Re: [Dri-devel] ATI Radeon 7500 Performance Issues

2002-06-25 Thread matt . nottingham

Michel =?ISO-8859-1?Q?D=E4nzer?= writes:
  On Sat, 2002-06-22 at 14:30, [EMAIL PROTECTED] wrote:
   One thing I noticed is that I did not have EnablePageFlip set to true
   in my X config file. Setting this raised the fps to 660. Is there any
   reason why this isn't the default?
  
  There are issues with it, e.g. when obscuring a 3D window.

OK

   That was with 3 xterms (2 of which were doing nothing, one running
   glxgears)  xemacs with no files loaded. Window manager was afterstep.
   So little 2D activity.
  
  Hmm. I'm running out of ideas then...

I was just looking through the files and came across this:

Jun 23 20:54:44 trinity kernel: [drm:radeon_freelist_get] *ERROR* returning NULL
!
Jun 23 20:55:45 trinity last message repeated 638 times
Jun 23 20:56:46 trinity last message repeated 195 times
Jun 23 20:57:49 trinity last message repeated 366 times
Jun 23 20:58:50 trinity last message repeated 354 times

But this doesn't seem to be generated when I do a glxgears, only
bzflag. Not sure what it means!

   Crystal space's walktest instantly dies with a
   
   walktest: t_imm_api.c:316: _tnl_end: Assertion `ctx-Driver.NeedFlush  0x1' 
 failed.
  
  I also see that.

I've reported it to the Crystal Space mailing list but have yet to get
anything back.

Thanks,

Matt


---
This sf.net email is sponsored by: Jabber Inc.
Don't miss the IM event of the season | Special offer for OSDN members! 
JabConf 2002, Aug. 20-22, Keystone, CO http://www.jabberconf.com/osdn
___
Dri-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/dri-devel



Re: [Dri-devel] ATI Radeon 7500 Performance Issues

2002-06-25 Thread Keith Whitwell


 Jun 23 20:55:45 trinity last message repeated 638 times
 Jun 23 20:56:46 trinity last message repeated 195 times
 Jun 23 20:57:49 trinity last message repeated 366 times
 Jun 23 20:58:50 trinity last message repeated 354 times
 
 But this doesn't seem to be generated when I do a glxgears, only
 bzflag. Not sure what it means!

The driver is running out of dma buffers - this might be normal, or not.

Crystal space's walktest instantly dies with a

walktest: t_imm_api.c:316: _tnl_end: Assertion `ctx-Driver.NeedFlush  0x1' 
failed.
   
   I also see that.
 
 I've reported it to the Crystal Space mailing list but have yet to get
 anything back.

This is definitely a bug in the driver, not a crystal space problem.  If you 
run with 'RADEON_NO_VTXFMT=t' it should go away.  I don't have time to track 
it down right now.

Keith



---
This sf.net email is sponsored by: Jabber Inc.
Don't miss the IM event of the season | Special offer for OSDN members! 
JabConf 2002, Aug. 20-22, Keystone, CO http://www.jabberconf.com/osdn
___
Dri-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/dri-devel



Re: [Dri-devel] ATI Radeon 7500 Performance Issues

2002-06-24 Thread Michel Dänzer

On Sat, 2002-06-22 at 14:30, [EMAIL PROTECTED] wrote:
 Michel =?ISO-8859-1?Q?D=E4nzer?= writes:
   On Thu, 2002-06-20 at 18:43, [EMAIL PROTECTED] wrote: 
=?ISO-8859-1?Q?D=E4nzer?= writes:
ed, 2002-06-19 at 16:59, Adam K Kirchhoff wrote: 

 19 Jun 2002, Michel [ISO-8859-1] Dänzer wrote:

On Wed, 2002-06-19 at 16:29, Malverian ... wrote:
 I'm having some pretty serious speed issues with my ATI Radeon 7500 64MB 
 DDR. And my hopes are in you guys for helping me iron out the kinks.
 
 glxinfo says that direct rendering is enabled
 X startup has no warnings or errors
 
 I'm getting 500FPS in glxgears

That's exactly the speed I get on my XP 1800+ So we both must be
making the same mistake or something.
   
