[llvm-testresults] Grawp nightly tester results

Evan Cheng evan.cheng at apple.com
Fri Sep 4 18:28:55 PDT 2009


On Sep 4, 2009, at 6:14 PM, Daniel Dunbar wrote:

> So, ditching 16-bit is a win?

I can't tell from the results. enc-pc1 got a lot faster, but mason and  
Dhrystone got slower. I'll leave it to Dan to make the full analysis.

>
> And do we know whats up with the JIT regressions? Evan, is this the
> first run with the changes to the compilation callback?

What JIT regressions? All of the following failures are crashing in  
the loop unswitching pass. Devang is looking into it.

Evan


>
> - Daniel
>
>>
>> New Test Failures:
>> Externals/SPEC/CFP2006/470_lbm/470_lbm [JIT, CBE, JIT codegen, LLC,  
>> LLC-BETA, LLC compile, Bitcode, GCCAS, LLC_BETA compile]
>> Externals/SPEC/CINT2006/403_gcc/403_gcc [JIT, JIT codegen, LLC, LLC- 
>> BETA, LLC compile, Bitcode, GCCAS, LLC_BETA compile]
>> Externals/SPEC/CINT2006/445_gobmk/445_gobmk [JIT, CBE, JIT codegen,  
>> LLC, LLC-BETA, LLC compile, Bitcode, GCCAS, LLC_BETA compile]
>> MultiSource/Applications/JM/ldecod/ldecod [JIT, CBE, JIT codegen,  
>> LLC, LLC-BETA, LLC compile, Bitcode, GCCAS, LLC_BETA compile]
>> MultiSource/Benchmarks/Prolangs-C/unix-tbl/unix-tbl [JIT, CBE, JIT  
>> codegen, LLC, LLC-BETA, LLC compile, Bitcode, GCCAS, LLC_BETA  
>> compile]
>>
>> Added Tests:
>>
>> Removed Tests:
>>
>> Significant Changes in Test Results:
>> LLC-BETA compile:
>>  Externals/SPEC/CFP2000/177_mesa/177_mesa: -32.20% (7.5199 => 5.0983)
>>  Externals/SPEC/CFP2000/188_ammp/188_ammp: -9.90% (1.5340 => 1.3821)
>>  Externals/SPEC/CFP2006/433_milc/433_milc: -10.86% (0.8879 => 0.7915)
>>  Externals/SPEC/CINT2006/464_h264ref/464_h264ref: -6.50% (5.9258 =>  
>> 5.5404)
>>  MultiSource/Applications/JM/lencod/lencod: -11.97% (7.7828 =>  
>> 6.8510)
>>  MultiSource/Applications/SPASS/SPASS: -12.42% (6.5489 => 5.7352)
>>  MultiSource/Applications/hbd/hbd: -9.01% (0.4473 => 0.4070)
>>  MultiSource/Benchmarks/FreeBench/pifft/pifft: -37.28% (0.5788 =>  
>> 0.3630)
>>  MultiSource/Benchmarks/MiBench/consumer-jpeg/consumer-jpeg: -7.48%  
>> (0.9996 => 0.9248)
>>  MultiSource/Benchmarks/MiBench/consumer-lame/consumer-lame:  
>> -15.16% (1.4610 => 1.2395)
>>  MultiSource/Benchmarks/MiBench/consumer-typeset/consumer-typeset:  
>> -14.09% (7.1303 => 6.1255)
>>  MultiSource/Benchmarks/MiBench/telecomm-gsm/telecomm-gsm: -5.30%  
>> (0.3097 => 0.2933)
>>  MultiSource/Benchmarks/PAQ8p/paq8p: -71.58% (3.3815 => 0.9611)
>>  MultiSource/Benchmarks/Prolangs-C/gnugo/gnugo: -56.70% (0.4649 =>  
>> 0.2013)
>>  MultiSource/Benchmarks/VersaBench/dbms/dbms: -12.53% (0.2346 =>  
>> 0.2052)
>>  MultiSource/Benchmarks/mafft/pairlocalalign: -11.51% (1.9353 =>  
>> 1.7125)
>>  MultiSource/Benchmarks/mediabench/gsm/toast/toast: -7.11% (0.3093  
>> => 0.2873)
>>  MultiSource/Benchmarks/mediabench/jpeg/jpeg-6a/cjpeg: -6.40%  
>> (0.8172 => 0.7649)
>>  MultiSource/Benchmarks/tramp3d-v4/tramp3d-v4: -8.14% (5.2939 =>  
>> 4.8631)
>> JIT:
>>  Externals/SPEC/CINT2000/176_gcc/176_gcc: 19.98% (17.1200 => 20.5400)
>>  Externals/SPEC/CINT2000/252_eon/252_eon: 8.06% (2.1100 => 2.2800)
