I installed the minimals and use MKIV. It seems that it is 1,5 to 2 times as slow as the MKII I used in texlive. Is this possible, or should there be something else that is responsible for this? -- Cecil Westerhof
Am 14.03.2011 um 11:23 schrieb Cecil Westerhof:
I installed the minimals and use MKIV. It seems that it is 1,5 to 2 times as slow as the MKII I used in texlive. Is this possible,
Yes, MkIV is slower.
or should there be something else that is responsible for this?
No, for simple document with lots of text MkIV is slower but there are parts (e.g. MetaPost) where MKIV is a lot faster than MkII. Wolfgang
2011/3/14 Wolfgang Schuster
I installed the minimals and use MKIV. It seems that it is 1,5 to 2 times as slow as the MKII I used in texlive. Is this possible,
Yes, MkIV is slower.
Okay, thanks. I also updated my system. So that was the other possibility. No, for simple document with lots of text MkIV is slower but there are
parts (e.g. MetaPost) where MKIV is a lot faster than MkII.
My document is mostly text. -- Cecil Westerhof
On Mon, 14 Mar 2011, Cecil Westerhof wrote:
2011/3/14 Wolfgang Schuster
I installed the minimals and use MKIV. It seems that it is 1,5 to 2 times as slow as the MKII I used in texlive. Is this possible,
Yes, MkIV is slower.
Okay, thanks. I also updated my system. So that was the other possibility.
In my experience context --silent=\* <filename> speeds up compilation by a factor of 1.5 (but you loose all the logging messages) Aditya
On Monday 14 March 2011 14:02:18 Aditya Mahajan wrote:
On Mon, 14 Mar 2011, Cecil Westerhof wrote:
2011/3/14 Wolfgang Schuster
I installed the minimals and use MKIV. It seems that it is 1,5 to 2 times
as slow as the MKII I used in texlive. Is this possible,
Yes, MkIV is slower.
Okay, thanks. I also updated my system. So that was the other possibility.
In my experience
context --silent=\* <filename>
speeds up compilation by a factor of 1.5 (but you loose all the logging messages)
or context --batch <filename>
2011/3/14 Alan BRASLAU
On Monday 14 March 2011 14:02:18 Aditya Mahajan wrote:
On Mon, 14 Mar 2011, Cecil Westerhof wrote:
2011/3/14 Wolfgang Schuster
I installed the minimals and use MKIV. It seems that it is 1,5 to 2 times
as slow as the MKII I used in texlive. Is this possible,
Yes, MkIV is slower. context --silent=\* <filename>
speeds up compilation by a factor of 1.5 (but you loose all the logging messages)
or
context --batch <filename>
Both did not work for me. But removing --purgeall from my script reduced the time from 30 seconds to 10. And because I generate (at the moment) 5 personalised versions, this is a difference from 2:44 to 0:56. -- Cecil Westerhof
On Mon, Mar 14, 2011 at 14:47, Cecil Westerhof wrote:
Both did not work for me. But removing --purgeall from my script reduced the time from 30 seconds to 10.
When you don't use --purgeall, ConTeXt calculates different things (for example table of contents, cross-references etc.) and stores them to temporary files. Next time when you compile the same document without too many changes it simply reuses the old data. But if you remove the temporary files with --purgeall, ConTeXt has to recalculate everything from scratch. Out of curiosity I checked that on my own document and realized exactly the same thing. Compile time dropped from 27 to 9 seconds, but only because ConTeXt had to read and typeset the document three times (I thought it usually did it twice). If you remove all the temporary files and call context without --purgeall, it will also take 30 seconds to typeset everything; it is only the second and all the subsequent runs that finish the job faster. Mojca PS: You would get the same kind of behaviour in MKII (however if MKII only runs twice and if there is a speed factor of 1.5, you could declare MKII being "three times faster" which does make some difference when compilation time is long). PPS (not to be taken (too) seriously): But I wouldn't be surprized if, say, two years from now you would try to repeat the experiment just to find out that MKIV became faster. (Unlikely to happen, but imaging Taco coming to idea to use all the four processor cores of your new machine and Hans reducing the number of required runs from three to two plus some extra optimizations.)
You can't have it Good, Cheap, and Fast all at once ;) On Mon, Mar 14, 2011 at 8:52 PM, Mojca Miklavec < mojca.miklavec.lists@gmail.com> wrote:
On Mon, Mar 14, 2011 at 14:47, Cecil Westerhof wrote:
Both did not work for me. But removing --purgeall from my script reduced
the
time from 30 seconds to 10.
When you don't use --purgeall, ConTeXt calculates different things (for example table of contents, cross-references etc.) and stores them to temporary files. Next time when you compile the same document without too many changes it simply reuses the old data.
But if you remove the temporary files with --purgeall, ConTeXt has to recalculate everything from scratch. Out of curiosity I checked that on my own document and realized exactly the same thing. Compile time dropped from 27 to 9 seconds, but only because ConTeXt had to read and typeset the document three times (I thought it usually did it twice).
