Python is slow on SuSE 8.2 vs other OS's
I benchmarked Python on SuSE 8.2 as I normally do with a new OS on my laptop. My somnec.py routine runs at 450sec on 8.2, vs. 300 sec on FreeBSD, vs. 250 sec on Windows. I have a 1Ghz P4 on a Vaio FX290. Somnec is a routine which mainly eats up cpu time doing variable width Romberg integration. I am wondering if there is a way to compile Python from SRPM and then give it compiler flags like -O2 -march=pentiumpro? I've never done that before so maybe I will RTFM :) On FreeBSD I just put make.conf into /etc. Is there an equivalent for Linux? Thanks, Rob. -- ------------------------- Numeric Python EM Project www.pythonemproject.com
On Sun, 24 Mar 2002 09:43:41 -0800 rob <rob3@pythonemproject.com> wrote:
I benchmarked Python on SuSE 8.2 as I normally do with a new OS on my
there is no evidence python will work on suse 8.2 heheheh is not out yet
Landy Roman wrote:
On Sun, 24 Mar 2002 09:43:41 -0800 rob <rob3@pythonemproject.com> wrote:
I benchmarked Python on SuSE 8.2 as I normally do with a new OS on my
there is no evidence python will work on suse 8.2
heheheh is not out yet
Huh? Python 2.1 was installed from the CDROM. Rob. -- ------------------------- Numeric Python EM Project www.pythonemproject.com
rob wrote:
Landy Roman wrote:
On Sun, 24 Mar 2002 09:43:41 -0800 rob <rob3@pythonemproject.com> wrote:
I benchmarked Python on SuSE 8.2 as I normally do with a new OS on my
there is no evidence python will work on suse 8.2
heheheh is not out yet
Huh? Python 2.1 was installed from the CDROM. Rob.
--
I mean Suse 7.3. I am getting my distros mixed up, as I just jettisoned Mandrake 8.2 :) :)
------------------------- Numeric Python EM Project
www.pythonemproject.com
-- ------------------------- Numeric Python EM Project www.pythonemproject.com
Can anyone comment on the factors that could potentially contribute to the empirical evidence presented regarding the run-time performance of Python on the 7.3 SuSE kit? I would be interested (as others might be, too) in a discussion of the factors that would cause this... Especially considering the hardware appears to be a constant in this evaluation... The reason I am interested is because I am seeing an increase in the number of computationally-intensive "shops" ambracing Python as a cornerstone development tool. Hence, I like to know where it runs fastest... ...and possibly, why. JK At 9:43 AM -0800 3/24/02, rob wrote:
From: rob <rob3@pythonemproject.com> X-Accept-Language: en To: "suse-linux-e@suse.com" <suse-linux-e@suse.com> X-Loop-Detect: 1 Subject: [SLE] Python is slow on SuSE 8.2 vs other OS's Status:
I benchmarked Python on SuSE 7.3 as I normally do with a new OS on my laptop. My somnec.py routine runs at 450sec on 8.2, vs. 300 sec on FreeBSD, vs. 250 sec on Windows. I have a 1Ghz P4 on a Vaio FX290. Somnec is a routine which mainly eats up cpu time doing variable width Romberg integration.
I am wondering if there is a way to compile Python from SRPM and then give it compiler flags like -O2 -march=pentiumpro? I've never done that before so maybe I will RTFM :)
On FreeBSD I just put make.conf into /etc. Is there an equivalent for Linux? Thanks, Rob.
-- ------------------------- Numeric Python EM Project
www.pythonemproject.com
John Koenig wrote:
Can anyone comment on the factors that could potentially contribute to the empirical evidence presented regarding the run-time performance of Python on the 7.3 SuSE kit?
I would be interested (as others might be, too) in a discussion of the factors that would cause this... Especially considering the hardware appears to be a constant in this evaluation...
The reason I am interested is because I am seeing an increase in the number of computationally-intensive "shops" ambracing Python as a cornerstone development tool. Hence, I like to know where it runs fastest...
...and possibly, why.
JK
At 9:43 AM -0800 3/24/02, rob wrote:
From: rob <rob3@pythonemproject.com> X-Accept-Language: en To: "suse-linux-e@suse.com" <suse-linux-e@suse.com> X-Loop-Detect: 1 Subject: [SLE] Python is slow on SuSE 8.2 vs other OS's Status:
I benchmarked Python on SuSE 7.3 as I normally do with a new OS on my laptop. My somnec.py routine runs at 450sec on 8.2, vs. 300 sec on FreeBSD, vs. 250 sec on Windows. I have a 1Ghz P4 on a Vaio FX290. Somnec is a routine which mainly eats up cpu time doing variable width Romberg integration.
