similar to: AutoFDO sample profiles v. SelectInst,

On Fri, Aug 12, 2016 at 12:15 PM, Xinliang David Li via llvm-dev < llvm-dev at lists.llvm.org> wrote: > +dehao. > > There are two potential problems: > > 1) the branch gets eliminated in the binary that is being profiled, so > there is no profile data > This seems like a fundamental problem for PGO. Maybe it is also responsible for this bug:

Hi, all Recently I'm trying to use LLVM PGO and autoFDO. However I have some problems in the process. LLVM source code is updated on April 9th. Operating system is SUSE x86_64 1. Problems in instrumentation based PGO: clang -O2 -fprofile-instr-generate test.c -o a.out ./a.out (then default.profraw is generated) clang -O2 -fprofile-instr-use=default.profraw test.c -o a.out

Hi all, Here I include an RFC for control flow sensitive AutoFDO (FS-AFDO). This is a joint work with David Li. Questions and feedback are welcome. Thanks, Rong ============= [RFC] Control Flow Sensitive AutoFDO (FS-AFDO) 1. Motivation AFDO profile is derived from PMU samples from running an earlier build binary. PMU samples are indexed by the IP addresses. An offline tool uses the debug

On 2/7/2017 20:02, Kostya Serebryany wrote: > ... > > My understanding is that BNDCU is the cheapest possible instruction, > just like XOR or ADD, > so the overhead should be relatively small. > Still my guesstimate would be >= 5% since stores are very numerous. > And such overhead will be on top of whatever overhead SafeStack has. > Do you have any measurements to

Hello; I was trying to create an executable from the "hmmer" benchmark source code files after applying some optimization. I am running the following script - #!/bin/bash while read line; do clang -S -emit-llvm -DNDEBUG -DSPEC_CPU -o $line'.o' $line'.c' done < files.txt while read line; do opt -block-placement $line'.o' -o $line'.o' done <

With recent bug fixes and performance tunings, AutoFDO at llvm has reached a usable state. To evaluate performance, we used O3/-fprofile-use/-fprofile-sample-use respectively to optimize clang itself, and measure its speed. clang built with -fprofile-use is ~20% faster than clang built with O3 clang built with -fprofile-sample-use is ~10% faster than clang built with O3 AutoFDO can deliver 50%

Michael Spencer via llvm-dev <llvm-dev at lists.llvm.org> writes: > I've recently implemented profile guided section layout in llvm + lld using > the Call-Chain Clustering (C³) heuristic from > https://research.fb.com/wp-content/uploads/2017/01/cgo2017-hfsort-final1.pdf > . In the programs I've tested it on I've gotten from 0% to 5% performance > improvement over

Hello! (+cc Adam Nemet, since he presented on optimization remarks at LLVM Dev Mtg 2016) I have a large C++ program, which I am compiling using a sampling profile generated via perf and AutoFDO. I'd like to use this profile in order to show the hotness of each code path that is displayed in the new optimization remarks viewer tool ( https://www.youtube.com/watch?v=qq0q1hfzidg). It seems,

Hi, I am getting this when running make-test :- $ /usr/src/llvm-test-2.5/configure checking for spec95 benchmark sources... no, not found in /home/vadve/shared/ben chmarks/spec95/benchspec checking for spec2000 benchmark sources... no, not found in /home/vadve/shared/b enchmarks/speccpu2000/benchspec checking for spec2006 benchmark sources... no, not found in /home/vadve/shared/b

Hi Rong, This is a very interesting proposal. We've also observed profile quality degradation from CFG destructive pass like loop rotate, and I can see how this proposal would help improve quality of profile that drives later optimization passes in the pipeline. I have a few questions. * How does this affect today's AutoFDO? Specifically, can users upgrade compiler with FS-AutoFDO

On Mon, Feb 23, 2009 at 6:09 PM, Dale Johannesen <dalej at apple.com> wrote: > These benchmarks are not distributed with llvm (which doesn't have the > legal right to distribute Spec, for example). If you have them from > another source, you need to configure --with-externals=<path> Okay, thanks. Thre should be better warning/error reporting though. Aaron > >

These benchmarks are not distributed with llvm (which doesn't have the legal right to distribute Spec, for example). If you have them from another source, you need to configure --with-externals=<path> On Feb 23, 2009, at 10:02 AMPST, Aaron Gray wrote: > Hi, > > I am getting this when running make-test :- > > $ /usr/src/llvm-test-2.5/configure > checking for

I am trying to work through the usage of thinlto for distributed builds. Here is the simple thinlto usage, just add -flto=thin everywhere, easy: clang++ -flto=thin -O3 -c -o CreateWay_.o -DSPEC_CPU -DNDEBUG -DSPEC_CPU_LITTLE_ENDIAN -Wno-dangling-else CreateWay_.cpp clang++ -flto=thin -O3 -c -o Places_.o -DSPEC_CPU -DNDEBUG -DSPEC_CPU_LITTLE_ENDIAN

On Mon, Feb 23, 2009 at 6:22 PM, Aaron Gray <aaronngray.lists at googlemail.com > wrote: > On Mon, Feb 23, 2009 at 6:09 PM, Dale Johannesen <dalej at apple.com> wrote: > >> These benchmarks are not distributed with llvm (which doesn't have the >> legal right to distribute Spec, for example). If you have them from >> another source, you need to configure

Thanks Teresa Yes it is astar, happen to send a tar of the sources but they are just copies from the spec distribution The ld command is: GNU ld (GNU Binutils) 2.29.1 Thanks for the guidance on path names. The prefix-replace just effects the string written to the object files right? So we could post-process that file with other tools as well, correct? Thanks again --david From: Teresa Johnson

Fails with gold too: Library-native.o:Library.cpp:regway: error: undefined reference to 'vtable for regwayobj' /home/dcallahan/fbsource/fbcode/third-party-buck/platform007/tools/binutils/bin/gold/ld: the vtable symbol may be undefined because the class is missing its key function clang-8: error: linker command failed with exit code 1 (use -v to see invocation) From: Teresa Johnson

Hi, We'd like to propose new Loop Distribution pass. The main motivation is to allow partial vectorization of loops. One such example is the main loop of 456.hmmer in SpecINT_2006. The current version of the patch improves hmmer by 24% on ARM64 and 18% on X86. The goal of the pass is to distribute a loop that can't be vectorized because of memory dependence cycles. The pass splits

Hi David, > Are the data below all collected when only one function is picked for > instrumentation? Yes, here is a list of the benchmarks and selected functions. +-----------------+----------------------------------------------------------------------------------------------+ | blks | _Z19BlkSchlsEqEuroNoDivfffffif |

On 08/16/2013 01:42 PM, Renato Golin wrote: > On 16 August 2013 20:02, reed kotler <rkotler at mips.com > <mailto:rkotler at mips.com>> wrote: > > -std=gnu89 is not valid for c++ > > > I think the point here is that this is the default std for GCC but not > Clang, so you have to force clang to behave like GCC. For C++, you'll > have to force

Hi, It was suggested that I post my question regarding a LLVM 3.4 performance regression to this mailing list, rather than stackoverflow. So here is the link: https://stackoverflow.com/questions/22902034/llvm-3-4-performance-regressed Thanks :) Jens -- Jens Tröger http://savage.light-speed.de/