![]() |
|
|
#13 | |
|
Registered User
Join Date: Dec 2007
Posts: 63
|
Quote:
/usr/bin/nvidia-settings -a InitialPixmapPlacement=0 -a GlyphCache=1 |
|
|
|
|
|
|
#14 |
|
Registered User
Join Date: Feb 2008
Posts: 5
|
I'm using 6600gt (so, no GlyphCache) and found what FF3 is much slower that FF2 on this page: http://fastra.ua.ac.be/en/specs.html .
I think this is nvidia problem because: 1. Users with integrated intel video have no problems at all (yes, I'm awaiting Larrabee) 2. CPU being eaten by X server, not firefox 3. Seems what windows users have no such problem too |
|
|
|
| Sponsored Ads - Guests Only | |
|
|
|
|
#15 |
|
Registered User
Join Date: Sep 2006
Posts: 2
|
On my machine (Geforce 8600GT, 171.06 drivers), scrolling on that example page (http://shame.tuxfamily.org/repo/?cat=1) makes the X server CPU usage go up a lot and everything becomes fairly unresponsive. Setting to InitialPixmapPlacement=0 with Geforce 8600GT makes it less slow, but Firefox itself is then slow with it instead. The X server doesn't do anything abnormal or stop responding in that case. It seems like Firefox must be exposing a regression in the driver's logic in that case.
|
|
|
|
|
|
#16 | |
|
Registered User
Join Date: Feb 2004
Posts: 585
|
Quote:
lg Clemens |
|
|
|
|
|
|
#17 |
|
Registered User
Join Date: Jun 2008
Posts: 7
|
After a little chat with firefox dev, they told me the main difference between FF2 and FF3 for the rendering speed is Xrender, which normaly boost the rendering engine, but can have side effects with some cards.
I decided to run a benchmark of NVIDIA drivers with the xrenderbenchmark, the results are here : => http://www.larsen-b.com/Article/290.html As you can figure, the NVIDIA drivers are pretty slow when it has to deal with large pixels area .. |
|
|
|
|
|
#18 | |
|
Registered User
Join Date: May 2008
Posts: 19
|
Quote:
|
|
|
|
|
|
|
#19 | |
|
Registered User
Join Date: Feb 2008
Posts: 272
|
Quote:
in the second benchmark nvidia is slow because it is not accelerated by the gpu. According to nvidia these are not commonly used xrender operations therefore they don't bother accelerating them (but probably the reason 2d on nvidia sucks). Seeing as an ati card with can perform nearly as well as an intel card then frankly I don't see much of a reason to stay with nvidia in the future. Now I just need to find someone with an ati 2xxx series to see results like those with my own eyes. EDIT: Just read the specs on the test system, don't tell me a Q6600 was used on card and an E6600 on another...which would completely invalidate the results |
|
|
|
|
|
|
#20 |
|
Registered User
Join Date: Oct 2007
Posts: 14
|
I just installed FF 3.0 on an Opensuse 10.3 system. I use a Geforce 7800 GTX with Nvidias driver 173.14.05. (KDE version 3.5.9.9).
I do not see any performance problems with FF 3.0 and scrolling large pages. I tried e.g. the page "http://yro.slashdot.org/yro/08/06/15/1258238.shtml" which was mentioned in the contribution of "Blg P" above. On the contrary: FF 3.0 seems to be significantly faster than FF 2.0. |
|
|
|
|
|
#21 |
|
Registered User
Join Date: May 2008
Posts: 74
|
my firefox had ui freezing issues until i updated xulrunner - perhaps this isn't an nvidia issue?
|
|
|
|
|
|
#22 |
|
Registered User
Join Date: May 2004
Posts: 663
|
|
|
|
|
|
|
#23 | |
|
Registered User
Join Date: Dec 2007
Posts: 63
|
Quote:
/usr/bin/nvidia-settings -a InitialPixmapPlacement=2 -a GlyphCache=1 It is little faster on some nvidia cards (but not on GF8xxx) with: /usr/bin/nvidia-settings -a InitialPixmapPlacement=0 -a GlyphCache=1 |
|
|
|
|
|
|
#24 | |
|
Registered User
Join Date: Jun 2008
Posts: 7
|
Quote:
The Intel benchmark was run on a E6600, but the ATI and Nvidia one were both run on a Q6600 (same mobo, ram, or distro) ... (Intel only sell GMA on mobo, so I can't benchmark it on the same system, that's it.) And remember Q6600 and E6600 are quite the same, only with more core. but multi core aren't used here. Bye .. |
|
|
|
|
![]() |
| Most Popular NVIDIA Based Graphics Cards | |
|
|
| Currently Active Users Viewing This Thread: 1 (0 members and 1 guests) | |
| Thread Tools | |
|
|