"><div dir="ltr"><div>Glad to hear that there is an interest in a function splitting pass. There are advantages to splitting functions at different stages as you've already noted.</div><div>- Having a target independent function splitting scales well to LTO, ThinLTO, supporting multiple architectures and offers ease of maintenance.</div><div>- While HCS+merge-function helps significantly reduce the codesize, in many cases the outlined functions tend to have identical function bodies (e.g., assert-fail etc); they can be deduplicated by linker with careful function naming. This reduces code-size regardless of function merging and across the entire program. This technique should also help Machine Function splitter in some cases but at some cost to the link time.</div><div><div><div>- It will be difficult to reduce argument setup and restore code in the HCS except in some cases like tail call, internal function calls, non-returning function calls etc. Having frame setup however should help with debugging IMO.</div><div></div></div></div><div><br></div><div>> <span style="color:rgb(0,0,0);white-space:pre-wrap">Recent work at Google has shown that aggressive, profile-driven inlining </span><span style="color:rgb(0,0,0);white-space:pre-wrap">for performance has led to significant code bloat and icache </span><span style="color:rgb(0,0,0);white-space:pre-wrap">fragmentation</span></div><pre style="white-space:pre-wrap;color:rgb(0,0,0)"><span style="font-family:Arial,Helvetica,sans-serif;color:rgb(34,34,34)">Because the inliner can be too aggressive at times and can negatively affect icache-miss etc, HCS can be integrated with inliner to assist in partial inlining (For example: split the callee before inlining). Rodrigo (@rcorcs) has some ideas around that and we've been exploring that as part of GSoC project.</span></pre><div>We've performance numbers from Firefox which shows ~5% performance improvement with HCS (cc: Ruijie @rjf). Vedant also reported performance numbers across iOS and Swift benchmarks in the past. I could find (<a href="https://github.com/apple/swift/pull/21016">https://github.com/apple/swift/pull/21016</a>) which reported decent performance improvement in core ios Frameworks.</div><div><br></div><div>That said we have been working on improving the cost model which I think will help alleviate many of the limitations that we typically don't have in a Machine Splitting optimization. I'd like to hear your ideas on how the cost model can be improved.</div><div><br></div><div>> <span style="color:rgb(0,0,0);white-space:pre-wrap">In contrast, the machine function splitter extracts cold code into a </span><span style="color:rgb(0,0,0);white-space:pre-wrap">separate section. </span></div><div>HCS also adds a section prefix to all the cold functions. It is possible that the cold functions are still in the same section as the hot one depending on the linker. Ruijie has a patch to move all the cold functions to a separate section, we are still evaluating the results (<a href="https://github.com/ruijiefang/llvm-hcs/commit/4966997e135050c99b4adc0aa971242af99ba832">https://github.com/ruijiefang/llvm-hcs/commit/4966997e135050c99b4adc0aa971242af99ba832</a>). In case it is not difficult to rerun the experiments, it'll help to see the numbers with the llvm-trunk and this patch from @rjf.</div><div><br></div><div>> <span style="color:rgb(0,0,0);white-space:pre-wrap">Furthermore, this may not play well with</span><span style="color:rgb(0,0,0);white-space:pre-wrap"> optimization passes such as MachineOutliner.</span></div><div><span style="color:rgb(0,0,0);white-space:pre-wrap">Can you share an example where HCS and Machine Outliner don't play well together?</span></div><div><span style="color:rgb(0,0,0);white-space:pre-wrap"><br></span></div><div><span style="color:rgb(0,0,0);white-space:pre-wrap">> </span><span style="color:rgb(0,0,0);white-space:pre-wrap">Synergistic optimizations are harder to reason about due to the pass</span><span style="color:rgb(0,0,0);white-space:pre-wrap"> timing. For example, inlining can be more aggressive if any cold code </span><span style="color:rgb(0,0,0);white-space:pre-wrap">introduced is trimmed.</span></div><div><font color="#000000"><span style="white-space:pre-wrap">How does this regress workloads if we have profile information and cold portions of a callee is outlined? Does the inliner always regress workloads we are evaluating?</span></font></div><div><div><div><br></div><div><div>PS: One correction I'd like to make is HCS splits SEME regions (Thanks to Vedant). In case some SEME aren't getting outlined, there are CFG transformations to make them friendly to HCS. I'd love to see such an example, as that'd motivate some of the future work.</div><div></div><br></div></div></div><div><div dir="ltr" class="gmail_signature"><div dir="ltr"><div><div dir="ltr"><div><div dir="ltr">Aditya Kumar<div>Compiler Engineer</div><div><a href="https://bitsimplify.com" target="_blank">https://bitsimplify.com</a><br></div></div></div></div></div></div></div></div><br></div></div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Tue, Aug 4, 2020 at 5:31 PM Snehasish Kumar <<a href="mailto:snehasishk@google.com">snehasishk@google.