On Tue, Jan 09, 2007 at 07:52:10PM +0100, Edwin Leuven wrote:
Abdelrazak Younes wrote:
Could you profile this instead:
lyx -e text UserGuide.lyx
another try:
% cumulative self self total
time seconds secondscalls ms/call ms/call name
16.00 0.04
On Tue, Jan 09, 2007 at 07:52:10PM +0100, Edwin Leuven wrote:
> Abdelrazak Younes wrote:
> >Could you profile this instead:
> >lyx -e text UserGuide.lyx
>
> another try:
>
> % cumulative self self total
> time seconds secondscalls ms/call ms/call name
> 16.00
Jean-Marc Lasgouttes wrote:
Abdelrazak == Abdelrazak Younes [EMAIL PROTECTED] writes:
Abdelrazak And got rid of this. So if this profile is correct, the
Abdelrazak attached patch will save more than 8% time at file
Abdelrazak reading. Could you try it out please Edwin?
This looks like a very
Abdelrazak Younes wrote:
Jean-Marc Lasgouttes wrote:
Abdelrazak == Abdelrazak Younes
[EMAIL PROTECTED] writes:
Abdelrazak And got rid of this. So if this profile is correct, the
Abdelrazak attached patch will save more than 8% time at file
Abdelrazak reading. Could you try it out please
Abdelrazak == Abdelrazak Younes [EMAIL PROTECTED] writes:
Abdelrazak But I do not trust cygwin much on this because it seems
Abdelrazak much quicker at the normal console. Could you try it on
Abdelrazak Linux.
Unfortunately, most of my time is already taken by my bitching at
others' work :)
Georg Baum wrote:
Abdelrazak Younes wrote:
Jean-Marc Lasgouttes wrote:
Abdelrazak == Abdelrazak Younes
[EMAIL PROTECTED] writes:
Abdelrazak And got rid of this. So if this profile is correct, the
Abdelrazak attached patch will save more than 8% time at file
Abdelrazak reading. Could you try
On Thu, Jan 11, 2007 at 03:11:31PM +0100, Abdelrazak Younes wrote:
Indeed. On cygwin there's one second improvement (from 7s to 6s) when
loading a really big file (4megs). I repeated the test multiple times to
make sure that the file was already in the cache.
But I do not trust cygwin
Abdelrazak Younes wrote:
You are a complicated guy Georg. I was answering to your false
assumption so, in this regard, you misread the patch. I cannot read your
mind and understand what you don't like.
I might be complicated, but I still do not think that I misread anything. I
only tried to
Jean-Marc Lasgouttes wrote:
"Abdelrazak" == Abdelrazak Younes <[EMAIL PROTECTED]> writes:
Abdelrazak> And got rid of this. So if this profile is correct, the
Abdelrazak> attached patch will save more than 8% time at file
Abdelrazak> reading. Could you try it out please Edwin?
This looks like
Abdelrazak Younes wrote:
> Jean-Marc Lasgouttes wrote:
>>> "Abdelrazak" == Abdelrazak Younes
>>> <[EMAIL PROTECTED]> writes:
>>
>> Abdelrazak> And got rid of this. So if this profile is correct, the
>> Abdelrazak> attached patch will save more than 8% time at file
>> Abdelrazak> reading.
> "Abdelrazak" == Abdelrazak Younes <[EMAIL PROTECTED]> writes:
Abdelrazak> But I do not trust cygwin much on this because it seems
Abdelrazak> much quicker at the "normal" console. Could you try it on
Abdelrazak> Linux.
Unfortunately, most of my time is already taken by my bitching at
Georg Baum wrote:
Abdelrazak Younes wrote:
Jean-Marc Lasgouttes wrote:
"Abdelrazak" == Abdelrazak Younes
<[EMAIL PROTECTED]> writes:
Abdelrazak> And got rid of this. So if this profile is correct, the
Abdelrazak> attached patch will save more than 8% time at file
Abdelrazak> reading. Could
On Thu, Jan 11, 2007 at 03:11:31PM +0100, Abdelrazak Younes wrote:
> Indeed. On cygwin there's one second improvement (from 7s to 6s) when
> loading a really big file (4megs). I repeated the test multiple times to
> make sure that the file was already in the cache.