If you remove all the temporary files and call context without --purgeall, it will also take 30 seconds to typeset everything; it is only the second and all the subsequent runs that finish the job faster.
Mojca
PS: You would get the same kind of behaviour in MKII (however if MKII only runs twice and if there is a speed factor of 1.5, you could declare MKII being "three times faster" which does make some difference when compilation time is long).
PPS (not to be taken (too) seriously): But I wouldn't be surprized if, say, two years from now you would try to repeat the experiment just to find out that MKIV became faster. (Unlikely to happen, but imaging Taco coming to idea to use all the four processor cores of your new machine and Hans reducing the number of required runs from three to two plus some extra optimizations.)
___________________________________________________________________________________ If your question is of interest to others as well, please add an entry to the Wiki!
maillist : ntg-context@ntg.nl / http://www.ntg.nl/mailman/listinfo/ntg-context webpage : http://www.pragma-ade.nl / http://tex.aanhet.net archive : http://foundry.supelec.fr/projects/contextrev/ wiki : http://contextgarden.net
___________________________________________________________________________________
Am 2011-03-14 um 20:52 schrieb Mojca Miklavec:
PPS (not to be taken (too) seriously): But I wouldn't be surprized if, say, two years from now you would try to repeat the experiment just to find out that MKIV became faster. (Unlikely to happen, but imaging Taco coming to idea to use all the four processor cores of your new machine and Hans reducing the number of required runs from three to two plus some extra optimizations.)
You can’t use several threads (or even processor cores) at once for linear tasks like TeXing. We already had that discussion several times. (It might be possible to re-write TeX again to parallelize some sub- tasks, but as far as I can imagine, this would yield minor speed gains after a huge programming effort.) Greetlings from Lake Constance! Hraban --- http://www.fiee.net/texnique/ http://wiki.contextgarden.net https://www.cacert.org (I'm an assurer)
On 14-3-2011 8:52, Mojca Miklavec wrote:
PS: You would get the same kind of behaviour in MKII (however if MKII only runs twice and if there is a speed factor of 1.5, you could declare MKII being "three times faster" which does make some difference when compilation time is long).
depends on the changes ... in an automated flow, if you knwo what happens you can limit it with a flag
PPS (not to be taken (too) seriously): But I wouldn't be surprized if, say, two years from now you would try to repeat the experiment just to find out that MKIV became faster. (Unlikely to happen, but imaging
sure, i'm current redoing the font code a bit and gain some time (and memory in mkiv) .. but don't expect miracles Hans ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | voip: 087 875 68 74 | www.pragma-ade.com | www.pragma-pod.nl -----------------------------------------------------------------
On Wed, Mar 16, 2011 at 5:39 PM, Hans Hagen
On 14-3-2011 8:52, Mojca Miklavec wrote:
PS: You would get the same kind of behaviour in MKII (however if MKII only runs twice and if there is a speed factor of 1.5, you could declare MKII being "three times faster" which does make some difference when compilation time is long).
depends on the changes ... in an automated flow, if you knwo what happens you can limit it with a flag
PPS (not to be taken (too) seriously): But I wouldn't be surprized if, say, two years from now you would try to repeat the experiment just to find out that MKIV became faster. (Unlikely to happen, but imaging
sure, i'm current redoing the font code a bit and gain some time (and memory in mkiv) .. but don't expect miracles
Maybe luajit can improve things considerably. -- luigi
On 16-3-2011 5:52, luigi scarso wrote:
Maybe luajit can improve things considerably.
i wonder ... we don't do that many calculations light userdata might help some (is on the agenda to be looke dinto) Hans ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | voip: 087 875 68 74 | www.pragma-ade.com | www.pragma-pod.nl -----------------------------------------------------------------
On 2011-03-14 <11:23:55>, Cecil Westerhof wrote:
I installed the minimals and use MKIV. It seems that it is 1,5 to 2 times as slow as the MKII I used in texlive. Is this possible, or should there be something else that is responsible for this?
Hi Cecil, regarding luatex/mkiv performance there are some comparisons between different tex machines that Hans presented at tug 2009: http://river-valley.tv/the-luatex-project-halfway-to-version-1/ Of course, with respect to the rapid development these might be out of date by now. Best regards, Philipp
-- Cecil Westerhof
___________________________________________________________________________________ If your question is of interest to others as well, please add an entry to the Wiki!
maillist : ntg-context@ntg.nl / http://www.ntg.nl/mailman/listinfo/ntg-context webpage : http://www.pragma-ade.nl / http://tex.aanhet.net archive : http://foundry.supelec.fr/projects/contextrev/ wiki : http://contextgarden.net ___________________________________________________________________________________
participants (10)
-
Aditya Mahajan -
Alan BRASLAU -
Cecil Westerhof -
Hans Hagen -
Henning Hraban Ramm -
John Haltiwanger -
luigi scarso -
Mojca Miklavec -
Philipp Gesang -
Wolfgang Schuster