Benchmarking newLISP

Machine-specific discussion
Unix, Linux, OS X, OS/2, Windows, ..?
cormullion
Posts: 2038
Joined: Tue Nov 29, 2005 8:28 pm
Location: latiitude 50N longitude 3W
Contact:

Benchmarking newLISP

Post by cormullion »

Just thought I'd pull these out of the forums and sort them, for fun.

Here are the results of running qa-bench from the source distribution on various pieces of kit:

Code: Select all

$ newlisp ./newlisp-.../qa-specific-tests/qa-bench
New series: newLISP version 10.6.x

Code: Select all

0.79 ; MacPro "recycle bin"; 2.3GHz Intel Core i5, OS X 10.9, newLISP v10.6.0-64-bit - kanen 
0.90 ; 2.7GHz Intel Core i5 iMac, OS X 10.9, newLISP v10.6.2-64-bit - cormullion
1.53 ; 2.0 GHz Intel Core 2 Duo, OS X 10.9, newLISP v10.6.2-64-bit - cormullion
9.29 ; 1.5GHz A9 (ARM) dual core Android, newLISP v10.6.0-64-bit — ralph ronnquist
Old series: newLISP versions up to 10.?

Code: Select all

0.33 ; 2.7GHz Intel Core i5 iMac, 64 bit newLISP 10.3.2 - cormullion
0.45 ; 2.4GHz Intel Core i5 MacBook Pro, 64 bit newLISP  - itistoday
0.5  ; 2.2Ghz AMD Phenom(tm) 9550 Quad-Core Processor 64-bit on Linux IPv4 - pjot
0.55 ; Windows XP at AMD Phenom II X2 545, 3 GHz - Cyril 
0.6  ; 2 x 3.2 GHz Quad-Core Intel Xeon - joe 
0.63 ; FreeBSD at NFSHOST poss. 2.8 GHZ CPU - lutz
0.7  ; FreeBSD at NFSHOST probably the same on a bad day - cormullion
0.71 ; Mac OS X 2.0 GHz Intel Core 2 Duo, 64-bit version of newLISP - cormullion
0.75 ; MacBook Pro 2.4 GHz Intel Core 2 Duo running 32-bit newLISP - hilti
0.8  ; AMD 64 3200+ - newdep
0.89 ; zLinux (for the IBM mainframe) - jopython 
0.9  ; Mac OS X 2.0 GHz Intel Core 2 Duo, 32-bit newLISP- cormullion
1.00 ; Mac OS X 1.83 GHz Intel Core 2 Duo - Lutz
1.1 ; Windows Vista 64 at Intel Pentium D 940, 3.2 Ghz. - kazimir
1.36 ; Pentium 4, 2Ghz running Ubuntu 9.04 - robert gorenc
2.24 ; Sun Sparc 1350MHZ processor - jopython
3.25 ; Windows XP at Intel Pentium III, 800 MHz - Cyril
3.40 ; Nokia N900 at 950 MHz - hilti's "numbercruncher"
5.15 ; Nokia N900 at 700 MHz
5.37 ; Raspberry Pi 900 mHz (overclocked with the raspi-config tool) - Hilti
5.44 ; Mac OS X 1GHz PowerPC G4 (eMac) - cormullion
6.72 ; Raspberry Pi 700 mHz  256 MB RAM - Hilti
9.52 ; Sun Sparc Ultra-2 - lutz
13.7 ; Nokia N810 armv61 - newdep
30.64 ; Pentium 90, running DamnSmallLinux - robert gorenc
50.0 ; Intel Pentium 120 - P54CQS - 120MHz - xytroxon

xytroxon
Posts: 296
Joined: Tue Nov 06, 2007 3:59 pm
Contact:

Re: Benchmarking newLISP

Post by xytroxon »

I upgraded over the summer...

2.1GHz AMD Athlon II Dual-Core P320

Image

Number One... Engage...