>
> But I do not trust
Abdelrazak Younes wrote:
> You are a complicated guy Georg. I was answering to your false
> assumption so, in this regard, you misread the patch. I cannot read your
> mind and understand what you don't like.
I might be complicated, but I still do not think that I misread anything. I
only tried
Edwin == Edwin Leuven [EMAIL PROTECTED] writes:
Edwin Abdelrazak Younes wrote:
Could you profile this instead: lyx -e text UserGuide.lyx
Edwin another try:
Since you are using oprofile, you could try to obtain a call graph
instead of a flat profile. This is possible if your linux is new
Jean-Marc Lasgouttes wrote:
Edwin == Edwin Leuven [EMAIL PROTECTED] writes:
Edwin Abdelrazak Younes wrote:
Could you profile this instead: lyx -e text UserGuide.lyx
Edwin another try:
Since you are using oprofile, you could try to obtain a call graph
instead of a flat profile. This is
Edwin == Edwin Leuven [EMAIL PROTECTED] writes:
Edwin Jean-Marc Lasgouttes wrote:
Edwin == Edwin Leuven [EMAIL PROTECTED] writes:
Edwin Abdelrazak Younes wrote:
Could you profile this instead: lyx -e text UserGuide.lyx
Edwin another try:
Since you are using oprofile, you could try to
> "Edwin" == Edwin Leuven <[EMAIL PROTECTED]> writes:
Edwin> Abdelrazak Younes wrote:
>> Could you profile this instead: lyx -e text UserGuide.lyx
Edwin> another try:
Since you are using oprofile, you could try to obtain a call graph
instead of a flat profile. This is possible if your linux
Jean-Marc Lasgouttes wrote:
"Edwin" == Edwin Leuven <[EMAIL PROTECTED]> writes:
Edwin> Abdelrazak Younes wrote:
Could you profile this instead: lyx -e text UserGuide.lyx
Edwin> another try:
Since you are using oprofile, you could try to obtain a call graph
instead of a flat profile. This
> "Edwin" == Edwin Leuven <[EMAIL PROTECTED]> writes:
Edwin> Jean-Marc Lasgouttes wrote:
>>> "Edwin" == Edwin Leuven <[EMAIL PROTECTED]> writes:
>>
Edwin> Abdelrazak Younes wrote:
Could you profile this instead: lyx -e text UserGuide.lyx
>>
Edwin> another try:
>> Since you are using
Abdelrazak Younes wrote:
Edwin Leuven wrote:
Abdelrazak Younes wrote:
Could you profile this instead:
lyx -e text UserGuide.lyx
then you get this:
Each sample counts as 0.01 seconds.
% cumulative self self total
time seconds secondscalls ms/call ms/call
Abdelrazak Younes wrote:
Could you profile this instead:
lyx -e text UserGuide.lyx
another try:
% cumulative self self total
time seconds secondscalls ms/call ms/call name
16.00 0.04 0.0424961 0.00 0.00
lyx::LyXLex::Pimpl::nextToken()
although loading and saving feels snappier, i find that selecting text
(especially in a large doc like the userguide) is slow on linux, the
selection lags a bit behind the mouse when i move the mouse quickly.
if i just change the selection (with the mouse) in a single paragraph
for a while
Abdelrazak Younes wrote:
Edwin Leuven wrote:
Abdelrazak Younes wrote:
Could you profile this instead:
lyx -e text UserGuide.lyx
then you get this:
Each sample counts as 0.01 seconds.
