[llvm-testresults] bwilson__llvm-gcc_PROD__i386 nightly tester results

1 view
Skip to first unread message

llvm-tes...@cs.uiuc.edu

unread,
Oct 5, 2011, 3:41:34 AM10/5/11
to llvm-tes...@cs.uiuc.edu

bwilson__llvm-gcc_PROD__i386 nightly tester results

URLhttp://llvm.org/perf/db_default/simple/nts/325/
Nicknamebwilson__llvm-gcc_PROD__i386:4
Namecurlew.apple.com

Run ID Order Start Time End Time
Current325 02011-10-05 06:59:172011-10-05 07:42:09
Previous324 02011-10-04 06:59:202011-10-04 07:42:08

Tests Summary

Status Group#
Unchanged Tests961
Total Tests961

llvm-tes...@cs.uiuc.edu

unread,
Oct 6, 2011, 3:40:30 AM10/6/11
to llvm-tes...@cs.uiuc.edu

bwilson__llvm-gcc_PROD__i386 nightly tester results

URLhttp://llvm.org/perf/db_default/simple/nts/326/
Nicknamebwilson__llvm-gcc_PROD__i386:4
Namecurlew.apple.com

Run ID Order Start Time End Time

Current326 02011-10-06 06:58:542011-10-06 07:41:21
Previous

325 02011-10-05 06:59:172011-10-05 07:42:09

Tests Summary

Status Group#
Performance Regressions9
Performance Improvements11
Unchanged Tests941
Total Tests961

Changes Detail

Performance Regressions ΔPreviousCurrent σ
SingleSource/Benchmarks/Shootout/lists.exec2.51%5.88506.03270.0124
MultiSource/Applications/lambda-0_1_3/lambda.exec2.06%6.49056.62410.0009
External/SPEC/CINT2000/255_vortex/255_vortex.exec1.74%5.01365.10060.0046
MultiSource/Applications/sqlite3/sqlite3.compile1.72%14.400414.64740.0278
External/SPEC/CINT2000/164_gzip/164_gzip.exec1.50%13.435413.63740.0706
External/SPEC/CFP2000/177_mesa/177_mesa.compile1.34%14.419514.61260.0745
External/SPEC/CINT95/134_perl/134_perl.compile1.18%6.35116.42600.0068
External/SPEC/CINT2006/400_perlbench/400_perlbench.exec1.15%9.40499.51260.0133
MultiSource/Applications/JM/lencod/lencod.compile1.05%15.332215.49320.0164

Performance Improvements ΔPreviousCurrent σ
MultiSource/Benchmarks/Prolangs-C/cdecl/cdecl.compile-21.57%0.43300.33960.0010
SingleSource/Benchmarks/McGill/chomp.exec-12.39%1.06000.92870.0005
MultiSource/Benchmarks/Ptrdist/yacr2/yacr2.exec-7.56%0.90760.83900.0001
MultiSource/Applications/hexxagon/hexxagon.exec-7.41%14.519513.44330.0010
External/SPEC/CINT2000/175_vpr/175_vpr.exec-2.18%4.62124.52050.0005
SingleSource/Benchmarks/Misc-C++/stepanov_container.exec-1.90%3.94763.87250.0224
MultiSource/Applications/aha/aha.exec-1.48%3.59663.54350.0006
MultiSource/Applications/SPASS/SPASS.exec-1.42%9.07208.94300.0394
SingleSource/Benchmarks/Misc-C++-EH/spirit.compile-1.23%8.63578.52940.0290
External/SPEC/CFP2000/188_ammp/188_ammp.exec-1.21%17.669617.45520.0304
MultiSource/Benchmarks/tramp3d-v4/tramp3d-v4.compile-1.06%25.261124.99370.0950

llvm-tes...@cs.uiuc.edu

unread,
Oct 7, 2011, 3:39:22 AM10/7/11
to llvm-tes...@cs.uiuc.edu

bwilson__llvm-gcc_PROD__i386 nightly tester results

URLhttp://llvm.org/perf/db_default/simple/nts/327/
Nicknamebwilson__llvm-gcc_PROD__i386:4
Namecurlew.apple.com