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr"><div class="gmail_default" style="font-size:small"><span id="gmail-m_-2371679712666550029gmail-docs-internal-guid-773c91dc-7fff-a76a-fefe-62de2a5f322c"><h1 style="line-height:1.38;text-align:justify;margin-top:20pt;margin-bottom:6pt"><span style="background-color:transparent;color:rgb(0,0,0);font-family:Arial;font-size:11pt;white-space:pre-wrap;font-weight:normal">Greetings,</span><br></h1><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="background-color:transparent;color:rgb(0,0,0);font-family:Arial;font-size:11pt;white-space:pre-wrap">We present “Machine Function Splitter”, a codegen optimization pass which splits functions into hot and cold parts. This pass leverages the basic block sections feature recently introduced in LLVM from the Propeller project. The pass targets functions with profile coverage, identifies cold blocks and moves them to a separate section. The linker groups all cold blocks across functions together, decreasing fragmentation and improving icache and itlb utilization. Our experiments show >2% performance improvement on clang bootstrap, ~1% improvement on Google workloads and 1.6% mean performance improvement on SPEC IntRate 2017.</span><br></p><h2 dir="ltr" style="line-height:1.38;text-align:justify;margin-top:18pt;margin-bottom:6pt"><span style="font-size:16pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-weight:400;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Motivation</span></h2><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Recent work at Google has shown that aggressive, profile-driven inlining for performance has led to significant code bloat and icache fragmentation (</span><a href="https://research.google/pubs/pub48320/" style="text-decoration-line:none" target="_blank"><span style="font-size:11pt;font-family:Arial;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;text-decoration-line:underline;vertical-align:baseline;white-space:pre-wrap">AsmDB - Ayers et al ‘2019</span></a><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">). We find that most functions 5 KiB or larger have inlined children more than 10 layers deep bringing in exponentially more code at each inline level, not all of which is necessarily hot. Generally, in roughly half of even the hottest functions, more than 50% of the code bytes are never executed, but likely to be in the cache.</span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Function splitting is a well known compiler transformation primarily targeting improved code locality to improve performance. LLVM has</span><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> a middle-end, target agnostic </span><a href="https://llvm.org/devmtg/2019-10/slides/Kumar-HotColdSplitting.pdf" style="text-decoration-line:none" target="_blank"><span style="font-size:11pt;font-family:Arial;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;text-decoration-line:underline;vertical-align:baseline;white-space:pre-wrap">hot cold splitting pass</span></a><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> as well as a </span><a href="https://github.com/llvm/llvm-project/blob/master/llvm/lib/Transforms/IPO/PartialInlining.cpp" style="text-decoration-line:none" target="_blank"><span style="font-size:11pt;font-family:Arial;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;text-decoration-line:underline;vertical-align:baseline;white-space:pre-wrap">partial inlining pass</span></a><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> which performs similar transformations, as noted by the authors in </span><a href="https://lists.llvm.org/pipermail/llvm-dev/2020-June/142429.html" style="text-decoration-line:none" target="_blank"><span style="font-size:11pt;font-family:Arial;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;text-decoration-line:underline;vertical-align:baseline;white-space:pre-wrap">a recent email thread</span></a><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">. However, due to the timing of the respective passes as well as the code extraction techniques employed, the overall gains on large, complex applications leave headroom for improvement. By deferring function splitting to the codegen phase we can maximize the opportunity to remove cold code as well as refine the code extraction technique. Furthermore, by performing function splitting very late, earlier passes can perform more aggressive optimizations.</span></p><h2 dir="ltr" style="line-height:1.38;text-align:justify;margin-top:18pt;margin-bottom:6pt"><span style="font-size:16pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-weight:400;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Implementation</span></h2><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">We propose a new machine function splitting pass which leverages the </span><a href="https://reviews.llvm.org/D68063" style="text-decoration-line:none" target="_blank"><span style="font-size:11pt;font-family:Arial;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;text-decoration-line:underline;vertical-align:baseline;white-space:pre-wrap">basic block sections feature</span></a><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> to split functions without the caveats of code extraction in the middle-end. The pass uses profile information to identify cold basic blocks very late in LLVM CodeGen, after regalloc and all other machine passes have executed. This allows our implementation to be precise in its assessment of cold regions while maximizing opportunity.</span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Each function is split into two parts. The hot cluster includes the function entry and all blocks which are not cold. All the cold blocks are grouped together as a </span><a href="https://github.com/llvm/llvm-project/blob/5934df0c9abe94fc450fbcf0ceca21cf838840e9/llvm/include/llvm/CodeGen/MachineBasicBlock.h#L63" style="text-decoration-line:none" target="_blank"><span style="font-size:11pt;font-family:Arial;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;text-decoration-line:underline;vertical-align:baseline;white-space:pre-wrap">Cold Section cluster</span></a><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">. With basic block sections, the cold blocks are assigned appropriate debug and call frame information and emitted as part of the .text.unlikely section. Unlike </span><a href="https://lists.llvm.org/pipermail/llvm-dev/2019-September/135393.html" style="text-decoration-line:none" target="_blank"><span style="font-size:11pt;font-family:Arial;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;text-decoration-line:underline;vertical-align:baseline;white-space:pre-wrap">Propeller</span></a><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">, which is presently the main user of the basic block sections feature, this pass does not require an additional round of profiling and uses existing instrumentation based FDO or CSFDO profile information.