[1]> (load "report.lisp")
;; Loading file report.lisp ...
Benchmark                 Reference  Armed
-------------------------------------------------------------------------------------
COMPILER                 [      0.60]  -1.00
LOAD-FASL                [      0.29]  -1.00
SUM-PERMUTATIONS         [      1.46]   1.41
WALK-LIST/SEQ            [      0.01]  -1.00
WALK-LIST/MESS           [      0.07]  -1.00
BOYER                    [     20.26]   0.03
BROWSE                   [      0.70]   0.29
DDERIV                   [      0.82]   0.22
DERIV                    [      0.93]   0.20
DESTRUCTIVE              [      0.79]   0.34
DIV2-TEST-1              [      1.27]   0.17
DIV2-TEST-2              [      1.76]   0.14
FFT                      [      1.88]   0.14
FRPOLY/FIXNUM            [      1.90]   0.36
FRPOLY/BIGNUM            [      0.72]   0.33
FRPOLY/FLOAT             [      1.94]   0.25
PUZZLE                   [      4.44]   0.76
TAK                      [      2.28]   0.37
CTAK                     [      1.48]   1.58
TRTAK                    [      2.27]   0.34
TAKL                     [      2.46]   0.16
STAK                     [      1.60]   0.48
FPRINT/UGLY              [      0.60]   2.15
FPRINT/PRETTY            [      0.83]   8.97
TRAVERSE                 [      3.50]   0.32
TRIANGLE                 [      8.37]   0.45
RICHARDS                 [      4.79]   0.84
FACTORIAL                [      0.26]   0.63
FIB                      [      1.08]   0.17
FIB-RATIO                [      0.02]  10.05
ACKERMANN                [      8.35]   0.26
MANDELBROT/COMPLEX       [      0.47]   1.13
MANDELBROT/DFLOAT        [      0.39]   0.28
MRG32K3A                 [      5.76]   0.27
CRC40                    [     12.48]   0.78
BIGNUM/ELEM-100-1000     [      0.02]  10.50
BIGNUM/ELEM-1000-100     [      0.07]   3.30
BIGNUM/ELEM-10000-1      [      0.07]   3.68
BIGNUM/PARI-100-10       [      0.01]   4.71
BIGNUM/PARI-200-5        [      0.03]   2.92
PI-DECIMAL/SMALL         [      0.67]  21.21
PI-DECIMAL/BIG           [      0.48]  52.27
PI-ATAN                  [      0.65]   1.32
PI-RATIOS                [      0.31]   7.04
HASH-STRINGS             [      0.66]   0.65
HASH-INTEGERS            [      1.25]   0.19
SLURP-LINES              [      6.15]   1.87
BOEHM-GC                 [      5.91]   0.42
DEFLATE-FILE             [      1.52]   0.41
1D-ARRAYS                [      0.31]   0.25
2D-ARRAYS                [      9.01]   0.57
3D-ARRAYS                [     16.85]   1.10
BITVECTORS               [      2.06]   0.14
BENCH-STRINGS            [      7.57]   0.05
fill-strings/adjustable  [     11.91]   0.17
STRING-CONCAT            [     76.70]   0.36
SEARCH-SEQUENCE          [      3.71]   0.39
CLOS/defclass            [      0.27]   2.81
CLOS/defmethod           [      0.07]   8.00
CLOS/instantiate         [      1.63]   3.75
CLOS/simple-instantiate  [      1.03]   5.26
CLOS/methodcalls         [      1.93]   2.29
CLOS/method+after        [      1.29]   2.31
EQL-SPECIALIZED-FIB      [      0.69]  -1.00
Reference time in first column is in seconds; other columns are relative
Reference implementation: CLISP 2.47 (2008-10-23) (built on x86-7.fedora.phx.redhat.com)
Impl Armed: Armed Bear Common Lisp 0.13.0-dev
=== Test machine ===
   Machine-type: X86_64
   Machine-version: X86_64
 
Bye.
[root@titan cl-bench]#
----- Original Message -----
From: dmiles@users.sourceforge.net
To: armedbear-j-devel@lists.sourceforge.net
Cc: kino coursey
Sent: Thursday, December 04, 2008 10:39 AM
Subject: Re: Benchmark numbers for IKVM

Someone offline was asking about the interpretation of these numbers.
So I will unpack a couple results:
 
Benchmark                 Reference     CLISP  IKVM   JAVA6
BOYER                    [      1.25]  16.11   2.16   1.02
ABCL on jdk1.7  took 1.25 seconds to run.
CLISP  took 16.11 times jdk1.7 to run totaling 20.16 seconds - This says ABCL is 16 times faster than CLISP at BOYER
IKVM running ABCL took 2.16 times jdk1.7 which totals to 2.70 seconds
JAVA6 is ABCL on jdk1.6 which took 1.02 times jdk1.6 which totals 1.27 seconds.  Which says jdk1.6 is 2% slower than jdk1.7 at BOYER
 
Benchmark                 Reference     CLISP  IKVM   JAVA6
CLOS/instantiate         [      6.68]   0.24  41.29   1.07
ABCL on jdk1.7  took 6.68 seconds to run.
CLISP  took 0.24 times jdk1.7 to run totaling 1.60 seconds - This says CLISP is 4 times faster than ABCL at this test
IKVM running ABCL took 41.29 times jdk1.7 which totaled a long 275 seconds
JAVA6 is ABCL on jdk1.6 which took 1.07 times jdk1.6 Which says jdk1.6 is 7% slower than on jdk1.7
 
 
> I think I understand the clisp and abcl parts, but what are java17, ikvm and java6 ?

JAVA17 is ABCL on jdk1.7 the default on my system. 
JAVA6 is ABCL on jdk1.6  -- I would have called it JAVA16 The column heading was only going to show first five letters.
 
> I'm guessing it's no coincidence that the java6 column is always close to 1 ?

JAVA6 Is pretty much 1.0 +/- .10 due to the fact that there is not much difference in jdk1.7 and a very recent update of jdk1.6.
 
 
Here is a web formated view of the results
 
https://sourceforge.net/mailarchive/attachment.php?list_name=armedbear-j-devel&message_id=CFAC93BD509C41908F6399FD3E51B0F4%40lotus&counter=1