% cumulative self self total
time seconds secondscalls ms/call ms/call
Abdelrazak Younes wrote:
Could you profile this instead:
lyx -e text UserGuide.lyx
another try:
% cumulative self self total
time seconds secondscalls ms/call ms/call name
16.00 0.04 0.0424961 0.00 0.00
lyx::LyXLex::Pimpl::nextToken()
although loading and saving feels snappier, i find that selecting text
(especially in a large doc like the userguide) is slow on linux, the
selection lags a bit behind the mouse when i move the mouse quickly.
if i just change the selection (with the mouse) in a single paragraph
for a while
Abdelrazak == Abdelrazak Younes [EMAIL PROTECTED] writes:
% cumulative self self total time seconds seconds calls ms/call
ms/call name 35.43 0.79 0.79 2002198 0.00 0.00
lyx::Paragraph::getFontSettings(lyx::BufferParams const, int)
const
Abdelrazak Ouch, 35%! I am sure this could be
Ouch, 35%! I am sure this could be optimized!
wrong profile. correct one here:
http://article.gmane.org/gmane.editors.lyx.devel/75394
Edwin Leuven wrote:
Michael Gerz wrote:
Edwin,
could you please check whether the attached patch saves any time in
lyx::Changes::lookup(int) const and lyx::Changes::merge() during
opening/saving a file?
this is open (save below):
Each sample counts as 0.01 seconds.
% cumulative
copy/paste (using the clipboard) from konsole ;-)
file open:
Each sample counts as 0.01 seconds.
% cumulative self self total
time seconds secondscalls ms/call ms/call name
9.09 0.02 0.02 184108 0.00 0.00
lyx::Paragraph::setFont(int,
Edwin Leuven wrote:
copy/paste (using the clipboard) from konsole ;-)
Thanks Edwin but I suspect that those number are not very meaningful
because of the Gui loading etc.
Could you profile this instead:
lyx -e text UserGuide.lyx
or Any other big file that you have.
Abdel.
Abdelrazak Younes [EMAIL PROTECTED] writes:
Each sample counts as 0.01 seconds.
% cumulative self self total
time seconds secondscalls ms/call ms/call name
7.41 0.02 0.02 268500 0.00 0.00
lyx::InsetBase::inMathed() const
I don't
Edwin,
is there any need to further optimize things? I still see some change
tracking-related stuff on top of the reports but I wonder whether it
really matters...
Michael
Abdelrazak Younes wrote:
Could you profile this instead:
lyx -e text UserGuide.lyx
then you get this:
Each sample counts as 0.01 seconds.
% cumulative self self total
time seconds secondscalls ms/call ms/call name
100.00 0.01 0.01110.00
Edwin Leuven wrote:
Abdelrazak Younes wrote:
Could you profile this instead:
lyx -e text UserGuide.lyx
then you get this:
Each sample counts as 0.01 seconds.
% cumulative self self total
time seconds secondscalls ms/call ms/call name
100.00 0.01
> "Abdelrazak" == Abdelrazak Younes <[EMAIL PROTECTED]> writes:
>> % cumulative self self total time seconds seconds calls ms/call
>> ms/call name 35.43 0.79 0.79 2002198 0.00 0.00
>> lyx::Paragraph::getFontSettings(lyx::BufferParams const&, int)
>> const
Abdelrazak> Ouch, 35%! I am sure
> Ouch, 35%! I am sure this could be optimized!
wrong profile. correct one here:
http://article.gmane.org/gmane.editors.lyx.devel/75394
Edwin Leuven wrote:
Michael Gerz wrote:
Edwin,
could you please check whether the attached patch saves any time in
lyx::Changes::lookup(int) const and lyx::Changes::merge() during
opening/saving a file?
this is open (save below):
Each sample counts as 0.01 seconds.
% cumulative
copy/paste (using the clipboard) from konsole ;-)
file open:
Each sample counts as 0.01 seconds.
% cumulative self self total
time seconds secondscalls ms/call ms/call name
9.09 0.02 0.02 184108 0.00 0.00
lyx::Paragraph::setFont(int,
Edwin Leuven wrote:
copy/paste (using the clipboard) from konsole ;-)
Thanks Edwin but I suspect that those number are not very meaningful
because of the Gui loading etc.