</span></p><br><div><img src="cid:ii_kdgmbku10" alt="Machine Function Splitter.png" style="margin-right: 0px;" width="770" height="577"><br></div><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"></span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">In the illustration above, the functions foo and bar contain a cold block each, index 5 and E respectively. We show a possible layout for these functions which optimizes for fall throughs. Note that all the blocks are kept in a contiguous region described by the symbols foo and bar. Using the machine function splitter, the cold blocks (5 and E) are moved to a separate section. These blocks can then be grouped along with other cold blocks (and functions) in a separate output section in the final binary. The key highlights of this approach are:</span></p><ul style="margin-top:0px;margin-bottom:0px"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Profile driven, profile type agnostic approach.</span></p></li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Cold basic blocks are split out using jumps.</span></p></li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">No additional instructions are added to the function for setup/teardown.</span></p></li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Runs as the last step before emitting assembly, no analysis/optimizations are hindered.</span></p></li></ul><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Exceptions</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">All eh pads are grouped together regardless of their coldness and are part of the original function. There are outstanding issues with splitting eh pads if they reside in separate sections in the binary. This remains as part of future work.</span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">DebugInfo and CFI </span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Debug information and CFI directives are updated and kept consistent by the underlying basic block sections framework. Support added in the following patches</span></p><ul style="margin-top:0px;margin-bottom:0px"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">DebugInfo (</span><a href="https://reviews.llvm.org/D78851" style="text-decoration-line:none" target="_blank"><span style="font-size:11pt;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;text-decoration-line:underline;vertical-align:baseline;white-space:pre-wrap">https://reviews.llvm.org/D78851</span></a><span style="font-size:11pt;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">)</span></p></li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">CFI (</span><a href="https://reviews.llvm.org/D79978" style="text-decoration-line:none" target="_blank"><span style="font-size:11pt;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;text-decoration-line:underline;vertical-align:baseline;white-space:pre-wrap">https://reviews.llvm.org/D79978</span></a><span style="font-size:11pt;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">).</span></p></li></ul><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-weight:700;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"><br></span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-weight:700;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Distinction between Machine Function Splitter and Propeller</span></p><p dir="ltr" style="line-height:1.2;text-align:justify;margin-top:0pt;margin-bottom:10pt"><span style="font-size:11pt;font-family:Arial;color:rgb(32,33,36);font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"><br></span><span style="font-size:11pt;font-family:Arial;color:rgb(32,33,36);font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Full Propeller optimizations include function splitting and layout optimizations, however it requires an additional round of profiling using perf on top of the peak (FDO/CSFDO + ThinLTO) binary. In this work we experiment with applying function splitting using the instrumented profile in the build instead of adding an additional round of profiling.</span></p><p dir="ltr" style="line-height:1.2;text-align:justify;margin-top:0pt;margin-bottom:10pt"><a href="https://lists.llvm.org/pipermail/llvm-dev/2019-September/135393.html" style="text-decoration-line:none" target="_blank"><span style="font-size:11pt;font-family:Arial;font-variant-numeric:normal;font-variant-east-asian:normal;text-decoration-line:underline;vertical-align:baseline;white-space:pre-wrap">Link to Propeller RFC</span></a></p><p dir="ltr" style="line-height:1.2;text-align:justify;margin-top:0pt;margin-bottom:10pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-weight:700;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"><br></span></p><p dir="ltr" style="line-height:1.2;text-align:justify;margin-top:0pt;margin-bottom:10pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-weight:700;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Split Binary Characteristics</span></p><p dir="ltr" style="line-height:1.2;text-align:justify;margin-top:0pt;margin-bottom:10pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Binaries produced by the compiler with function splitting enabled contain additional symbols. A function which has been split into a hot