perm filename STAK.TIM[TIM,LSP]12 blob
sn#759501 filedate 1984-06-26 generic text, type C, neo UTF8
COMMENT ⊗ VALID 00017 PAGES
C REC PAGE DESCRIPTION
C00001 00001
C00002 00002 (fasload stak)
C00005 00003 STAK
C00006 00004 NIL
C00008 00005 Franz
C00016 00006 SCORE Dec 11, 1983 interlisp
C00017 00007 PSL SCORE 1/10/84 - See later page
C00018 00008 DEC780CL
C00020 00009 InterLisp Vax 780
C00021 00010 PSL-20 3.3
C00022 00011 PSL-Cray 3.2
C00023 00012 PSL-750 3.2
C00024 00013 PSL-780 3.2
C00025 00014 PSL-DN600 3.2
C00026 00015 PSL-DN300 3.2
C00027 00016 S-1
C00028 00017 PERQ 6/18/84
C00029 ENDMK
C⊗;
(fasload stak)
(timit)
Timing performed on Saturday 06/25/83 at 21:44:55.
Cpu (- GC) Time = 3.499
Elapsed Time = 4.9
Wholine Time = 4.4166667
GC Time = 0.0
Load Average Before = 0.183229327
Load Average After = 0.213666797
Average Load Average = 0.198448062
NIL
(timit)
Timing performed on Saturday 06/25/83 at 21:45:40.
Cpu (- GC) Time = 3.498
Elapsed Time = 5.46666664
Wholine Time = 5.0
GC Time = 0.0
Load Average Before = 0.272845864
Load Average After = 0.304480433
Average Load Average = 0.28866315
NIL
(timit)
Timing performed on Saturday 06/25/83 at 21:46:02.
Cpu (- GC) Time = 3.498
Elapsed Time = 5.26666665
Wholine Time = 4.76666665
GC Time = 0.0
Load Average Before = 0.30024767
Load Average After = 0.329322577
Average Load Average = 0.314785123
NIL
NEW CPU
(fasload stak)
(timit)
Timing performed on Thursday 01/26/84 at 12:56:51.
Cpu (- GC) Time = 3.499
Elapsed Time = 7.31666666
Wholine Time = 4.65
GC Time = 0.0
Load Average Before = 0.47693181
Load Average After = 0.531651616
Average Load Average = 0.50429171
NIL
Timing performed on Thursday 01/26/84 at 12:57:05.
Cpu (- GC) Time = 3.499
Elapsed Time = 6.6
Wholine Time = 4.78333336
GC Time = 0.0
Load Average Before = 0.5377115
Load Average After = 0.57184744
Average Load Average = 0.55477947
NIL
;;; STAK
D2
7/5/83 without interrupts
Elapsed 12.4
CPU 12.4
D3
7/5/83
Elapsed 1.89
CPU 1.89
D1
1/25/84 with interrupts
Elapsed 4.66
CPU 4.66
;;; NIL
STAK
Fixnum-only arithmetic.
cpu=23.15,elapsed=23.24,pf=0
I've been going through the nil binding and value-cell code recently,
and have found one of the typical brainos of large systems. The
binding stuff seems to think that it is doing the world a favor by
saving and restoring all registers it uses. I don't think there is a
single piece of code which accepts this favor. (More typically, kernel
subroutines just "document" the registers they save, but i have found
this to be more of a pain than it is worth if i have to go and fix
something. I have reason to believe that this has hurt maclisp at
points, as i think that various functions/subroutines have to push/
pop registers which earlier versions of them did not use.)