Could you profile this instead:
lyx -e text UserGuide.lyx
or Any other big file that you have.
Abdel.
Abdelrazak Younes <[EMAIL PROTECTED]> writes:
> > Each sample counts as 0.01 seconds.
> > % cumulative self self total
> > time seconds secondscalls ms/call ms/call name
> > 7.41 0.02 0.02 268500 0.00 0.00
> > lyx::InsetBase::inMathed()
Edwin,
is there any need to further optimize things? I still see some change
tracking-related stuff on top of the reports but I wonder whether it
really matters...
Michael
Abdelrazak Younes wrote:
Could you profile this instead:
lyx -e text UserGuide.lyx
then you get this:
Each sample counts as 0.01 seconds.
% cumulative self self total
time seconds secondscalls ms/call ms/call name
100.00 0.01 0.01110.00
Edwin Leuven wrote:
Abdelrazak Younes wrote:
Could you profile this instead:
lyx -e text UserGuide.lyx
then you get this:
Each sample counts as 0.01 seconds.
% cumulative self self total
time seconds secondscalls ms/call ms/call name
100.00 0.01
the following is the profile i get when opening the user guide:
(full profile here: http://leuven.ecodip.net/lyx/profile.zip)
was wondering why lyx::Changes::set(lyx::Change const, int, int)
is called...
Flat profile:
Each sample counts as 0.01 seconds.
% cumulative self
Edwin Leuven schrieb:
the following is the profile i get when opening the user guide:
(full profile here: http://leuven.ecodip.net/lyx/profile.zip)
was wondering why lyx::Changes::set(lyx::Change const, int, int)
is called...
Flat profile:
Each sample counts as 0.01 seconds.
%
Michael Gerz wrote:
was wondering why lyx::Changes::set(lyx::Change const, int, int)
is called...
Every time you insert a character, the corresponding change has to be
set (even if it is unchanged, because it may be inserted in a region of
deleted/inserted text).
i was expecting that
Edwin Leuven schrieb:
Michael Gerz wrote:
was wondering why lyx::Changes::set(lyx::Change const, int, int)
is called...
Every time you insert a character, the corresponding change has to be
set (even if it is unchanged, because it may be inserted in a region
of deleted/inserted text).
i
Michael Gerz wrote:
With optimizations or in debugging mode?
aargh! it is compiling with --enable-optimization=-O2 now. will report
back...
Edwin Leuven wrote:
Michael Gerz wrote:
With optimizations or in debugging mode?
aargh! it is compiling with --enable-optimization=-O2 now. will report
opening and saving the userguide takes about 1 sec
opening profile:
% cumulative self self total
time seconds
Edwin,
could you please check whether the attached patch saves any time in
lyx::Changes::lookup(int) const and lyx::Changes::merge() during
opening/saving a file?
Michael
Index: changes.C
===
--- changes.C (Revision 16573)
Michael Gerz wrote:
Edwin,
could you please check whether the attached patch saves any time in
lyx::Changes::lookup(int) const and lyx::Changes::merge() during
opening/saving a file?
this is open (save below):
Each sample counts as 0.01 seconds.
% cumulative self self
Edwin Leuven schrieb:
Michael Gerz wrote:
Edwin,
could you please check whether the attached patch saves any time in
lyx::Changes::lookup(int) const and lyx::Changes::merge() during
opening/saving a file?
this is open (save below):
Each sample counts as 0.01 seconds.
% cumulative
Michael Gerz wrote:
Edwin Leuven schrieb:
Michael Gerz wrote:
Edwin,
could you please check whether the attached patch saves any time in
lyx::Changes::lookup(int) const and lyx::Changes::merge() during
opening/saving a file?
this is open (save below):
In other words: The speed-up is
Am Sonntag, 7. Januar 2007 19:42 schrieb Edwin Leuven:
Michael Gerz wrote:
In other words: The speed-up is signficant, right?
yeah, i'd suggest you put it in...
