Re: speedup buffer parsing by a factor of 3

2002-05-29 Thread Lars Gullik Bjønnes
Jean-Marc Lasgouttes <[EMAIL PROTECTED]> writes: | Lars Gullik Bjønnes wrote: >> Andre Poenitz <[EMAIL PROTECTED]> writes: >> | On Wed, May 29, 2002 at 06:43:33PM +0200, Lars Gullik Bjønnes >> wrote: >> I'd say that the performance hit is negligible, and most of it can be fixed with some

Re: speedup buffer parsing by a factor of 3

2002-05-29 Thread Jean-Marc Lasgouttes
Lars Gullik Bjønnes wrote: > Andre Poenitz <[EMAIL PROTECTED]> writes: > > | On Wed, May 29, 2002 at 06:43:33PM +0200, Lars Gullik Bjønnes wrote: > >>>I'd say that the performance hit is negligible, and most of it can be >>>fixed with some more clever structurs and algorithms. >> > | We spend 25

Re: speedup buffer parsing by a factor of 3

2002-05-29 Thread Lars Gullik Bjønnes
Andre Poenitz <[EMAIL PROTECTED]> writes: | On Wed, May 29, 2002 at 06:43:33PM +0200, Lars Gullik Bjønnes wrote: >> I'd say that the performance hit is negligible, and most of it can be >> fixed with some more clever structurs and algorithms. > | We spend 25% of the LyX related UserGuide loading

Re: speedup buffer parsing by a factor of 3

2002-05-29 Thread Andre Poenitz
On Wed, May 29, 2002 at 06:43:33PM +0200, Lars Gullik Bjønnes wrote: > I'd say that the performance hit is negligible, and most of it can be > fixed with some more clever structurs and algorithms. We spend 25% of the LyX related UserGuide loading time in LyXTextClass::operator[] if I read the gpr

Re: speedup buffer parsing by a factor of 3

2002-05-29 Thread Lars Gullik Bjønnes
Jean-Marc Lasgouttes <[EMAIL PROTECTED]> writes: >> "Lars" == Lars Gullik Bjønnes <[EMAIL PROTECTED]> writes: > | Lars> We have other thing to do to speed up things before we begin | Lars> with tricks like these... > | Probably. > | Lars> _and_ we must assume that the implementation that we

Re: speedup buffer parsing by a factor of 3

2002-05-29 Thread Lars Gullik Bjønnes
Jean-Marc Lasgouttes <[EMAIL PROTECTED]> writes: >> "Andre" == Andre Poenitz <[EMAIL PROTECTED]> writes: > | Andre> On Wed, May 29, 2002 at 05:51:25PM +0200, Lars Gullik Bjønnes | Andre> wrote: >>> | Apart from that: Why does a paragraph carry its layout as string >>> and not as | index in th

Re: speedup buffer parsing by a factor of 3

2002-05-29 Thread Jean-Marc Lasgouttes
> "Lars" == Lars Gullik Bjønnes <[EMAIL PROTECTED]> writes: Lars> We have other thing to do to speed up things before we begin Lars> with tricks like these... Probably. Lars> _and_ we must assume that the implementation that we Lars> use will use the fastest way of comparing two strings.

Re: speedup buffer parsing by a factor of 3

2002-05-29 Thread Jean-Marc Lasgouttes
> "Andre" == Andre Poenitz <[EMAIL PROTECTED]> writes: Andre> On Wed, May 29, 2002 at 05:51:25PM +0200, Lars Gullik Bjønnes Andre> wrote: >> | Apart from that: Why does a paragraph carry its layout as string >> and not as | index in the layout table? >> >> Why does it not carry a pointer to

Re: speedup buffer parsing by a factor of 3

2002-05-29 Thread Andre Poenitz
On Wed, May 29, 2002 at 05:51:25PM +0200, Lars Gullik Bjønnes wrote: > | Apart from that: Why does a paragraph carry its layout as string and not as > | index in the layout table? > > Why does it not carry a pointer to the actual layout object? I don't know ;-) Andre' -- Those who desire to g

Re: speedup buffer parsing by a factor of 3

2002-05-29 Thread Lars Gullik Bjønnes
Andre Poenitz <[EMAIL PROTECTED]> writes: >> You can't be much faster than that, if this is really the bottleneck. > | I think we have a right to use a "const_string" class... > | Apart from that: Why does a paragraph carry its layout as string and not as | index in the layout table? Why does it

Re: speedup buffer parsing by a factor of 3

2002-05-29 Thread John Levon
On Wed, May 29, 2002 at 05:40:06PM +0200, Jean-Marc Lasgouttes wrote: > There is not much on this site. Check the CVS ... it's /very/ simple i.e. just rdtsc calls (only useful if you've x86/Linux of course ...) regards john -- "A Mini Cooper ? I wouldn't be seen dead in one of those !"

