[Openmcl-devel] cl-bench Windows 32-bit performance

Daniel Weinreb dlw at google.com
Fri Jun 17 03:53:11 PDT 2011


Sorry to reply so long after the original mail.

If you are looking at these figures, please be very careful
to understand what each benchmark measures.  These
are microbenchmarks, some of which exercise only
particular features.  For example, ctak is a program
that spends all its time in catch/throw, which is never,
in practice, a performance bottleneck.  Others are
written in a style that is rarely used any more in Lisp,
doing things old ways.  And whatever you do,
DON'T compute a mean or median; the benchmarks
are not comparable.  Just general advice about
interpretation.

-- Dan

On Wed, Apr 13, 2011 at 9:25 PM, Brandon Van Every <bvanevery at gmail.com>wrote:

> http://trac.clozure.com/ccl/wiki/HowFastAreWe presents some results
> using cl-bench.  How to obtain and use cl-bench is documented at
> http://www.cliki.net/Performance_Benchmarks .  I added a section on
> how to run the benchmarks using a Windows .bat file.  Here are my
> initial results.  I'm running Windows Vista on a Dell Inspiron 1720
> laptop with a 2GHz Intel Core 2 Duo processor and 2GB RAM.  I left
> system tray items, Firefox, and a few text files running in the
> background, but I did not bring any of those things to the foreground
> while running the tests.  This is a machine with a "typical background
> load" on it; it has not been specially configured to maximize
> benchmarking results.
>
> The raw hardware specs of my machine are roughly comparable to
> "MacBook Pro, Intel Core 2 Duo @2.33 GHz" machine on the Cliki page.
> The reference implementation for that benchmark is "SBCL 1.0.2 (binary
> download)".  Roughly eyeballing it, I find my "Clozure Common Lisp
> Version 1.7-dev-r14645M-trunk  (WindowsX8632)" benchmark is 2x..5x
> slower in many cases.  It's also slower than the "OpenMCL Version
> 1.1-pre-061231 (DarwinX8664)" maybe by 1.5x..4x depending.  The
> comparison math is difficult as I have to mentally renormalize the
> numbers; I will try to work up a comparison to SBCL.  SBCL's support
> of Windows is in a seemingly endless Alpha, so I'm not sure if I can
> actually bench it, but I'll try.
>
> I am wondering whether the 32-bit version is significantly slower than
> the 64-bit version.  Hmm, another possibility is I am using the "dev"
> version.
>
>
> Cheers,
> Brandon Van Every
>
>
> ;; -*- lisp -*-  Clozure Common Lisp Version 1.7-dev-r14645M-trunk
> (WindowsX8632)
> ;;
> ;; Implementation *features*:
> ;; (:PRIMARY-CLASSES :COMMON-LISP :OPENMCL :CCL :CCL-1.2 :CCL-1.3
> :CCL-1.4 :CCL-1.5 :CCL-1.6 :CLOZURE :CLOZURE-COMMON-LISP :ANSI-CL
> :OPENMCL-UNICODE-STRINGS :OPENMCL-NATIVE-THREADS :OPENMCL-PARTIAL-MOP
> :MCL-COMMON-MOP-SUBSET :OPENMCL-MOP-2 :OPENMCL-PRIVATE-HASH-TABLES
> :X8632-TARGET :X8632-HOST :X86 :X86-TARGET :X86-HOST :WINDOWS-HOST
> :WINDOWS-TARGET :WIN32-TARGET :WIN32-HOST :32-BIT-TARGET :32-BIT-HOST
> :WINDOWS :LITTLE-ENDIAN-TARGET :LITTLE-ENDIAN-HOST)
> ;;
> ;; Function                      real     user     sys       consed
> ;; ----------------------------------------------------------------
> ;; COMPILER                      3.05     2.98     0.00            0
> ;; LOAD-FASL                     0.34     0.28     0.00            0
> ;; SUM-PERMUTATIONS              5.71     5.43     0.00            0
> ;; WALK-LIST/SEQ                 0.06     0.06     0.00            0
> ;; WALK-LIST/MESS                0.07     0.05     0.00            0
> ;; BOYER                        13.31    13.09     0.00            0