-- xytroxon ;)
"Many computers can print only capital letters, so we shall not use lowercase letters."
-- Let's Talk Lisp (c) 1976

hilti
Posts: 140
Joined: Sun Apr 19, 2009 10:09 pm
Location: Hannover, Germany
Contact:

Re: Benchmarking newLISP

Post by hilti »

What kind of benchmark was run? I'd like to contribute my numbers just for fun and doing some tests on my "numbercruncher" (a Nokia N900) ;-)

Cheers
Hilti
--()o Dragonfly web framework for newLISP
http://dragonfly.apptruck.de

cormullion
Posts: 2038
Joined: Tue Nov 29, 2005 8:28 pm
Location: latiitude 50N longitude 3W
Contact:

Re: Benchmarking newLISP

Post by cormullion »

This is running qa-bench from the source distribution:

$ newlisp ./newlisp-10.2.8/qa-specific-tests/qa-bench

Lutz - I found that my new 64 bit version of newLISP was consistently faster - 0.7 rather than 0.9. Cool!

Cyril
Posts: 183
Joined: Tue Oct 30, 2007 6:27 pm
Location: Moscow, Russia
Contact:

Re: Benchmarking newLISP

Post by Cyril »

xytroxon wrote:I upgraded over the summer...
I am in upgrade just now! This is probably the very last message I write from the my old box. So...

Code: Select all

3.25 ; Windows XP at Intel Pentium III, 800 MHz - Cyril
And the result from the my new box will be in a few hours! ;-)

Update:

Code: Select all

0.55 ; Windows XP at AMD Phenom II X2 545, 3 GHz - Cyril
With newLISP you can grow your lists from the right side!

cormullion
Posts: 2038
Joined: Tue Nov 29, 2005 8:28 pm
Location: latiitude 50N longitude 3W
Contact:

Re: Benchmarking newLISP

Post by cormullion »

If only all upgrades showed so much improvement... :)

itistoday
Posts: 429
Joined: Sun Dec 02, 2007 5:10 pm
Contact:

Re: Benchmarking newLISP

Post by itistoday »

Nice, 64-bit is definitely a little bit faster:

Code: Select all

$ ./newlisp qa-specific-tests/qa-bench 

Benchmarking all non I/O primitives
    1016.4 ms
>>>>> Performance ratio: 0.45 (1.0 on Mac OS X, 1.83 GHz Intel Core 2 Duo)
2.4GHz Intel Core i5 MacBook Pro here. :-)
Get your Objective newLISP groove on.

Kazimir Majorinc
Posts: 388
Joined: Thu May 08, 2008 1:24 am
Location: Croatia
Contact:

Re: Benchmarking newLISP

Post by Kazimir Majorinc »

1.1, Windows Vista 64 at Intel Pentium D 940, 3.2 Ghz.

I just concluded that I do not need new hardware for next few years.
qa-bench could be integrated in Newlisp core as function, people like benchmarks.

Robert Gorenc
Posts: 12
Joined: Tue Oct 27, 2009 9:01 pm
Location: Croatia

Re: Benchmarking newLISP

Post by Robert Gorenc »

30.64 on Pentium 90, running DamnSmallLinux (old gran'pa is still running :-) )
and
1.36 Pentium 4, 2Ghz running Ubuntu 9.04

hilti
Posts: 140
Joined: Sun Apr 19, 2009 10:09 pm
Location: Hannover, Germany
Contact:

Re: Benchmarking newLISP

Post by hilti »

Hmm...I'm always getting this message when running the qa-bench script.

Code: Select all

 newlisp ./newlisp-10.2.8/qa-specific-tests/qa-bench

Benchmarking all non I/O primitives
    1608,4 ms

ERR: value expected in function mul : .5
--()o Dragonfly web framework for newLISP
http://dragonfly.apptruck.de

cormullion
Posts: 2038
Joined: Tue Nov 29, 2005 8:28 pm
Location: latiitude 50N longitude 3W
Contact:

Re: Benchmarking newLISP

Post by cormullion »

Looks like a locale problem - commas for decimal points, then decimal points not recognized?