Re: speedup buffer parsing by a factor of 3

2002-05-29 Thread Andre Poenitz
On Wed, May 29, 2002 at 05:14:21PM +0200, Lars Gullik Bjønnes wrote: > | I wonder if we should replace the layout vector by a map indexed by the > | layout name... > > Can you wait a bit with this one? Too late. But it does not improve things, so I won't bother you with a patch... Andre' --

Re: speedup buffer parsing by a factor of 3

2002-05-29 Thread Lars Gullik Bjønnes
Jean-Marc Lasgouttes <[EMAIL PROTECTED]> writes: >> "Andre" == Andre Poenitz <[EMAIL PROTECTED]> writes: > | Andre> On Wed, May 29, 2002 at 04:53:56PM +0200, Jean-Marc Lasgouttes | Andre> wrote: >>> It should. Note that there is a lot of time needed to load the >>> fonts from the X server tha

Re: speedup buffer parsing by a factor of 3

2002-05-29 Thread Andre Poenitz
On Wed, May 29, 2002 at 05:19:08PM +0200, Jean-Marc Lasgouttes wrote: > Andre> LyXTextClass::operator[] is pretty expensive. > > This is because of string comparisons which are done for the cached > value. I do not think we can go much faster, except if we used a > smarter operator== for strings

Re: speedup buffer parsing by a factor of 3

2002-05-29 Thread Jean-Marc Lasgouttes
> "John" == John Levon <[EMAIL PROTECTED]> writes: John> On Wed, May 29, 2002 at 05:28:52PM +0200, Jean-Marc Lasgouttes John> wrote: >> This seems realy strange: when eon loads the user guide, I would >> say hat the messages about loading fonts are visible more than 0.5% >> of the time... Jo

Re: speedup buffer parsing by a factor of 3

2002-05-29 Thread John Levon
On Wed, May 29, 2002 at 05:28:52PM +0200, Jean-Marc Lasgouttes wrote: > This seems realy strange: when eon loads the user guide, I would say > hat the messages about loading fonts are visible more than 0.5% of the > time... I suppose I should build an optimised build and some X11 libraries with

Re: speedup buffer parsing by a factor of 3

2002-05-29 Thread Jean-Marc Lasgouttes
> "John" == John Levon <[EMAIL PROTECTED]> writes: John> On Wed, May 29, 2002 at 05:10:16PM +0200, Jean-Marc Lasgouttes John> wrote: >> This would be useful to actually profile libX11, no? All I want is >> see the time spent in loading fonts from the server. John> Well this will show you whe

Re: speedup buffer parsing by a factor of 3

2002-05-29 Thread John Levon
On Wed, May 29, 2002 at 05:10:16PM +0200, Jean-Marc Lasgouttes wrote: > This would be useful to actually profile libX11, no? All I want is > see the time spent in loading fonts from the server. Well this will show you where time was spent inside libX11, so if you only use LyX when you do it, it

Re: speedup buffer parsing by a factor of 3

2002-05-29 Thread Lars Gullik Bjønnes
Andre Poenitz <[EMAIL PROTECTED]> writes: | On Wed, May 29, 2002 at 04:53:56PM +0200, Jean-Marc Lasgouttes wrote: >> It should. Note that there is a lot of time needed to load the fonts >> from the X server that gprof does not take in account. I tried to >> install the gprof-aware libc packages,

Re: speedup buffer parsing by a factor of 3

2002-05-29 Thread Jean-Marc Lasgouttes
> "Andre" == Andre Poenitz <[EMAIL PROTECTED]> writes: Andre> On Wed, May 29, 2002 at 04:53:56PM +0200, Jean-Marc Lasgouttes Andre> wrote: >> It should. Note that there is a lot of time needed to load the >> fonts from the X server that gprof does not take in account. I >> tried to install th

Re: speedup buffer parsing by a factor of 3

2002-05-29 Thread Jean-Marc Lasgouttes
> "John" == John Levon <[EMAIL PROTECTED]> writes: John> On Wed, May 29, 2002 at 04:53:56PM +0200, Jean-Marc Lasgouttes John> wrote: >> It should. Note that there is a lot of time needed to load the >> fonts from the X server that gprof does not take in account. I >> tried to install the gpro

Re: speedup buffer parsing by a factor of 3

2002-05-29 Thread Andre Poenitz
On Wed, May 29, 2002 at 04:53:56PM +0200, Jean-Marc Lasgouttes wrote: > It should. Note that there is a lot of time needed to load the fonts > from the X server that gprof does not take in account. I tried to > install the gprof-aware libc packages, but I do not know how to use > this. LyXTextCla

Re: speedup buffer parsing by a factor of 3

2002-05-29 Thread John Levon
On Wed, May 29, 2002 at 04:53:56PM +0200, Jean-Marc Lasgouttes wrote: > It should. Note that there is a lot of time needed to load the fonts > from the X server that gprof does not take in account. I tried to > install the gprof-aware libc packages, but I do not know how to use > this. export LD

Re: speedup buffer parsing by a factor of 3

2002-05-29 Thread Jean-Marc Lasgouttes
> "Andre" == Andre Poenitz <[EMAIL PROTECTED]> writes: Andre> This tranlates to 15% overall performance boost when simply Andre> loading the UserGuide? Andre> Not bad... It should. Note that there is a lot of time needed to load the fonts from the X server that gprof does not take in accoun

speedup buffer parsing by a factor of 3

2002-05-29 Thread Andre Poenitz
This tranlates to 15% overall performance boost when simply loading the UserGuide? Not bad... Andre' -- Those who desire to give up Freedom in order to gain Security, will not have, nor do they deserve, either one. (T. Jefferson)