> ;; BROWSE                        0.81     0.73     0.00            0
> ;; DDERIV                        0.90     0.95     0.00            0
> ;; DERIV                         0.97     0.90     0.00            0
> ;; DESTRUCTIVE                   0.67     0.64     0.00            0
> ;; DIV2-TEST-1                   1.39     1.23     0.00            0
> ;; DIV2-TEST-2                   1.58     1.48     0.00            0
> ;; FFT                           0.29     0.23     0.00            0
> ;; FRPOLY/FIXNUM                 1.03     1.03     0.00            0
> ;; FRPOLY/BIGNUM                 1.21     1.19     0.00            0
> ;; FRPOLY/FLOAT                  1.98     1.86     0.00            0
> ;; PUZZLE                        1.92     1.81     0.00            0
> ;; TAK                           0.82     0.87     0.00            0
> ;; CTAK                          1.35     1.50     0.00            0
> ;; TRTAK                         0.83     0.83     0.00            0
> ;; TAKL                          1.57     1.58     0.00            0
> ;; STAK                          1.60     1.73     0.00            0
> ;; FPRINT/UGLY                  10.48    10.62     0.00            0
> ;; FPRINT/PRETTY                 7.12     7.43     0.00            0
> ;; TRAVERSE                      1.92     1.75     0.00            0
> ;; TRIANGLE                      2.05     2.01     0.00            0
> ;; RICHARDS                      4.51     4.46     0.00            0
> ;; FACTORIAL                     1.44     1.42     0.00            0
> ;; FIB                           0.31     0.31     0.00            0
> ;; FIB-RATIO                     0.14     0.14     0.00            0
> ;; ACKERMANN                     2.62     2.73     0.00            0
> ;; MANDELBROT/COMPLEX            0.72     0.69     0.00            0
> ;; MANDELBROT/DFLOAT             0.35     0.39     0.00            0
> ;; MRG32K3A                     14.75    14.71     0.00            0
> ;; CRC40                        44.38    44.15     0.00            0
> ;; BIGNUM/ELEM-100-1000          2.26     2.31     0.00            0
> ;; BIGNUM/ELEM-1000-100         16.50    16.16     0.00            0
> ;; BIGNUM/ELEM-10000-1          18.34    17.83     0.00            0
> ;; BIGNUM/PARI-100-10            8.12     8.10     0.00            0
> ;; BIGNUM/PARI-200-5           126.43   125.46     0.00            0
> ;; PI-DECIMAL/SMALL            106.09   106.36     0.00            0
> ;; PI-DECIMAL/BIG              414.84   415.35     0.00            0
> ;; PI-ATAN                       7.38     7.11     0.00            0
> ;; PI-RATIOS                    13.75    13.91     0.00            0
> ;; HASH-STRINGS                 11.31    11.05     0.00            0
> ;; HASH-INTEGERS                 4.51     4.34     0.00            0
> ;; SLURP-LINES                   0.01     0.02     0.00            0
> ;; BOEHM-GC                     10.92    10.64     0.00            0
> ;; DEFLATE-FILE                  1.27     1.31     0.00            0
> ;; 1D-ARRAYS                     0.07     0.08     0.00            0
> ;; 2D-ARRAYS                     4.92     4.91     0.00            0
> ;; 3D-ARRAYS                    10.62    10.34     0.00            0
> ;; BITVECTORS                    1.84     1.75     0.00            0
> ;; BENCH-STRINGS                 5.99     5.91     0.00            0
> ;; fill-strings/adjustable      43.07    42.73     0.00            0
> ;; STRING-CONCAT               121.96   119.59     0.00            0