Run ID Order Start Time End Time

Current327 02011-10-07 06:58:392011-10-07 07:41:06
Previous

326 02011-10-06 06:58:542011-10-06 07:41:21

Tests Summary

Status Group#
Performance Regressions2
Unchanged Tests959
Total Tests961

Changes Detail

Performance Regressions ΔPreviousCurrent σ

SingleSource/Benchmarks/Shootout-C++/methcall.exec3.36%7.46627.71680.0745
SingleSource/Benchmarks/Shootout/lists.exec2.85%6.03276.20490.0545

llvm-tes...@cs.uiuc.edu

unread,
Oct 8, 2011, 3:39:50 AM10/8/11
to llvm-tes...@cs.uiuc.edu

bwilson__llvm-gcc_PROD__i386 nightly tester results

URLhttp://llvm.org/perf/db_default/simple/nts/328/
Nicknamebwilson__llvm-gcc_PROD__i386:4
Namecurlew.apple.com

Run ID Order Start Time End Time

Current328 02011-10-08 06:59:192011-10-08 07:41:39
Previous

327 02011-10-07 06:58:392011-10-07 07:41:06

Tests Summary

Status Group#
Performance Regressions2

Performance Improvements1
Unchanged Tests958
Total Tests961

Changes Detail

Performance Regressions ΔPreviousCurrent σ

SingleSource/Benchmarks/Shootout-C++/ackermann.exec6.96%1.81271.93890.0014
External/Povray/povray.exec1.61%4.30964.37900.0095

Performance Improvements ΔPreviousCurrent σ

MultiSource/Benchmarks/Bullet/bullet.exec-3.89%6.45866.20760.0084

llvm-tes...@cs.uiuc.edu

unread,
Oct 9, 2011, 3:38:48 AM10/9/11
to llvm-tes...@cs.uiuc.edu

bwilson__llvm-gcc_PROD__i386 nightly tester results

URLhttp://llvm.org/perf/db_default/simple/nts/329/
Nicknamebwilson__llvm-gcc_PROD__i386:4
Namecurlew.apple.com

Run ID Order Start Time End Time

Current329 02011-10-09 06:58:152011-10-09 07:40:39
Previous

328 02011-10-08 06:59:192011-10-08 07:41:39

Tests Summary

Status Group#
Performance Regressions1
Performance Improvements1
Unchanged Tests959
Total Tests961

Changes Detail

Performance Regressions ΔPreviousCurrent σ

SingleSource/Benchmarks/Shootout/lists.exec6.48%6.15026.54850.1270

Performance Improvements ΔPreviousCurrent σ

External/SPEC/CFP2000/177_mesa/177_mesa.compile-2.43%14.727514.36920.1100

llvm-tes...@cs.uiuc.edu

unread,
Oct 10, 2011, 3:39:09 AM10/10/11
to llvm-tes...@cs.uiuc.edu

bwilson__llvm-gcc_PROD__i386 nightly tester results

URLhttp://llvm.org/perf/db_default/simple/nts/330/
Nicknamebwilson__llvm-gcc_PROD__i386:4
Namecurlew.apple.com

Run ID Order Start Time End Time

Current330 02011-10-10 06:58:302011-10-10 07:41:01
Previous

329 02011-10-09 06:58:152011-10-09 07:40:39

Tests Summary

Status Group#
Performance Improvements1
Unchanged Tests960
Total Tests961

Changes Detail

Performance Improvements ΔPreviousCurrent σ
SingleSource/Benchmarks/Shootout-C++/methcall.exec-4.08%7.69267.37890.0938

llvm-tes...@cs.uiuc.edu

unread,
Oct 11, 2011, 3:39:52 AM10/11/11
to llvm-tes...@cs.uiuc.edu

bwilson__llvm-gcc_PROD__i386 nightly tester results

URLhttp://llvm.org/perf/db_default/simple/nts/331/
Nicknamebwilson__llvm-gcc_PROD__i386:4
Namecurlew.apple.com

