similar to: FW: [LLVMdev] Building 'as' for Olden

On Sun, 8 Dec 2002, Chris Lattner wrote: > Olden itself doesn't need it. _Compiling_ Olden requires the assembler to > turn llvm .s files into llvm .o files. Why not just use the bytecode > files that Misha posted? Right now I can't even get the LLVM assembly file to produce from Olden. The bytecodes that Misha posted are some ancient Olden version that doesn't work

On Sun, 8 Dec 2002, Juan Nicolas Ruiz wrote: > does anyone knows how to get the Olden BM bytecode files? http://llvm.cs.uiuc.edu/LLVMPrograms/Olden/ -- Misha Brukman :: uiuc.edu/~brukman

When I try to compile the olden bc's off the webpage I am getting undefined symbols. My compile script works as follows: #!/bin/sh echo "Generating assembly file." llc -f $1.bc echo "Assembling s file." /opt/SUNWspro/bin/cc -xarch=v9 -lm $1.s -o $1 echo "Removing s file." rm $1.s and here is an example of the symbol issues: Undefined first referenced symbol

does anyone knows how to get the Olden BM bytecode files? We're low on disk space and I don't want to build the whole test branch (even if i can, which I doubt, it failed last time i tried).

On Sun, 8 Dec 2002, Chris Lattner wrote: > If you recompile the bytecode files now, (go into each directory and type > 'make final' they should be ok. Perhaps someone can do this and post the > results for everyone to use... I put up what I could compile here: http://tank.cs.uiuc.edu/Olden/ However, I could not get mst to compile: % make final dis: bytecode didn't read

I've also been trying to compile the Olden benchmarks in /test/Programs/MultiSource/Olden-* but to no effect. I just did a full CVS update -d on the llvm tree but when I try compiling say Olden-bh I get this error: gmake: *** No rule to make target `Output/args.rbc', needed by `Output/bh.linked.rbc'. Stop. We've also tried downloading the Olden benchmarks directly from the URL

On Sun, 8 Dec 2002, Chris Lattner wrote: > > For each bmk, the undefined symbols are coming from args.c as far as I can > > tell. Do I need to link in args.c's bc file for each of them or am I > > doing something terribly wrong? > > These benchmarks were compile a long time ago in a galaxy far far away. > In this galaxy, uninitialized globals (int X;) were handled

> What exactly does Olden require from 'as'. I can't build the entire 'as' > because there's no more disk space left. Olden itself doesn't need it. _Compiling_ Olden requires the assembler to turn llvm .s files into llvm .o files. Why not just use the bytecode files that Misha posted? -Chris -- http://llvm.cs.uiuc.edu/ http://www.nondot.org/~sabre/Projects/

What exactly does Olden require from 'as'. I can't build the entire 'as' because there's no more disk space left. Wojciech

The following should work: % cd llvm/test/Programs/MultiSource % gmake DISABLE_CBE=1 DISABLE_LLC=1 --Vikram http://www.cs.uiuc.edu/~vadve > -----Original Message----- > From: llvmdev-admin at cs.uiuc.edu > [mailto:llvmdev-admin at cs.uiuc.edu] On Behalf Of Juan Nicolas Ruiz > Sent: Sunday, December 08, 2002 2:20 PM > To: llvmdev at cs.uiuc.edu > Subject: [LLVMdev] Olden BM

Hi, Below you can see the updated benchmark results for the new SLP-vectorizer. As you can see, there is a small number of compile time regressions, a single major runtime *regression, and many performance gains. There is a tiny increase in code size: 30k for the whole test-suite. Based on the numbers below I would like to enable the SLP-vectorizer by default for -O3. Please let me know if you

On Jul 13, 2013, at 11:30 PM, Nadav Rotem <nrotem at apple.com> wrote: > Hi, > > LLVM’s SLP-vectorizer is a new pass that combines similar independent instructions in a straight-line code. It is currently not enabled by default, and people who want to experiment with it can use the clang command line flag “-fslp-vectorize”. I ran LLVM’s test suite with and without the SLP

Sorry for not posting sooner. I forgot to send an update with the results. I also have some benchmark data. It confirms much of what you posted -- binary size increase is essentially 0, performance increases across the board. It looks really good to me. However, there was one crash that I'd like to check if it still fires. Will update later today (feel free to ping me if you don't hear

Hi, Sorry for the delay in response. I measured the code size change and noticed small changes in both directions for individual programs. I found a 30k binary size growth for the entire testsuite + SPEC. I attached an updated performance report that includes both compile time and performance measurements. Thanks, Nadav On Jul 14, 2013, at 10:55 PM, Nadav Rotem <nrotem at apple.com>

Hi I'm trying to compile the "bh" C program from the Olden benchmark to one bc file. # compile source files into an LLVM bitcode file llvm-gcc -emit-llvm -c args.c -o args.bc -w -DTORONTO llvm-gcc -emit-llvm -c newbh.c -o newbh.bc -w -DTORONTO llvm-gcc -emit-llvm -c util.c -o util.bc -w -DTORONTO llvm-gcc -emit-llvm -c walksub.c -o walksub.bc -w -DTORONTO # To link files together

On Apr 16, 2012, at 3:30 PM, Chandler Carruth <chandlerc at google.com> wrote: > Does sinking these into the DAGCombine layer help? How much does it break? I tried disabling just the InstCombine transforms that hide shl instructions behind bitmasks. Even though DAGCombine has the same transforms, it causes some pretty bad regressions: External/SPEC/CINT95/147_vortex/147_vortex

On Jul 14, 2013, at 9:52 PM, Chris Lattner <clattner at apple.com> wrote: > > On Jul 13, 2013, at 11:30 PM, Nadav Rotem <nrotem at apple.com> wrote: > >> Hi, >> >> LLVM’s SLP-vectorizer is a new pass that combines similar independent instructions in a straight-line code. It is currently not enabled by default, and people who want to experiment with it

Cool! What changes have you seen to generated code size? I'll take it for a spin on our benchmarks. On Sat, Jul 13, 2013 at 11:30 PM, Nadav Rotem <nrotem at apple.com> wrote: > Hi, > > LLVM’s SLP-vectorizer is a new pass that combines similar independent > instructions in a straight-line code. It is currently not enabled by > default, and people who want to

Hi, LLVM’s SLP-vectorizer is a new pass that combines similar independent instructions in a straight-line code. It is currently not enabled by default, and people who want to experiment with it can use the clang command line flag “-fslp-vectorize”. I ran LLVM’s test suite with and without the SLP vectorizer on a Sandybridge mac (using SSE4, w/o AVX). Based on my performance measurements

Hello, I was interested in how much Type-Based Alias Analysis helps to optimize code. For that purpose, I've compared three sets of benchmarks: compiled without TBAA, compiled with a default TBAA metadata format, and compiled with new TBAA metadata format. As a set of benchmarks, I've used the LLVM test suite (http://llvm.org/docs/TestingGuide.html#test-suite-overview) which has a lot of