[ANN] Exupery 0.10 released, now on Windows and Linux

>
> Exupery 0.10 is now released. There are prebuilt VM's available for
> both Windows and Linux. This release now provides a measurable speed
> improvement for the compilerBenchmark macro benchmark due to work on
> dynamic primitive inlining.
>
> Instructions for installation and a link to a pre-built image is
> here:
>
>   http://minnow.cc.gatech.edu/squeak/3945
>
>
>     Benchmarks on my Athlon 64 3500+
>     =========================================================
>     arithmaticLoopBenchmark  1398 compiled   92 ratio: 15.196
>     bytecodeBenchmark        2134 compiled  469 ratio:  4.550
>     sendBenchmark            1580 compiled  697 ratio:  2.267
>     doLoopsBenchmark         1090 compiled  840 ratio:  1.298
>     largeExplorers            334 compiled  358 ratio:  0.933
>     compilerBenchmark         733 compiled  705 ratio:  1.040
>     Cumulative Time          4167 compiled 1448 ratio   2.878
>
>     1,067,222,511 bytecodes/sec; 16,716,421 sends/sec
>
>     Benchmarks on Andy's Mobile Pentium 3
>     =========================================================
>     arithmaticLoopBenchmark  2487 compiled  285 ratio:  8.726
>     bytecodeBenchmark        4271 compiled 1255 ratio:  3.403
>     sendBenchmark            3482 compiled 1772 ratio:  1.965
>     doLoopsBenchmark         2078 compiled 1663 ratio:  1.250
>     largeExplorers           2224 compiled 1683 ratio:  1.321
>     compilerBenchmark        2093 compiled 1712 ratio:  1.223
>     Cumulative Time         12903 compiled 4971 ratio   2.596
>
>     Benchmarks from my Pentium-M laptop
>     =========================================================
>     arithmaticLoopBenchmark 1003 compiled  191 ratio:   5.251
>     bytecodeBenchmark       1773 compiled  683 ratio:   2.596
>     sendBenchmark           1446 compiled  922 ratio:   1.568
>     doLoopsBenchmark         991 compiled  918 ratio:   1.080
>     largeExplorers           418 compiled  441 ratio:   0.948
>     compilerBenchmark        718 compiled  683 ratio:   1.051
>     Cumulative Time         3773 compiled 2015 ratio    1.872
>
> It's interesting that on Andy's machine Exupery is providing a nice
> performance improvement for largeExplorers while on my machine there
> is a 7% performance loss. The loss is due to the interpreter inlining
> Point>>@ into the main interpreter loop while Exupery executes it as
> a normal primitive. Andy's benchmarks are promising enough for a 1.0,
> pity relative performance isn't so high on the other two machines.
>
> There is a mailing list for those interested in the project here:
>
>   http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/exupery
>
> Many thanks to Andy Tween for doing the Windows port and building
> the official Windows VM. Thanks also to Patrick Mauritz for doing
> a Solaris x86 port which was the first OS port.
>
> Bryce
> _______________________________________________
> Exupery mailing list
> [hidden email]
> http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/exupery
>

> First of all - let me thank you to the work you have done, Bruce!
>
> I have tested Exupery with provided image on 3 different machines. And
> here are the numbers:
>
> 1) Intel Pentium 4 Mobile 1.8 GHz (Win VM) - look at doLoopsBenchmark results!