;;; Franz
∂09-Oct-83 1349 jkf@ucbkim stak
Received: from UCBKIM by SU-AI with TCP/SMTP; 9 Oct 83 13:48:58 PDT
Received: by ucbkim.ARPA (4.6/4.2)
id AA20699; Sun, 9 Oct 83 13:52:26 PDT
Date: Sun, 9 Oct 83 13:52:26 PDT
From: John Foderaro (on an h19-u) <jkf@ucbkim>
Message-Id: <8310092052.AA20699@ucbkim.ARPA>
To: rpg@su-ai
Subject: stak
Cc:
--- Benchmark stak run on ucbkim at Sat Oct 8 02:36:17 PDT 1983 by jkf
--- cpu usage: 2:36am up 3:06, 0 users, load average: 1.22, 1.22, 1.17
Franz Lisp, Opus 38.81
=> [fasl stak.o]
t
=> benchmark: test (file stak) , tranlinks: on, localf: no
executing form: (tak 18 12 6)
begin (43 173)
end (422 173)
runs 1
avg cpu time 6.316666666666667
benchmark: test (file stak) , tranlinks: off, localf: no
executing form: (tak 18 12 6)
begin (424 173)
end (1489 173)
runs 1
avg cpu time 17.75
nil
=> Franz Lisp, Opus 38.81
=> [fasl stak-l.o]
t
=> benchmark: test (file stak) , tranlinks: on, localf: yes
executing form: (tak 18 12 6)
begin (42 174)
end (232 174)
runs 1
avg cpu time 3.166666666666667
benchmark: test (file stak) , tranlinks: off, localf: yes
executing form: (tak 18 12 6)
begin (233 174)
end (422 174)
runs 1
avg cpu time 3.15
nil
=> Franz Lisp, Opus 38.81
=> [load stak.l]
[fasl benchmac.o]
t
=> benchmark: test (file stak) , tranlinks: on, interpreted
executing form: (tak 18 12 6)
begin (58 175)
end (4744 175)
runs 1
avg cpu time 78.1
benchmark: test (file stak) , tranlinks: off, interpreted
executing form: (tak 18 12 6)
begin (4746 175)
end (9472 175)
runs 1
avg cpu time 78.76666666666667
nil
=> --- cpu usage: 2:39am up 3:10, 0 users, load average: 1.29, 1.18, 1.16
--- end of benchmark stak
--- Benchmark stak run on ucbmatisse at Fri Oct 7 13:01:04 PDT 1983 by jkf
--- cpu usage: 1:01pm up 4 days, 47 mins, 4 users, load average: 1.08, 1.15, 1.28
Franz Lisp, Opus 38.81
=> [fasl stak.o]
t
=> benchmark: test (file stak) , tranlinks: on, localf: no
executing form: (tak 18 12 6)
begin (123 202)
end (794 202)
runs 1
avg cpu time 11.18333333333333
benchmark: test (file stak) , tranlinks: off, localf: no
executing form: (tak 18 12 6)
begin (797 202)
end (2356 202)
runs 1
avg cpu time 25.98333333333333
nil
=> Franz Lisp, Opus 38.81
=> [fasl stak-l.o]
t
=> benchmark: test (file stak) , tranlinks: on, localf: yes
executing form: (tak 18 12 6)
begin (122 200)
end (425 200)
runs 1
avg cpu time 5.05
benchmark: test (file stak) , tranlinks: off, localf: yes
executing form: (tak 18 12 6)
begin (427 200)
end (728 200)
runs 1
avg cpu time 5.016666666666667
nil
=> Franz Lisp, Opus 38.81
=> [load stak.l]
[fasl benchmac.o]
t
=> benchmark: test (file stak) , tranlinks: on, interpreted
executing form: (tak 18 12 6)
begin (142 197)
end (8130 197)
runs 1
avg cpu time 133.1333333333333
benchmark: test (file stak) , tranlinks: off, interpreted
executing form: (tak 18 12 6)
begin (8134 197)
end (16133 197)
runs 1
avg cpu time 133.3166666666667
nil
=> --- cpu usage: 1:07pm up 4 days, 53 mins, 4 users, load average: 1.52, 1.25, 1.27
--- end of benchmark stak
--- Benchmark stak run on ucbmike at Fri Oct 7 14:23:04 PDT 1983 by jkf
--- cpu usage: 2:23pm up 4:54, 0 users, load average: 1.01, 1.03, 1.03
Franz Lisp, Opus 38.79
-> [fasl stak.o]
t
-> benchmark: test (file stak) , tranlinks: on, localf: no
executing form: (tak 18 12 6)
begin (90 324)
end (760 324)
runs 1
avg cpu time 11.16666666666667
benchmark: test (file stak) , tranlinks: off, localf: no
executing form: (tak 18 12 6)
begin (764 324)
end (2444 324)
runs 1
avg cpu time 28.0
nil
-> Franz Lisp, Opus 38.79
-> [fasl stak-l.o]
t
-> benchmark: test (file stak) , tranlinks: on, localf: yes
executing form: (tak 18 12 6)