> ;; SEARCH-SEQUENCE               5.07     5.18     0.00            0
> ;; CLOS/defclass                 1.00     0.64     0.00            0
> ;; CLOS/defmethod                0.29     0.34     0.00            0
> ;; CLOS/instantiate              6.38     6.47     0.00            0
> ;; CLOS/simple-instantiate      11.40    11.36     0.00            0
> ;; CLOS/methodcalls              6.22     5.99     0.00            0
> ;; CLOS/method+after             3.45     3.48     0.00            0
> ;; CLOS/complex-methods          1.54     1.43     0.00            0
> ;; EQL-SPECIALIZED-FIB           1.57     1.62     0.00            0
>
> ("Clozure Common Lisp Version 1.7-dev-r14645M-trunk  (WindowsX8632)"
> ("EQL-SPECIALIZED-FIB" 157/100 1623/1000 0 0) ("CLOS/complex-methods"
> 309/200 287/200 0 0) ("CLOS/method+after" 1723/500 3479/1000 0 0)
> ("CLOS/methodcalls" 6223/1000 5991/1000 0 0)
> ("CLOS/simple-instantiate" 2849/250 11357/1000 0 0)
> ("CLOS/instantiate" 6381/1000 6473/1000 0 0) ("CLOS/defmethod" 73/250
> 343/1000 0 0) ("CLOS/defclass" 997/1000 16/25 0 0) ("SEARCH-SEQUENCE"
> 1013/200 259/50 0 0) ("STRING-CONCAT" 121957/1000 119591/1000 0 0)
> ("fill-strings/adjustable" 43071/1000 5341/125 0 0) ("BENCH-STRINGS"
> 599/100 739/125 0 0) ("BITVECTORS" 461/250 1747/1000 0 0) ("3D-ARRAYS"
> 531/50 5171/500 0 0) ("2D-ARRAYS" 4923/1000 2457/500 0 0) ("1D-ARRAYS"
> 69/1000 39/500 0 0) ("DEFLATE-FILE" 637/500 131/100 0 0) ("BOEHM-GC"
> 5461/500 266/25 0 0) ("SLURP-LINES" 11/1000 2/125 0 0)
> ("HASH-INTEGERS" 4511/1000 542/125 0 0) ("HASH-STRINGS" 1131/100
> 2209/200 0 0) ("PI-RATIOS" 6873/500 2783/200 0 0) ("PI-ATAN" 7383/1000
> 7113/1000 0 0) ("PI-DECIMAL/BIG" 10371/25 415353/1000 0 0)
> ("PI-DECIMAL/SMALL" 53047/500 106361/1000 0 0) ("BIGNUM/PARI-200-5"
> 15804/125 15682/125 0 0) ("BIGNUM/PARI-100-10" 8121/1000 1012/125 0 0)
> ("BIGNUM/ELEM-10000-1" 917/50 17831/1000 0 0) ("BIGNUM/ELEM-1000-100"
> 2062/125 8081/500 0 0) ("BIGNUM/ELEM-100-1000" 451/200 2309/1000 0 0)
> ("CRC40" 5548/125 44147/1000 0 0) ("MRG32K3A" 7377/500 1471/100 0 0)
> ("MANDELBROT/DFLOAT" 173/500 39/100 0 0) ("MANDELBROT/COMPLEX" 361/500
> 86/125 0 0) ("ACKERMANN" 327/125 273/100 0 0) ("FIB-RATIO" 27/200 7/50
> 0 0) ("FIB" 313/1000 39/125 0 0) ("FACTORIAL" 359/250 71/50 0 0)
> ("RICHARDS" 4507/1000 2231/500 0 0) ("TRIANGLE" 2049/1000 503/250 0 0)
> ("TRAVERSE" 1923/1000 1747/1000 0 0) ("FPRINT/PRETTY" 7119/1000 297/40
> 0 0) ("FPRINT/UGLY" 2621/250 85/8 0 0) ("STAK" 8/5 433/250 0 0)
> ("TAKL" 157/100 63/40 0 0) ("TRTAK" 827/1000 827/1000 0 0) ("CTAK"
> 673/500 1497/1000 0 0) ("TAK" 41/50 873/1000 0 0) ("PUZZLE" 1923/1000
> 181/100 0 0) ("FRPOLY/FLOAT" 1983/1000 232/125 0 0) ("FRPOLY/BIGNUM"
> 121/100 593/500 0 0) ("FRPOLY/FIXNUM" 103/100 103/100 0 0) ("FFT"
> 291/1000 117/500 0 0) ("DIV2-TEST-2" 1577/1000 741/500 0 0)
> ("DIV2-TEST-1" 139/100 1231/1000 0 0) ("DESTRUCTIVE" 167/250 16/25 0
> 0) ("DERIV" 487/500 113/125 0 0) ("DDERIV" 449/500 119/125 0 0)
> ("BROWSE" 81/100 733/1000 0 0) ("BOYER" 2663/200 13089/1000 0 0)
> ("WALK-LIST/MESS" 9/125 47/1000 0 0) ("WALK-LIST/SEQ" 59/1000 31/500 0
> 0) ("SUM-PERMUTATIONS" 5707/1000 543/100 0 0) ("LOAD-FASL" 337/1000
> 7/25 0 0) ("COMPILER" 1523/500 2979/1000 0 0))
> _______________________________________________
> Openmcl-devel mailing list
> Openmcl-devel at clozure.com
> http://clozure.com/mailman/listinfo/openmcl-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.clozure.com/pipermail/openmcl-devel/attachments/20110617/dee2a775/attachment.htm>


More information about the Openmcl-devel mailing list