and cold part is non-contiguous. The symbol table entry for the hot part retains the symbol name of the original function with type FUNC. The symbol for the cold part contains a “.cold” suffix attached to the original symbol name, the type is not set for this symbol. Using a suffix has been the norm for such optimizations e.g. -hot-cold-split in LLVM and the prior GCC implementation detailed earlier. We expect standardized tooling to handle split functions appropriately, e.g demangling works as expected --</span></p><div dir="ltr" style="margin-left:0pt" align="left"><table style="border:none;border-collapse:collapse"><colgroup></colgroup><tbody><tr style="height:0pt"><td style="border-width:1pt;border-style:solid;border-color:rgb(224,224,224);vertical-align:top;background-color:rgb(250,250,250);padding:5pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Consolas,sans-serif;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">$ c</span><span style="font-size:10pt;font-family:Consolas,sans-serif;color:rgb(97,97,97);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">++</span><span style="font-size:10pt;font-family:Consolas,sans-serif;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">filt _Z3foov</span><span style="font-size:10pt;font-family:Consolas,sans-serif;color:rgb(97,97,97);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">.</span><span style="font-size:10pt;font-family:Consolas,sans-serif;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">cold</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Consolas,sans-serif;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">foo</span><span style="font-size:10pt;font-family:Consolas,sans-serif;color:rgb(97,97,97);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">()</span><span style="font-size:10pt;font-family:Consolas,sans-serif;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> </span><span style="font-size:10pt;font-family:Consolas,sans-serif;color:rgb(97,97,97);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">[</span><span style="font-size:10pt;font-family:Consolas,sans-serif;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">clone </span><span style="font-size:10pt;font-family:Consolas,sans-serif;color:rgb(97,97,97);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">.</span><span style="font-size:10pt;font-family:Consolas,sans-serif;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">cold</span><span style="font-size:10pt;font-family:Consolas,sans-serif;color:rgb(97,97,97);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">]</span></p></td></tr></tbody></table></div><br><h2 dir="ltr" style="line-height:1.38;text-align:justify;margin-top:18pt;margin-bottom:6pt"><span style="font-size:16pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-weight:400;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Contrast with HotColdSplit (HCS)</span></h2><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Function splitting in the middle-end in LLVM employs extraction of cold single-entry-single-exit (SESE) regions into separate functions. In general, the pass has been found to be impactful in reducing code size by deduplication of cold regions; however our experiments show it does not improve performance of large workloads. </span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">The key differences are:</span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-weight:700;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Extraction methodology and tradeoffs</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">HCS extracts cold code from SESE regions using a function call. This may incur a spill and fill of caller registers along with additional setup and teardown if live values modified in the cold region need to be communicated back to the original function. This has a couple of implications</span></p><ol style="margin-top:0px;margin-bottom:0px"><li dir="ltr" style="list-style-type:decimal;font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">The “residue” of each extracted region is non-trivial and there is a tradeoff between the amount of code that needs to be cold before it is profitable to extract. Thus the cost of mischaracterization is high.</span></p></li><li dir="ltr" style="list-style-type:decimal;font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Since each SESE region is extracted separately the net reduction in code size of the original function is less.</span><span style="font-size:11pt;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"><br><br></span></p></li></ol><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">In contrast, the machine function splitter extracts cold code into a separate section. Control is transferred to cold code via jumps. More often than not these jumps may already exist as part of the original layout thus incurring no additional cost. No additional instructions are inserted to accommodate splitting. Finally, no additional setup/teardown is necessary for live values modified in cold regions.</span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-weight:700;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Pass timing and interaction with other optimizations</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">The HCS pass is run on the IR in the optimizer. This allows it to be target agnostic and allow later stages to merge identical code if necessary. However, there are some drawbacks to this approach. In particular,</span></p><ol style="margin-top:0px;margin-bottom:0px"><li dir="ltr" style="list-style-type:decimal;font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Splitting early may miss opportunities introduced by later passes such as library call inlining and CFG simplification resulting from a combination of optimizations. Furthermore, this may not play well with optimization passes such as MachineOutliner.