If you do, please add a comment to the inlined functions that they are
inlined because of profiler measurements.
Georg
Edwin Leuven wrote:
Edwin Leuven wrote:
Michael Gerz wrote:
With optimizations or in debugging mode?
aargh! it is compiling with --enable-optimization=-O2 now. will report
What's that? A CMake settings? Or is this linux with oprofile?
opening and saving the userguide takes about 1 sec
Abdelrazak Younes wrote:
is this linux with oprofile?
yeah, mon péché mignon... ;-)
the following is the profile i get when opening the user guide:
(full profile here: http://leuven.ecodip.net/lyx/profile.zip)
was wondering why lyx::Changes::set(lyx::Change const&, int, int)
is called...
Flat profile:
Each sample counts as 0.01 seconds.
% cumulative self
Edwin Leuven schrieb:
the following is the profile i get when opening the user guide:
(full profile here: http://leuven.ecodip.net/lyx/profile.zip)
was wondering why lyx::Changes::set(lyx::Change const&, int, int)
is called...
Flat profile:
Each sample counts as 0.01 seconds.
%
Michael Gerz wrote:
was wondering why lyx::Changes::set(lyx::Change const&, int, int)
is called...
Every time you insert a character, the corresponding change has to be
set (even if it is unchanged, because it may be inserted in a region of
deleted/inserted text).
i was expecting that
Edwin Leuven schrieb:
Michael Gerz wrote:
was wondering why lyx::Changes::set(lyx::Change const&, int, int)
is called...
Every time you insert a character, the corresponding change has to be
set (even if it is unchanged, because it may be inserted in a region
of deleted/inserted text).
i
Michael Gerz wrote:
With optimizations or in debugging mode?
aargh! it is compiling with --enable-optimization=-O2 now. will report
back...
Edwin Leuven wrote:
Michael Gerz wrote:
With optimizations or in debugging mode?
aargh! it is compiling with --enable-optimization=-O2 now. will report
opening and saving the userguide takes about 1 sec
opening profile:
% cumulative self self total
time seconds
Edwin,
could you please check whether the attached patch saves any time in
lyx::Changes::lookup(int) const and lyx::Changes::merge() during
opening/saving a file?
Michael
Index: changes.C
===
--- changes.C (Revision 16573)
Michael Gerz wrote:
Edwin,
could you please check whether the attached patch saves any time in
lyx::Changes::lookup(int) const and lyx::Changes::merge() during
opening/saving a file?
this is open (save below):
Each sample counts as 0.01 seconds.
% cumulative self self
Edwin Leuven schrieb:
Michael Gerz wrote:
Edwin,
could you please check whether the attached patch saves any time in
lyx::Changes::lookup(int) const and lyx::Changes::merge() during
opening/saving a file?
this is open (save below):
Each sample counts as 0.01 seconds.
% cumulative
Michael Gerz wrote:
Edwin Leuven schrieb:
Michael Gerz wrote:
Edwin,
could you please check whether the attached patch saves any time in
lyx::Changes::lookup(int) const and lyx::Changes::merge() during
opening/saving a file?
this is open (save below):
>
In other words: The speed-up is
Am Sonntag, 7. Januar 2007 19:42 schrieb Edwin Leuven:
> Michael Gerz wrote:
> > In other words: The speed-up is signficant, right?
>
> yeah, i'd suggest you put it in...
If you do, please add a comment to the inlined functions that they are
inlined because of profiler measurements.
Georg
Edwin Leuven wrote:
Edwin Leuven wrote:
Michael Gerz wrote:
With optimizations or in debugging mode?
aargh! it is compiling with --enable-optimization=-O2 now. will report
What's that? A CMake settings? Or is this linux with oprofile?
opening and saving the userguide takes about 1 sec
Abdelrazak Younes wrote:
is this linux with oprofile?
yeah, mon péché mignon... ;-)
70 matches
Mail list logo