Run ID Order Start Time End Time

Current331 02011-10-11 06:59:182011-10-11 07:41:45
Previous

330 02011-10-10 06:58:302011-10-10 07:41:01

Tests Summary

Status Group#

llvm-tes...@cs.uiuc.edu

unread,
Oct 12, 2011, 3:40:12 AM10/12/11
to llvm-tes...@cs.uiuc.edu

bwilson__llvm-gcc_PROD__i386 nightly tester results

URLhttp://llvm.org/perf/db_default/simple/nts/332/
Nicknamebwilson__llvm-gcc_PROD__i386:4
Namecurlew.apple.com

Run ID Order Start Time End Time

Current332 02011-10-12 06:59:182011-10-12 07:42:03
Previous

331 02011-10-11 06:59:182011-10-11 07:41:45

Tests Summary

Status Group#
Performance Regressions17
Performance Improvements6
Added Tests2
Unchanged Tests938
Total Tests963

Changes Detail

Performance Regressions ΔPreviousCurrent σ

MultiSource/Applications/aha/aha.exec12.07%3.54243.96980.0002
MultiSource/Benchmarks/VersaBench/bmm/bmm.exec11.38%2.29702.55850.0007
External/SPEC/CFP2000/179_art/179_art.exec10.95%1.41901.57440.0013
MultiSource/Applications/JM/lencod/lencod.exec6.81%11.814912.61950.0267
External/SPEC/CINT2006/458_sjeng/458_sjeng.exec5.72%6.02636.37110.0076
MultiSource/Benchmarks/mafft/pairlocalalign.exec3.98%29.802230.98690.0810
MultiSource/Benchmarks/Fhourstones-3_1/fhourstones3_1.exec3.88%2.07292.15340.0010
External/SPEC/CINT2000/256_bzip2/256_bzip2.exec2.68%10.379210.65710.0138
External/SPEC/CINT2000/164_gzip/164_gzip.exec2.64%13.486613.84240.0737
External/SPEC/CINT2000/175_vpr/175_vpr.exec2.61%4.52204.63990.0007
External/SPEC/CINT2000/197_parser/197_parser.compile2.21%3.16923.23920.0075
MultiSource/Benchmarks/sim/sim.exec1.97%6.19976.32190.0004
MultiSource/Applications/sqlite3/sqlite3.exec1.30%4.66354.72410.0028
MultiSource/Benchmarks/Bullet/bullet.compile1.26%45.290745.86080.0695
External/SPEC/CINT2006/483_xalancbmk/483_xalancbmk.compile1.25%223.7703226.56410.1639
External/SPEC/CFP2006/450_soplex/450_soplex.compile1.04%30.593530.91240.0283
SingleSource/Benchmarks/CoyoteBench/almabench.exec1.00%12.054612.17540.0049

Performance Improvements ΔPreviousCurrent σ

SingleSource/Benchmarks/Misc/fbench.exec-8.84%2.08341.89920.0050
External/SPEC/CINT2006/400_perlbench/400_perlbench.exec-5.17%9.53759.04450.0241
External/SPEC/CINT2006/456_hmmer/456_hmmer.exec-2.78%5.36635.21690.0008
MultiSource/Applications/lua/lua.exec-2.50%28.877928.15500.0036
External/Povray/povray.exec-1.73%4.37984.30400.0271
MultiSource/Applications/lambda-0_1_3/lambda.exec-1.56%6.62496.52140.0019

Added Tests
SingleSource/Benchmarks/Stanford/FloatMM.compile
SingleSource/Benchmarks/Stanford/FloatMM.exec

Duncan Sands

unread,
Oct 12, 2011, 4:24:04 AM10/12/11
to llvm-tes...@cs.uiuc.edu, LLVMdev
Hi Bob, are these performance regressions real? They look pretty serious.

Ciao, Duncan.