As a temporary fix, you could try inserting a (set-locale "en_US") statement near the top...

hilti
Posts: 140
Joined: Sun Apr 19, 2009 10:09 pm
Location: Hannover, Germany
Contact:

Re: Benchmarking newLISP

Post by hilti »

Now it works - thanks a lot.

This is my MacBook Pro 2.4 GHz Intel Core 2 Duo running 32-bit newLISP

Code: Select all

Benchmarking all non I/O primitives
    1788.7 ms
>>>>> Performance ratio: 0.75 (1.0 on Mac OS X, 1.83 GHz Intel Core 2 Duo)
--()o Dragonfly web framework for newLISP
http://dragonfly.apptruck.de

hilti
Posts: 140
Joined: Sun Apr 19, 2009 10:09 pm
Location: Hannover, Germany
Contact:

Re: Benchmarking newLISP

Post by hilti »

Nokia N900 at 950 MHz - that's my "numbercruncher" ;-)

Code: Select all

8540.2 ms >>>>> Performance ratio: 3.40
Nokia N900 at 700 MHz results in

Code: Select all

12999.1 ms >>>>> Performance ratio: 5.15
--()o Dragonfly web framework for newLISP
http://dragonfly.apptruck.de

cormullion
Posts: 2038
Joined: Tue Nov 29, 2005 8:28 pm
Location: latiitude 50N longitude 3W
Contact:

Re: Benchmarking newLISP

Post by cormullion »

not bad for a phone, though!

If someone tells me how to compile newLISP for an iPod Touch, I have a jail-broken one lying around ...

cormullion
Posts: 2038
Joined: Tue Nov 29, 2005 8:28 pm
Location: latiitude 50N longitude 3W
Contact:

Re: Benchmarking newLISP

Post by cormullion »

Updated :)

hilti
Posts: 140
Joined: Sun Apr 19, 2009 10:09 pm
Location: Hannover, Germany
Contact:

Re: Benchmarking newLISP

Post by hilti »

Here's a result for newLISP 10.4.5 on a Raspberry Pi (700 mHZ, 256 MB RAM)

Code: Select all

pi@raspberrypi /tmp/newlisp-10.4.5/qa-specific-tests $ newlisp qa-bench 

>>>>> Benchmarking all non I/O primitives ... may take a while ...
   17001.4 ms
>>>>> Performance ratio: 6.72 (1.0 on Mac OSX, 1.83 GHz Core 2 Duo, newLISP v10.2.8)
Next step is to host some newLISP sites on my Raspberry Pi.

Update:
Now http://www.rundragonfly.com is running on the Raspberry Pi. Performance feels quite good.
Last edited by hilti on Sat Dec 29, 2012 9:58 am, edited 2 times in total.
--()o Dragonfly web framework for newLISP
http://dragonfly.apptruck.de

hilti
Posts: 140
Joined: Sun Apr 19, 2009 10:09 pm
Location: Hannover, Germany
Contact:

Re: Benchmarking newLISP

Post by hilti »

Here is a result at 900 mHz (overclocked with the raspi-config tool)

Code: Select all

>>>>> Benchmarking all non I/O primitives ... may take a while ...
   13575.5 ms
>>>>> Performance ratio: 5.37 (1.0 on Mac OSX, 1.83 GHz Core 2 Duo, newLISP v10.2.8)
--()o Dragonfly web framework for newLISP
http://dragonfly.apptruck.de

Ryon
Posts: 248
Joined: Thu Sep 26, 2002 12:57 am

Re: Benchmarking newLISP

Post by Ryon »

Forbidden

You don't have permission to access / on this server.
Apache/2.2.22 (Debian) Server at http://www.rundragonfly.com Port 80
But is this a useful purpose? :)

hilti
Posts: 140
Joined: Sun Apr 19, 2009 10:09 pm
Location: Hannover, Germany
Contact:

Re: Benchmarking newLISP

Post by hilti »

Corrected. I've just played around ;-)

I think newLISP and the Raspberry Pi are a wonderful couple, because

1. newLISP can be used for distributed computing tasks between several Raspberry Pi's
2. Uses less memory than other scripting languages (I think Python on a Pi is too much)
3. Is fast enough to serve simple dynamic websites (faster than PHP+MySQL on a Pi)

Has anyone a Raspberry Pi at home?