</span></p></li><li dir="ltr" style="list-style-type:decimal;font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Synergistic optimizations are harder to reason about due to the pass timing. For example, inlining can be more aggressive if any cold code introduced is trimmed.</span><span style="font-size:11pt;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"><br><br></span></p></li></ol><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">In contrast, the machine function splitter runs as the last step in codegen. This ensures that the opportunity for splitting is maximised without hindering existing analyses and synergistic decisions can be made in earlier optimization passes. We rely on accurate profile count propagation across optimizations to maximise opportunities. This works particularly well for instrumented profiles while improving the pass for sampled profiles is ongoing work.</span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">We have provided a contrived example in the Appendix which demonstrates the code generated for both approaches. The key differences are highlighted inline.</span></p><br><h2 dir="ltr" style="line-height:1.38;text-align:justify;margin-top:18pt;margin-bottom:6pt"><span style="font-size:16pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-weight:400;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Evaluation</span></h2><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">In this section, we present an in-depth evaluation of the impact on clang bootstrap and summary results for two google internal workloads, Search1 and Search2 as well overall results on the SPECInt 2017 benchmarks. All experiments are conducted on Intel Skylake based systems unless otherwise noted. Profile guided optimizations using instrumented profiles are enabled for all builds.</span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-weight:700;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">clang-bootstrap</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">We pick 500 compiler invocations from a bootstrap build of clang and then evaluate the performance of a PGO+ThinLTO optimized version with that of PGO+ThinLTO+Split compiler. For the latter, the final optimized build includes the machine function splitter.</span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Results: </span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">We observe a mean 2.33% improvement in end to end runtime. The improvements in runtime are driven by reduction in icache and TLB miss rates. The table below summarizes our experiment, each data point is averaged over multiple iterations. The observed variation for each metric is < 1%.</span></p><br><div dir="ltr" style="margin-left:0pt" align="left"><table style="border:none;border-collapse:collapse"><colgroup><col width="100"><col width="100"><col width="116"><col width="84"></colgroup><tbody><tr style="height:26.25pt"><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Event</span></p></td><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Split (MPKI)</span></p></td><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Baseline (MPKI)</span></p></td><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">% Reduction</span></p></td></tr><tr style="height:15.75pt"><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">itlb_miss</span></p></td><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">0.87</span></p></td><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">1.28</span></p></td><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">31.70</span></p></td></tr><tr style="height:15.75pt"><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">stlb_miss</span></p></td><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">0.08</span></p></td><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">0.12</span></p></td><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">32.51</span></p></td></tr><tr style="height:15.75pt"><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">l1i_miss</span></p></td><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">5.98</span></p></td><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">6.61</span></p></td><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">9.56</span></p></td></tr><tr style="height:15.75pt"><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">l2_miss</span></p></td><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">0.27</span></p></td><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">0.34</span></p></td><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">20.02</span></p></td></tr></tbody></table></div><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">In this experiment, the function splitting pass moved cold code from ~30K functions in .text and .text.hot. We present a comparison of the binary contents using </span><a href="https://github.com/google/bloaty" style="text-decoration-line:none" target="_blank"><span style="font-size:11pt;font-family:Arial;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;text-decoration-line:underline;vertical-align:baseline;white-space:pre-wrap">bloaty</span></a></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"><span style="width:100px;display:inline-block"></span></span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> FILE SIZE VM SIZE </span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> -------------- -------------- </span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> +23% +8.26Mi +23% +8.26Mi .text.unlikely</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> +6.5% +761Ki [ = ] 0 .strtab</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> +4.8% +247Ki +4.8% +247Ki .eh_frame</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> +6.1% +193Ki [ = ] 0 .symtab</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> +8.5% +63.1Ki +8.5% +63.1Ki .eh_frame_hdr</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> +0.3% +31.3Ki +0.3% +31.3Ki .rodata</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> +0.4% +3 [ = ] 0 [Unmapped]</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> -0.3% -8 -0.3% -8 .init_array</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> [ = ] 0 -33.3% -8 [LOAD #4 [RW]]</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> [ = ] 0 -0.2% -416 .bss</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> -57.1% -4.04Mi -57.1% -4.04Mi .text.hot</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> -48.4% -4.13Mi -48.4% -4.13Mi .text</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> +1.6% +1.35Mi +0.6% +430Ki TOTAL</span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">We see that 48% and 57% of code in .text and .text.hot respectively was moved to the .text.unlikely section. We also note a small increase in overall binary size due to the following reasons:</span></p><ul style="margin-top:0px;margin-bottom:0px"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Some additional jump instructions may be inserted.</span></p></li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Small increase in associated metadata, e.g. debug information.</span></p></li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Additional symbols of type foo.cold for cold parts.</span></p></li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Alignment requirements for both original and split function parts.</span></p></li></ul><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-style:italic;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Comparison with HotColdSplit</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">For the clang-bootstrap benchmark we also compared the performance of the hot-cold-split pass with split-machine-functions. We summarize the results for performance and the characteristics of the binary built by each pass in the table below. Each metric is presented as change vs the baseline, an FDO optimized build of clang.</span></p><br><div dir="ltr" style="margin-left:0pt" align="left"><table style="border:none;border-collapse:collapse"><colgroup><col width="100"><col width="173"><col width="171"></colgroup><tbody><tr style="height:15.75pt"><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><br></td><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Hot Cold Split</span></p></td><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Machine Function Splitter</span></p></td></tr><tr style="height:15.75pt"><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Performance</span></p></td><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">1.10%</span></p></td><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">2.65%</span></p></td></tr><tr style="height:15.75pt"><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">.text size</span></p></td><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">-41.5% -2.89Mi</span></p></td><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">-49.2% -3.43Mi</span></p></td></tr><tr style="height:15.75pt"><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">.text.hot size</span></p></td><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">-46.9% -2.52Mi</span></p></td><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">-57.1% -3.07Mi</span></p></td></tr><tr style="height:15.75pt"><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Full binary size</span></p></td><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">9.6% +7.56Mi</span></p></td><td style="border-width:0.75pt;border-style:solid;border-color:rgb(204,204,204);vertical-align:bottom;padding:2pt;overflow:hidden"><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">1.7% +1.37Mi</span></p></td></tr></tbody></table></div><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Note that the increase in overall binary size increase for HCS is due to the increase in .eh_frame (+61% +3.03Mi). HCS extracts each cold SESE region as a separate function whereas the machine function splitter extracts the cold code as a single region thus incurring a constant overhead per function.</span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-weight:700;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Google workloads</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">We evaluated the impact of function splitting on a couple of search workloads, Search1 and Search2. A key difference with respect to the clang experiment above is the use of huge pages for code. Overall, we find that on Intel Skylake the key benefit is from reduction of iTLB misses whereas on AMD the key benefit is from the reduction of icache misses. This is due to the fewer iTLB entries available for hugepages on Intel architectures. We find that overall throughput for Search1 and Search2 improve between 0.8% to 1.2%; a significant improvement on these benchmarks. The workloads are built with FDO and CSFDO respectively. On Intel Skylake, iTLB misses reduce by 16% to 35%, sTLB misses reduce by 62% to 67%. On AMD, L1 icache misses improve by 1.2% to 2.6% whereas L2 instruction misses improve by 4.8% to 5.1%. </span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-style:italic;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Comparison with HotColdSplit</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">An evaluation of the hot-cold-split pass did not yield performance improvements on google workloads.