On 10/12/11 09:40, llvm-tes...@cs.uiuc.edu wrote:
>
> bwilson__llvm-gcc_PROD__i386 nightly tester results
>
> URL http://llvm.org/perf/db_default/simple/nts/332/
> Nickname bwilson__llvm-gcc_PROD__i386:4
> Name curlew.apple.com
>
> Run ID Order Start Time End Time
> Current 332 0 2011-10-12 06:59:18 2011-10-12 07:42:03
> Previous 331 0 2011-10-11 06:59:18 2011-10-11 07:41:45
>
>

> Tests Summary
>
> Status Group #
> Performance Regressions 17
> Performance Improvements 6
> Added Tests 2
> Unchanged Tests 938

> *Total Tests* *963*

> _______________________________________________
> llvm-testresults mailing list
> llvm-tes...@cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-testresults

_______________________________________________
llvm-testresults mailing list
llvm-tes...@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-testresults

Bob Wilson

unread,
Oct 12, 2011, 12:03:40 PM10/12/11
to Duncan Sands, llvm-tes...@cs.uiuc.edu, LLVMdev
Yes, they are real. I re-ran the two tests with the biggest execution time regressions, and the results were completely reproducible.

llvm-tes...@cs.uiuc.edu

unread,
Oct 13, 2011, 3:40:21 AM10/13/11
to llvm-tes...@cs.uiuc.edu

bwilson__llvm-gcc_PROD__i386 nightly tester results

URLhttp://llvm.org/perf/db_default/simple/nts/333/
Nicknamebwilson__llvm-gcc_PROD__i386:4
Namecurlew.apple.com

Run ID Order Start Time End Time

Current333 02011-10-13 06:59:232011-10-13 07:42:15
Previous

332 02011-10-12 06:59:182011-10-12 07:42:03

Tests Summary

Status Group#
New Failures2
Performance Regressions5
Performance Improvements2
Unchanged Tests954
Total Tests963

Changes Detail

New Failures
External/SPEC/CINT2000/176_gcc/176_gcc.compile
External/SPEC/CINT2000/176_gcc/176_gcc.exec

Performance Regressions ΔPreviousCurrent σ

SingleSource/Benchmarks/McGill/chomp.exec14.19%0.92911.06090.0004
MultiSource/Benchmarks/Ptrdist/ks/ks.exec7.21%1.73871.86410.0139
MultiSource/Benchmarks/PAQ8p/paq8p.exec6.35%93.189899.11140.2146
MultiSource/Applications/hexxagon/hexxagon.exec3.50%13.315313.78180.0512
MultiSource/Applications/lambda-0_1_3/lambda.exec2.11%6.52146.65920.0412

Performance Improvements ΔPreviousCurrent σ

External/SPEC/CINT2000/255_vortex/255_vortex.exec-1.47%5.06244.98780.0038
External/SPEC/CFP2000/188_ammp/188_ammp.exec-1.06%17.527617.34190.0278

Evan Cheng

unread,
Oct 13, 2011, 2:30:02 PM10/13/11
to llvm-testresults
No one is paying much attention to i386 performance these days. We did not see the same kind of regressions on the Clang testers.

Evan

Added Tests
SingleSource/Benchmarks/Stanford/FloatMM.compile
SingleSource/Benchmarks/Stanford/FloatMM.exec

Evan Cheng

unread,
Oct 13, 2011, 2:48:43 PM10/13/11
to Bob Wilson, llvm-tes...@cs.uiuc.edu, LLVMdev
If I were to guess, I would say it's due to machine LICM changes. Although I can't understand why the regressions are not showing up on Clang testers.

Evan

> LLVM Developers mailing list
> LLV...@cs.uiuc.edu http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev

llvm-tes...@cs.uiuc.edu

unread,
Oct 14, 2011, 3:40:24 AM10/14/11
to llvm-tes...@cs.uiuc.edu

bwilson__llvm-gcc_PROD__i386 nightly tester results

URLhttp://llvm.org/perf/db_default/simple/nts/334/
Nicknamebwilson__llvm-gcc_PROD__i386:4
Namecurlew.apple.com

Run ID Order Start Time End Time

Current334 02011-10-14 06:59:322011-10-14 07:42:21
Previous

333 02011-10-13 06:59:232011-10-13 07:42:15

Tests Summary

Status Group#
New Passes2
Unchanged Tests961
Total Tests963

Changes Detail

New Passes
External/SPEC/CINT2000/176_gcc/176_gcc.compile
External/SPEC/CINT2000/176_gcc/176_gcc.exec

Reply all
Reply to author
Forward
0 new messages