   Make really sure you're using all the latest components.
 
 Looking at the output of ldd on `which glxgear` and checking their
 creation times showed that they are all what I'd expect. Also
 /usr/X11R6/lib/modules/{dri|drivers|extensions} also have the creation
 time I'd expect. Is there anything I should be looking for that I've
 missed?

No, sounds good.

 One thing I noticed is that I did not have EnablePageFlip set to true
 in my X config file. Setting this raised the fps to 660. Is there any
 reason why this isn't the default?

There are issues with it, e.g. when obscuring a 3D window.

   Also, 2D
   clients will still have an impact on 3D performance, although it should
   no longer be as bad as it used to be.
 
 That was with 3 xterms (2 of which were doing nothing, one running
 glxgears)  xemacs with no files loaded. Window manager was afterstep.
 So little 2D activity.

Hmm. I'm running out of ideas then...


   Anyhow, glxgears really isn't an important benchmark, how do 'real' apps
   perform?
 
 True, but it seems an easy way just to compare 'like' systems.

Only if the desktop environments are 'like' as well.


 Crystal space's walktest instantly dies with a
 
 walktest: t_imm_api.c:316: _tnl_end: Assertion `ctx-Driver.NeedFlush  0x1' failed.

I also see that.


-- 
Earthling Michel Dänzer (MrCooper)/ Debian GNU/Linux (powerpc) developer
XFree86 and DRI project member   /  CS student, Free Software enthusiast


---
Sponsored by:
ThinkGeek at http://www.ThinkGeek.com/
___
Dri-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/dri-devel



Re: [Dri-devel] ATI Radeon 7500 Performance Issues

2002-06-23 Thread Tim Smith

On Saturday 22 Jun 2002 1:30 pm, [EMAIL PROTECTED] scribed 
numinously:
 Michel =?ISO-8859-1?Q?D=E4nzer?= writes:
   On Thu, 2002-06-20 at 18:43, [EMAIL PROTECTED] wrote:
=?ISO-8859-1?Q?D=E4nzer?= writes:
ed, 2002-06-19 at 16:59, Adam K Kirchhoff wrote:
   
 19 Jun 2002, Michel [ISO-8859-1] Dänzer wrote:
   
On Wed, 2002-06-19 at 16:29, Malverian ... wrote:
 I'm having some pretty serious speed issues with my ATI Radeon
 7500 64MB DDR. And my hopes are in you guys for helping me iron
 out the kinks.

 glxinfo says that direct rendering is enabled
 X startup has no warnings or errors

 I'm getting 500FPS in glxgears
   
That's exactly the speed I get on my XP 1800+ So we both must be
making the same mistake or something.
  
   Make really sure you're using all the latest components.

 Looking at the output of ldd on `which glxgear` and checking their
 creation times showed that they are all what I'd expect. Also
 /usr/X11R6/lib/modules/{dri|drivers|extensions} also have the creation
 time I'd expect. Is there anything I should be looking for that I've
 missed?

 One thing I noticed is that I did not have EnablePageFlip set to true
 in my X config file. Setting this raised the fps to 660. Is there any
 reason why this isn't the default?

   Also, 2D
   clients will still have an impact on 3D performance, although it
   should no longer be as bad as it used to be.

 That was with 3 xterms (2 of which were doing nothing, one running
 glxgears)  xemacs with no files loaded. Window manager was afterstep.
 So little 2D activity.

I also get about 500fps with glxgears on my Radeon 7500 64MB SDR card on an 
Athlon 1200. Similarly, enabling page flipping raises the rate to around 
600fps. I'm running with the latest CVS HEAD.

I generally run with page flipping disabled, because when it's enabled I get 
jumpy animation and the appearance of the gears appearing to wobble in 
place.