</span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-weight:700;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">SPECInt 2017</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">We evaluated the impact of the machine function splitter on SPECInt 2017 using the int rate metrics. Overall, we found a 1.6% geomean intrate improvement for the benchmarks where performance improved (500.perlbench_r, 502.gcc_r, 505.mcf_r, 520.omnetpp_r). For the benchmarks that didn’t improve performance, the average degradation was 0.6% (523.xalancbmk_r, 525.x264_r, 531.deepsjeng_r, 541.leela_r). </span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">We note that the instruction footprint of SPEC workloads are smaller than most modern workloads and our work is primarily focused on reducing the footprint to improve performance. These experiments were performed on Intel Haswell machines.</span></p><br><h2 dir="ltr" style="line-height:1.38;text-align:justify;margin-top:18pt;margin-bottom:6pt"><span style="font-size:16pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-weight:400;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Appendix</span></h2><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Example to illustrate hot-cold-split and split-machine-functions</span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Input IR</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">```</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">@i = external global i32, align 4</span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">define i32 @foo(i32 %0, i32 %1) nounwind !prof !1 {</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> %3 = icmp eq i32 %0, 0</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> br i1 %3, label %6, label %4, !prof !2</span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">4: ; preds = %2</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> %5 = call i32 @L1()</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> br label %9</span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">6: ; preds = %2</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> %7 = call i32 @R1()</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> %8 = add nsw i32 %1, 1</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> br label %9</span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">9: ; preds = %6, %4</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> %10 = phi i32 [ %1, %4 ], [ %8, %6 ]</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> %11 = load i32, i32* @i, align 4</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> %12 = add nsw i32 %10, %11</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> store i32 %12, i32* @i, align 4</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> ret i32 %12</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">}</span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">declare i32 @L1()</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">declare i32 @R1() cold nounwind</span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">!1 = !{!"function_entry_count", i64 7}</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">!2 = !{!"branch_weights", i32 0, i32 7}</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">```</span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Code generated by Machine Function Splitter</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">$ llc < example.ll -mtriple=x86_64-unknown-linux-gnu -split-machine-functions</span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">```</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> .text</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> .file "<stdin>"</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> .globl foo # -- Begin function foo</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> .p2align 4, 0x90</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> .type foo,@function</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">foo: # @foo</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"># %bb.0:</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> pushq %rbx</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> movl %esi, %ebx</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> testl %edi, %edi</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> </span><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:rgb(255,255,0);font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">je foo.cold # Jump to cold code</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"># %bb.1:</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> callq L1</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">.LBB0_2:</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> addl i(%rip), %ebx</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> movl %ebx, i(%rip)</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> movl %ebx, %eax</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> popq %rbx</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> retq</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> .section .text.unlikely.foo,"ax",@progbits</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">foo.cold:</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> callq R1</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> </span><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:rgb(255,255,0);font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">incl %ebx # Directly increment value</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> jmp .LBB0_2</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">.LBB_END0_3:</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> .size foo.cold, .LBB_END0_3-foo.cold</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> .text</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">.Lfunc_end0:</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> .size foo, .Lfunc_end0-foo</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> # -- End