>
> arithmaticLoopBenchmark 2097 compiled 158 ratio: 13.272
> bytecodeBenchmark 2958 compiled 619 ratio: 4.778
> sendBenchmark 2718 compiled 1874 ratio: 1.449
> doLoopsBenchmark 1647 compiled 2965 ratio: 0.555
> largeExplorers 1156 compiled 1166 ratio: 0.991
> compilerBenchmark 1082 compiled 1085 ratio: 0.997
> Cumulative Time 8093.659 compiled 3693.635 ratio 2.190
>
> 2) AMD Sempron 3100+ 1.8GHz (Win VM) - this one was born to run Exupery :)
>
> arithmaticLoopBenchmark 1708 compiled 113 ratio: 15.115
> bytecodeBenchmark 2505 compiled 545 ratio: 4.596
> sendBenchmark 1871 compiled 841 ratio: 2.225
> doLoopsBenchmark 1203 compiled 809 ratio: 1.487
> largeExplorers 704 compiled 458 ratio: 1.536
> compilerBenchmark 701 compiled 774 ratio: 0.906
> Cumulative Time 5437.323 compiled 1715.399 ratio 3.169
>
> 3) Intel Pentium Celeron 2000+
>
> arithmaticLoopBenchmark 1862 compiled 125 ratio: 14.896
> bytecodeBenchmark 3115 compiled 732 ratio: 4.254
> sendBenchmark 2130 compiled 973 ratio: 2.189
> doLoopsBenchmark 1453 compiled 940 ratio: 1.546
> largeExplorers 872 compiled 832 ratio: 1.048
> compilerBenchmark 1078 compiled 1043 ratio: 1.034
> Cumulative Time 7005.573 compiled 2356.208 ratio 2.972
>
> Note: .tz extension means tar gzip format? If yes then having .tgz or
> .tar.gz extension allows easier extraction in more tools (Total
> Commander doesn't understand .tz, in WinZip and IZArc I had to rename
> ungzipped file, etc.)
>
> Bye!
>
> Jakub
>
> On 11/15/06, [hidden email] <[hidden email]> wrote:
> >
> > Exupery 0.10 is now released. There are prebuilt VM's available for
> > both Windows and Linux. This release now provides a measurable speed
> > improvement for the compilerBenchmark macro benchmark due to work on
> > dynamic primitive inlining.
> >
> > Instructions for installation and a link to a pre-built image is
> > here:
> >
> >   http://minnow.cc.gatech.edu/squeak/3945
> >
> >
> >     Benchmarks on my Athlon 64 3500+
> >     =========================================================
> >     arithmaticLoopBenchmark  1398 compiled   92 ratio: 15.196
> >     bytecodeBenchmark        2134 compiled  469 ratio:  4.550
> >     sendBenchmark            1580 compiled  697 ratio:  2.267
> >     doLoopsBenchmark         1090 compiled  840 ratio:  1.298
> >     largeExplorers            334 compiled  358 ratio:  0.933
> >     compilerBenchmark         733 compiled  705 ratio:  1.040
> >     Cumulative Time          4167 compiled 1448 ratio   2.878
> >
> >     1,067,222,511 bytecodes/sec; 16,716,421 sends/sec
> >
> >     Benchmarks on Andy's Mobile Pentium 3
> >     =========================================================
> >     arithmaticLoopBenchmark  2487 compiled  285 ratio:  8.726
> >     bytecodeBenchmark        4271 compiled 1255 ratio:  3.403
> >     sendBenchmark            3482 compiled 1772 ratio:  1.965
> >     doLoopsBenchmark         2078 compiled 1663 ratio:  1.250
> >     largeExplorers           2224 compiled 1683 ratio:  1.321
> >     compilerBenchmark        2093 compiled 1712 ratio:  1.223
> >     Cumulative Time         12903 compiled 4971 ratio   2.596
> >
> >     Benchmarks from my Pentium-M laptop
> >     =========================================================
> >     arithmaticLoopBenchmark 1003 compiled  191 ratio:   5.251
> >     bytecodeBenchmark       1773 compiled  683 ratio:   2.596
> >     sendBenchmark           1446 compiled  922 ratio:   1.568
> >     doLoopsBenchmark         991 compiled  918 ratio:   1.080
> >     largeExplorers           418 compiled  441 ratio:   0.948
> >     compilerBenchmark        718 compiled  683 ratio:   1.051
> >     Cumulative Time         3773 compiled 2015 ratio    1.872
> >
> > It's interesting that on Andy's machine Exupery is providing a nice
> > performance improvement for largeExplorers while on my machine there
> > is a 7% performance loss. The loss is due to the interpreter inlining
> > Point>>@ into the main interpreter loop while Exupery executes it as
> > a normal primitive. Andy's benchmarks are promising enough for a 1.0,
> > pity relative performance isn't so high on the other two machines.
> >
> > There is a mailing list for those interested in the project here:
> >
> >   http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/exupery
> >
> > Many thanks to Andy Tween for doing the Windows port and building
> > the official Windows VM. Thanks also to Patrick Mauritz for doing
> > a Solaris x86 port which was the first OS port.
> >
> > Bryce
> > _______________________________________________
> > Exupery mailing list
> > [hidden email]
> > http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/exupery
> >
>