>>  Externals/SPEC/CINT2000/253_perlbmk/253_perlbmk: 55.43% (12.5200  
>> => 19.4600)
>>  Externals/SPEC/CINT2000/300_twolf/300_twolf: 5.48% (6.0200 =>  
>> 6.3500)
>>  Externals/SPEC/CINT2006/400_perlbench/400_perlbench: 76.58%  
>> (17.1200 => 30.2300)
>>  Externals/SPEC/CINT2006/401_bzip2/401_bzip2: 5.18% (3.8600 =>  
>> 4.0600)
>>  Externals/SPEC/CINT2006/464_h264ref/464_h264ref: 5.05% (26.1600 =>  
>> 27.4800)
>>  MultiSource/Applications/ClamAV/clamscan: 10.80% (4.2600 => 4.7200)
>>  MultiSource/Applications/JM/lencod/lencod: 5.57% (14.9100 =>  
>> 15.7400)
>>  MultiSource/Applications/SPASS/SPASS: 58.90% (12.3600 => 19.6400)
>>  MultiSource/Applications/d/make_dparser: 21.09% (1.2800 => 1.5500)
>>  MultiSource/Applications/hbd/hbd: 7.89% (0.3800 => 0.4100)
>>  MultiSource/Applications/lemon/lemon: 27.06% (153.3700 => 194.8700)
>>  MultiSource/Applications/oggenc/oggenc: 9.93% (1.4100 => 1.5500)
>>  MultiSource/Applications/treecc/treecc: 11.63% (0.4300 => 0.4800)
>>  MultiSource/Benchmarks/ASCI_Purple/SMG2000/smg2000: 7.53% (7.9700  
>> => 8.5700)
>>  MultiSource/Benchmarks/MallocBench/espresso/espresso: 5.96%  
>> (2.3500 => 2.4900)
>>  MultiSource/Benchmarks/MiBench/consumer-jpeg/consumer-jpeg: 5.56%  
>> (0.5400 => 0.5700)
>>  MultiSource/Benchmarks/MiBench/consumer-lame/consumer-lame: 12.06%  
>> (1.4100 => 1.5800)
>>  MultiSource/Benchmarks/MiBench/consumer-typeset/consumer-typeset:  
>> 16.46% (5.6500 => 6.5800)
>>  MultiSource/Benchmarks/Prolangs-C/TimberWolfMC/timberwolfmc:  
>> 27.66% (0.4700 => 0.6000)
>>  MultiSource/Benchmarks/Prolangs-C/assembler/assembler: 25.00%  
>> (0.2800 => 0.3500)
>>  MultiSource/Benchmarks/Prolangs-C/bison/mybison: 35.06% (0.7700 =>  
>> 1.0400)
>>  MultiSource/Benchmarks/Ptrdist/bc/bc: 6.10% (1.6400 => 1.7400)
>>  MultiSource/Benchmarks/Ptrdist/yacr2/yacr2: 11.84% (1.5200 =>  
>> 1.7000)
>>  MultiSource/Benchmarks/mediabench/gsm/toast/toast: 12.00% (0.2500  
>> => 0.2800)
>>  MultiSource/Benchmarks/mediabench/mpeg2/mpeg2dec/mpeg2decode:  
>> 8.51% (0.4700 => 0.5100)
>>  SingleSource/Benchmarks/Adobe-C++/simple_types_constant_folding:  
>> 6.53% (2.9100 => 3.1000)
>> LLC compile:
>>  MultiSource/Benchmarks/MiBench/telecomm-gsm/telecomm-gsm: -5.16%  
>> (0.2943 => 0.2791)
>> LLC-BETA:
>>  MultiSource/Benchmarks/FreeBench/mason/mason: 13.64% (0.2200 =>  
>> 0.2500)
>>  MultiSource/Benchmarks/Trimaran/enc-pc1/enc-pc1: -48.19% (0.8300  
>> => 0.4300)
>>  SingleSource/Benchmarks/Dhrystone/dry: 22.22% (0.4500 => 0.5500)
>>  SingleSource/Benchmarks/Misc/fbench: -9.92% (2.4200 => 2.1800)
>>  SingleSource/UnitTests/Vector/SSE/sse_stepfft: -5.26% (0.5700 =>  
>> 0.5400)
>> CBE:
>>  SingleSource/Benchmarks/Dhrystone/fldry: 6.78% (1.1800 => 1.2600)
>> _______________________________________________
>> llvm-testresults mailing list
>> llvm-testresults at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-testresults
>>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-testresults/attachments/20090904/98b02c8c/attachment.html>


More information about the llvm-testresults mailing list