During my tests, I am running in 1024x768 24bit, with no other apps (not 
even a window manager)

-- 
Tim Smith ([EMAIL PROTECTED])
It's the frost. Sometimes it makes the blade stick.



---
Sponsored by:
ThinkGeek at http://www.ThinkGeek.com/
___
Dri-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/dri-devel



Re: [Dri-devel] ATI Radeon 7500 Performance Issues

2002-06-23 Thread matt . nottingham



  Anyhow, glxgears really isn't an important benchmark, how do 'real' =
apps
  perform?

OK, found the 't' key in bzflag - I get around 35 fps on my 7500 DDR,
1800+ Athlon, A7M266-D motherboard. I seem to remember someone saying
they were getting 150, but I can't remember what card/cpu they had.

I'll look into running the Mesa demo's this afternoon.

Thanks,

Matt
 


---
Sponsored by:
ThinkGeek at http://www.ThinkGeek.com/
___
Dri-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/dri-devel



Re: [Dri-devel] ATI Radeon 7500 Performance Issues

2002-06-23 Thread Michel Dänzer

On Sun, 2002-06-23 at 12:45, [EMAIL PROTECTED] wrote:
 
 
   Anyhow, glxgears really isn't an important benchmark, how do 'real' =
 apps
   perform?
 
 OK, found the 't' key in bzflag - I get around 35 fps on my 7500 DDR,
 1800+ Athlon, A7M266-D motherboard. I seem to remember someone saying
 they were getting 150, but I can't remember what card/cpu they had.

Could have been me. It gives me around 150 fps in 1280x854x32 on this
TiBook/667 with an M7 aka 7500 Mobility, and around 200 fps in
1280x960x32 on an Athlon 900 box at work with a 7500. That's with all
detail settings at max, 'except' dithering is off. I guess the extra 50
fps are due to bandwidth between chip and VRAM and/or the size of the
latter.


-- 
Earthling Michel Dänzer (MrCooper)/ Debian GNU/Linux (powerpc) developer
XFree86 and DRI project member   /  CS student, Free Software enthusiast


---
Sponsored by:
ThinkGeek at http://www.ThinkGeek.com/
___
Dri-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/dri-devel



Re: [Dri-devel] ATI Radeon 7500 Performance Issues

2002-06-23 Thread matt . nottingham

Michel =?ISO-8859-1?Q?D=E4nzer?= writes:
  On Sun, 2002-06-23 at 12:45, [EMAIL PROTECTED] wrote:
   
   
 Anyhow, glxgears really isn't an important benchmark, how do 'real' =
   apps
 perform?
   
   OK, found the 't' key in bzflag - I get around 35 fps on my 7500 DDR,
   1800+ Athlon, A7M266-D motherboard. I seem to remember someone saying
   they were getting 150, but I can't remember what card/cpu they had.
  
  Could have been me. It gives me around 150 fps in 1280x854x32 on this
  TiBook/667 with an M7 aka 7500 Mobility, and around 200 fps in
  1280x960x32 on an Athlon 900 box at work with a 7500. That's with all
  detail settings at max, 'except' dithering is off. I guess the extra 50
  fps are due to bandwidth between chip and VRAM and/or the size of the
  latter.
  

Just to say that the 35 fps is at 1280x1024x32. Turning dithering on
or off makes very little difference to the frame rate. All others are
at max.

Thanks,

Matt
 


---
Sponsored by:
ThinkGeek at http://www.ThinkGeek.com/
___
Dri-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/dri-devel



Re: [Dri-devel] ATI Radeon 7500 Performance Issues

2002-06-23 Thread Michel Dänzer

On Sun, 2002-06-23 at 15:44, [EMAIL PROTECTED] wrote:
 Michel =?ISO-8859-1?Q?D=E4nzer?= writes:
   On Sun, 2002-06-23 at 12:45, [EMAIL PROTECTED] wrote:


  Anyhow, glxgears really isn't an important benchmark, how do 'real' =
apps
  perform?