begin (93 324)
end (634 324)
runs 1
avg cpu time 9.016666666666667
benchmark: test (file stak) , tranlinks: off, localf: yes
executing form: (tak 18 12 6)
begin (636 324)
end (1177 324)
runs 1
avg cpu time 9.016666666666667
nil
-> Franz Lisp, Opus 38.79
-> [load stak.l]
[fasl benchmac.o]
t
-> benchmark: test (file stak) , tranlinks: on, interpreted
executing form: (tak 18 12 6)
begin (112 327)
end (9447 327)
runs 1
avg cpu time 155.5833333333333
benchmark: test (file stak) , tranlinks: off, interpreted
executing form: (tak 18 12 6)
begin (9452 327)
end (18798 327)
runs 1
avg cpu time 155.7666666666667
nil
-> --- cpu usage: 2:30pm up 5:02, 0 users, load average: 1.05, 1.03, 1.03
--- end of benchmark stak
;;; SCORE Dec 11, 1983 interlisp
block compiled
6.379 seconds
0.0 gc
;;; PSL SCORE 1/10/84 - See later page
Cpu time: 12 ms
21 lisp> (tak 18 12 6)
7
Cpu time: 2757 ms
22 lisp> (tak 18 12 6)
7
Cpu time: 2750 ms
23 lisp> (tak 18 12 6)
7
Cpu time: 2795 ms
24 lisp> (tak 18 12 6)
7
Cpu time: 2798 ms
25 lisp> (tak 18 12 6)
7
Cpu time: 2787 ms
26 lisp> (tak 18 12 6)
7
Cpu time: 2815 ms
27 lisp> (tak 18 12 6)
7
Cpu time: 2741 ms
28 lisp>
;;; DEC780CL
cpu + probability x gc
STAK 9.9
*************************************************************************
Name of VAXLisp VAXLisp VAXLisp VAXLisp
application U1.0-20 U1.0-20 U1.0-20 U1.0-20
Benchmark VMS V3.5 VMS V3.5 VMS V3.5 VMS V3.6
730 (No FPA) 750 (No FPA) 780 785 (No FPA)
-------------------------------------------------------------------------------
-------------------------------------------------------------------------------
Stak 20.96 6.21 4.11 2.92
;;; InterLisp Vax 780
STAK:
/lisp/rbates/lisp/dick/STAK.V;1
←(TIME (TAK 18 12 6]
0 conses
10.288 seconds
7
;;; PSL-20 3.3
CTAK: Takai test using catch and throw
Timing performed on DEC-20
23-Mar-84 05:15:00 .
........................................
Cpu (- GC) Time = 2.971 secs
Elapsed Time = 3.0 secs
GC Time = 0.0 secs
Load Average Before = 1.2
Load Average After = 1.2
Average Load Average = 1.2
;;; PSL-Cray 3.2
;;; Times are in milliseconds
09:27:33 016:01.501 STAK: Takai test using fluid binding$2 ε
09:27:53 016:04.913 Cpu (- GC) Time = 1119.51800000 secs$2 ε
09:27:55 016:05.433 Elapsed Time = 0. secs
09:27:56 016:05.953 GC Time = 0. secs
09:27:58 016:06.473 Load Average Before = 0
09:28:00 016:06.994 Load Average After = 0
09:28:02 016:07.514 Average Load Average = 0.
;;; PSL-750 3.2
Cpu (- GC) Time = 17.782 secs
Elapsed Time = 0.0 secs
GC Time = 0.0 secs
Load Average Before = 0
Load Average After = 0
Average Load Average = 0.0
;;; PSL-780 3.2
Cpu (- GC) Time = 7.106 secs
Elapsed Time = 0.0 secs
GC Time = 0.0 secs
Load Average Before = 0
Load Average After = 0
Average Load Average = 0.0
;;; PSL-DN600 3.2
Cpu (- GC) Time = 18.683 secs
Elapsed Time = 0.0 secs
GC Time = 0.0 secs
Load Average Before = 0
Load Average After = 0
Average Load Average = 0.0
;;; PSL-DN300 3.2
Cpu (- GC) Time = 19.44 secs
Elapsed Time = 0.0 secs
GC Time = 0.0 secs
Load Average Before = 0
Load Average After = 0
Average Load Average = 0.0
;;; S-1
∂09-Jun-84 0114 EJG@S1-A.ARPA S-1 STAK time
Received: from S1-A.ARPA by SU-AI.ARPA with TCP; 9 Jun 84 01:14:00 PDT
Date: 09 Jun 84 0110 PDT
From: Erik Gilbert <EJG@S1-A.ARPA>
Subject: S-1 STAK time
To: rpg@SU-AI.ARPA
CC: PMF@S1-A.ARPA
So, I patched around the EXCH instruction in SPECIAL-LOOKUP and
(TAK 18 12 6) now gets the right answer. The official time
reported by EVALT is 4.31 seconds (also for (tester) 100 iteration
version).
Keep those cards and benchmarks coming in! (Which ones next, coach?)
;;; PERQ 6/18/84
20.82 seconds (stopwatch)