I am wondering if there is a way to compile Python from SRPM and then give it compiler flags like -O2 -march=pentiumpro? I've never done that before so maybe I will RTFM :)
On FreeBSD I just put make.conf into /etc. Is there an equivalent for Linux? Thanks, Rob.
-- ------------------------- Numeric Python EM Project
www.pythonemproject.com
-- To unsubscribe send e-mail to suse-linux-e-unsubscribe@suse.com For additional commands send e-mail to suse-linux-e-help@suse.com Also check the FAQ at http://www.suse.com/support/faq and the archives at http://lists.suse.com
I am going to get the SRPM and edit the spec file for better compilation options. That will be interesting. I've found in general that the ActiveState Windows Python distributions are the fastest. One my lowly notebook with Win2k they even match my FreeBSD Athlon DDR system. At work I have an Intel P4 1.8Ghz on an Asus P4S. It runs the same program in 180s on NT. You can get the program from my website. Its a good one for general Python math benchmarking since it really doesn't exercise Numpy very much, since there is little to vectorize. Its in the neclib.tgz distribution. Rob. -- ------------------------- Numeric Python EM Project www.pythonemproject.com
Do you mean something like "rpm --rebuild --target=$(uname -m) python.src.rpm" ? söndagen den 24 mars 2002 18.43 skrev rob:
I benchmarked Python on SuSE 8.2 as I normally do with a new OS on my laptop. My somnec.py routine runs at 450sec on 8.2, vs. 300 sec on FreeBSD, vs. 250 sec on Windows. I have a 1Ghz P4 on a Vaio FX290. Somnec is a routine which mainly eats up cpu time doing variable width Romberg integration.
I am wondering if there is a way to compile Python from SRPM and then give it compiler flags like -O2 -march=pentiumpro? I've never done that before so maybe I will RTFM :)
On FreeBSD I just put make.conf into /etc. Is there an equivalent for Linux? Thanks, Rob.
Anders Dahlqvist wrote:
Do you mean something like "rpm --rebuild --target=$(uname -m) python.src.rpm" ?
söndagen den 24 mars 2002 18.43 skrev rob:
I benchmarked Python on SuSE 8.2 as I normally do with a new OS on my laptop. My somnec.py routine runs at 450sec on 8.2, vs. 300 sec on FreeBSD, vs. 250 sec on Windows. I have a 1Ghz P4 on a Vaio FX290. Somnec is a routine which mainly eats up cpu time doing variable width Romberg integration.
I am wondering if there is a way to compile Python from SRPM and then give it compiler flags like -O2 -march=pentiumpro? I've never done that before so maybe I will RTFM :)
On FreeBSD I just put make.conf into /etc. Is there an equivalent for Linux? Thanks, Rob.
-- To unsubscribe send e-mail to suse-linux-e-unsubscribe@suse.com For additional commands send e-mail to suse-linux-e-help@suse.com Also check the FAQ at http://www.suse.com/support/faq and the archives at http://lists.suse.com
Dear Anders, I forgot about your message, and tried rpm -bb --target=i686 python.spec It did everything except make an RPM. I followed the RPM manual as far as I can see. I wonder what went wrong? I do have w working python (i686 version), but it is just the binary in BUILD and in /var/tmp/python-root. Rob. -- ------------------------- Numeric Python EM Project www.pythonemproject.com
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On March 25, 2002 10:54 am, rob wrote: ...snip...
Dear Anders, I forgot about your message, and tried rpm -bb --target=i686 python.spec
It did everything except make an RPM. I followed the RPM manual as far as I can see. I wonder what went wrong? I do have w working python (i686 version), but it is just the binary in BUILD and in /var/tmp/python-root.
RPM *really* sucks at reporting errors. Sometimes you have to scroll up a bit to see a "File not found". I have no idea why it doesn't stick a summary of the errors at the bottom... More of Red Hat's "infinite wisdom" I guess. This is usually frustrating because there's no way to continue in RPM. You have to fix the error (hopefully) and rebuild from scratch. You can see where the 'make install' went in '/var/tmp/python-root' (or something like that) - -- James Oakley Engineering - SolutionInc Ltd. joakley@solutioninc.com http://www.solutioninc.com -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.0.6 (GNU/Linux) Comment: For info see http://www.gnupg.org iD8DBQE8nz/b+FOexA3koIgRAg9LAKCxSBmBzNSVN9pZkADQszsY9W1mIwCgqdMu dge3PwG8gODx5p4iKQoQdhg= =fmNi -----END PGP SIGNATURE-----
participants (6)
-
Anders Dahlqvist -
James Oakley -
John Koenig -
Landy Roman -
rob -
schuetzen - RKBA!