Cheers
Hilti
--()o Dragonfly web framework for newLISP
http://dragonfly.apptruck.de

cormullion
Posts: 2038
Joined: Tue Nov 29, 2005 8:28 pm
Location: latiitude 50N longitude 3W
Contact:

Re: Benchmarking newLISP

Post by cormullion »

Haven't got one, but I always thought it was a great idea. However I'd picked up the impression that it was really fast —people talking about HD video and things —but it's obviously not yet up to PC speed yet for general tasks. However, putting a few together might be a different story.

kanen
Posts: 145
Joined: Thu Mar 25, 2010 6:24 pm
Contact:

Re: Benchmarking newLISP

Post by kanen »

> total time: 1947.244
> Performance ratio: 0.79 (1.0 on MacOSX 10.9, 2.3GHz Intel Core i5, newLISP v10.6.0-64-bit)

This is on the Mac Pro 2013 Recycle Bin computer. :)
. Kanen Flowers http://kanen.me .

cormullion
Posts: 2038
Joined: Tue Nov 29, 2005 8:28 pm
Location: latiitude 50N longitude 3W
Contact:

Re: Benchmarking newLISP

Post by cormullion »

Seems a bit slow - presumably that's due to newLISP 10.6.0 rather than OSX 10.9 or the Recycle Bin...?

ralph.ronnquist
Posts: 228
Joined: Mon Jun 02, 2014 1:40 am
Location: Melbourne, Australia

Re: Benchmarking newLISP

Post by ralph.ronnquist »

This is my result trialling last week's 10.6.3 on a 1.5GHz A9 (ARM) dual core TV box (hardware "Amlogic Meson8B"):
>>>>> Benchmarking all non I/O primitives ... (may take a while)
>>>>> total time: 24184.55799999999
>>>>> Performance ratio: 9.29 (1.0 on MacOSX 10.9, 2.3GHz Intel Core i5, newLISP v10.6.0-64-bit)

cormullion
Posts: 2038
Joined: Tue Nov 29, 2005 8:28 pm
Location: latiitude 50N longitude 3W
Contact:

Re: Benchmarking newLISP

Post by cormullion »

newLISP 10.6 looks to be slower than older versions — I'm only getting 0.9 seconds now on 10.6, compared with 0.4 with newLISP 10.3. I was going to investigate, but the benchmark "qa-bench" in the current distribution isn't backwards compatible with earlier versions... Which makes the top post fairly meaningless now, since they're all different version numbers. :)

Lutz
Posts: 5289
Joined: Thu Sep 26, 2002 4:45 pm
Location: Pasadena, California
Contact:

Re: Benchmarking newLISP

Post by Lutz »

In my own benchmarks 10.6.x is definitely not slower, rather faster (< 1%) than previous versions. The benchmarks have changed over versions, and were recalibrated when changing to a new Mac mini in 2011 around 10.3.x. When recompiling older versions calibrated on an older 2007 Mac mini on a newer 2011 Mac mini, the older versions will give faster (lower ratios) because calibrated to an older model CPU.

Over the years newLISP only has gotten faster, never slower. The last, minor speedup in 10.4.7, when eliminating the strncat() C function for security reasons.

All benchmarks are done on Mac OS X, Linux and Windows XP. Linux is always the fastest clocking in at 0.93 to 0.94 compared to OS X

On OS X 10.10 Yosemete, newLISP has gotten slower and average of 1% comparing to OS X 10.9 Maverick.

Locked