function</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> .section ".note.GNU-stack","",@progbits</span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">```</span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">Code generated by Hot Cold Split</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">$ clang -c -O2 -S -mllvm --hot-cold-split -mllvm --hotcoldsplit-threshold=0 -x ir example.ll</span></p><br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">```</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> .text</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> .file "example.ll"</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> .globl foo # -- Begin function foo</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> .p2align 4, 0x90</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> .type foo,@function</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">foo: # @foo</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"># %bb.0:</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> pushq %rbx</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> subq $16, %rsp</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> movl %esi, %ebx</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> testl %edi, %edi</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> jne .LBB0_1</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:rgb(255,255,0);font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"># %bb.2: # Residue block in original function</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> leaq 12(%rsp), %rsi</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> </span><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:rgb(255,255,0);font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">movl %ebx, %edi # Pass param to increment</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> </span><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:rgb(255,255,0);font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">callq foo.cold.1 # Call to cold code</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> </span><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:rgb(255,255,0);font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">movl 12(%rsp), %ebx # Fill incremented value from stack</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">.LBB0_3:</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> addl i(%rip), %ebx</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> movl %ebx, i(%rip)</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> movl %ebx, %eax</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> addq $16, %rsp</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> popq %rbx</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> retq</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">.LBB0_1:</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> callq L1</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> jmp .LBB0_3</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">.Lfunc_end0:</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> .size foo, .Lfunc_end0-foo</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> # -- End function</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> .p2align 4, 0x90 # -- Begin function foo.cold.1</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> .type foo.cold.1,@function</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">foo.cold.1: # @foo.cold.1</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"># %bb.0: # %newFuncRoot</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> pushq %rbp</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> pushq %rbx</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> pushq %rax</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> movq %rsi, %rbx</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> movl %edi, %ebp</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> callq R1</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> incl %ebp</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> movl %ebp, (%rbx)</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> addq $8, %rsp</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> popq %rbx</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> popq %rbp</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> retq</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">.Lfunc_end1:</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> .size foo.cold.1, .Lfunc_end1-foo.cold.1</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> # -- End function</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> .cg_profile foo, L1, 0</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> .cg_profile foo, foo.cold.1, 7</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> .section ".note.GNU-stack","",@progbits</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> .addrsig</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:10pt;font-family:"Courier New";color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> .addrsig_sym foo.cold.1</span></p><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap">```</span></p><div class="gmail_default" style="font-size:small"><span><br></span></div>Thanks,</span></div><div class="gmail_default" style="font-size:small"><span>Snehasish Kumar</span></div><div class="gmail_default" style="font-size:small"><span>Software Engineer, Google<br><p dir="ltr" style="line-height:1.38;text-align:justify;margin-top:0pt;margin-bottom:0pt"><span style="font-size:11pt;font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-variant-numeric:normal;font-variant-east-asian:normal;vertical-align:baseline;white-space:pre-wrap"> </span></p></span><br></div></div>
</blockquote></div>