OK, found the 't' key in bzflag - I get around 35 fps on my 7500 DDR,
1800+ Athlon, A7M266-D motherboard. I seem to remember someone saying
they were getting 150, but I can't remember what card/cpu they had.
   
   Could have been me. It gives me around 150 fps in 1280x854x32 on this
   TiBook/667 with an M7 aka 7500 Mobility, and around 200 fps in
   1280x960x32 on an Athlon 900 box at work with a 7500. That's with all
   detail settings at max, 'except' dithering is off. I guess the extra 50
   fps are due to bandwidth between chip and VRAM and/or the size of the
   latter.
   
 
 Just to say that the 35 fps is at 1280x1024x32. Turning dithering on
 or off makes very little difference to the frame rate. All others are
 at max.

One additional note: those numbers were with bzflag 1.7e4. 1.7e6 only
gives around 100 fps on the TiBook. :/

Also, the numbers are from a simple map. On a more complicated map, when
lots of stuff like buildings with lots of walls is visible, the
framerate is significantly lower.


-- 
Earthling Michel Dänzer (MrCooper)/ Debian GNU/Linux (powerpc) developer
XFree86 and DRI project member   /  CS student, Free Software enthusiast


---
Sponsored by:
ThinkGeek at http://www.ThinkGeek.com/
___
Dri-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/dri-devel



Re: [Dri-devel] ATI Radeon 7500 Performance Issues

2002-06-22 Thread matt . nottingham

Michel =?ISO-8859-1?Q?D=E4nzer?= writes:
  On Thu, 2002-06-20 at 18:43, [EMAIL PROTECTED] wrote: 
   =?ISO-8859-1?Q?D=E4nzer?= writes:
   ed, 2002-06-19 at 16:59, Adam K Kirchhoff wrote: 
   
19 Jun 2002, Michel [ISO-8859-1] Dänzer wrote:
   
   On Wed, 2002-06-19 at 16:29, Malverian ... wrote:
I'm having some pretty serious speed issues with my ATI Radeon 7500 64MB 
DDR. And my hopes are in you guys for helping me iron out the kinks.

glxinfo says that direct rendering is enabled
X startup has no warnings or errors

I'm getting 500FPS in glxgears
   
   That's exactly the speed I get on my XP 1800+ So we both must be
   making the same mistake or something.
  
  Make really sure you're using all the latest components.

Looking at the output of ldd on `which glxgear` and checking their
creation times showed that they are all what I'd expect. Also
/usr/X11R6/lib/modules/{dri|drivers|extensions} also have the creation
time I'd expect. Is there anything I should be looking for that I've
missed?

One thing I noticed is that I did not have EnablePageFlip set to true
in my X config file. Setting this raised the fps to 660. Is there any
reason why this isn't the default?

  Also, 2D
  clients will still have an impact on 3D performance, although it should
  no longer be as bad as it used to be.

That was with 3 xterms (2 of which were doing nothing, one running
glxgears)  xemacs with no files loaded. Window manager was afterstep.
So little 2D activity.

  Anyhow, glxgears really isn't an important benchmark, how do 'real' apps
  perform?

True, but it seems an easy way just to compare 'like' systems.

Now that I've enable the page flipping, 'Emilia pinball' is now
playable in a window size 1024x768 which it wasn't before. 

bzflag plays well, but I haven't figure out yet how to make it give an
fps.

Crystal space's walktest instantly dies with a

walktest: t_imm_api.c:316: _tnl_end: Assertion `ctx-Driver.NeedFlush  0x1' failed.

It didn't used to before I used the TCL branch (now also main) branch.

  -- 
  Earthling Michel Dänzer (MrCooper)/ Debian GNU/Linux (powerpc) developer
  XFree86 and DRI project member   /  CS student, Free Software enthusiast


Thanks,

Matt


---
Sponsored by:
ThinkGeek at http://www.ThinkGeek.com/
___
Dri-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/dri-devel