mitea | 1 Jun 2003 01:44
Picon

nvidia cheated in 3dmark2003?

 
----------------------------------
Peace and love,
Tweety
YahooID: tweety_04_01
 
Dominic Curran | 1 Jun 2003 11:19
Picon
Favicon

WHERE TO FIND OPENGL INFORMATION - (Bi-weekly posting)

WHERE TO FIND OPENGL INFORMATION - (Bi-weekly posting)

The Official OpenGL Site - News, downloads, tutorials, books & links:-
http://www.opengl.org/

The archive for this mailing list can be found at:-
http://www.egroups.com/list/opengl-gamedev-l/

The OpenGL Repository - Downloads for OpenGL implementations:-
http://www.fortunecity.com/skyscraper/nuclear/274/

The OpenGL GameDev FAQ:-
http://www.3dgamedev.com/resources/openglfaq.txt

The EFnet OpenGL FAQ:-
http://www.geocities.com/SiliconValley/Park/5625/opengl/

The Omniverous Biped's FAQ:-
http://www.sjbaker.org/steve/omniv

OpenGL 1.0 Reference
http://www.cevis.uni-bremen.de/~uwe/opengl/

OpenGL 1.1 Reference - This is pretty much the Blue book on-line:-
http://tc1.chemie.uni-bielefeld.de/doc/OpenGL/hp/Reference.html

Information on the GLUT API:-
http://www.opengl.org/developers/documentation/glut.html

Apples OpenGL web site is at:-
http://developer.apple.com/opengl/

The Mesa 3-D graphics library:-
http://www.mesa3d.org

SGI OpenGL Sample Implementation (downloadable source):-
http://oss.sgi.com/projects/ogl-sample/

Some nice OpenGL tutorials for beginners:-
http://nehe.gamedev.net/

Nate Robins OpenGL Page (some tutorials and code)
http://www.xmission.com/~nate/opengl.html

The Charter for this mailing list can be found at OpenGl GameDev Site:-
http://www.geocities.com/SiliconValley/Hills/9956/OpenGL/

The GLSetup project page is at:-
http://www.glsetup.com/

OpenGL Usenet groups:-
news:comp.graphics.api.opengl 

This is a bi-weekly posting to OPENGL-GAMEDEV-L.

If you believe an amendment should be made to this posting please
contact Dominic at dominic_curran <at> yahoo.com
----- 
FAQ and OpenGL Resources at:
  http://www.geocities.com/SiliconValley/Hills/9956/OpenGL

--

-- 
Author: Dominic Curran
  INET: dcurran <at> ti.com

Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
San Diego, California        -- Mailing list and web hosting services
---------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru <at> fatcity.com (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB OPENGL-GAMEDEV-L
(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing).

Colin Fahey | 2 Jun 2003 04:50
Picon
Favicon

Re: nvidia cheated in 3dmark2003?

>>>Subject: nvidia cheated in 3dmark2003?
>>>> http://www.futuremark.com/companyinfo/3dmark03_audit_report.pdf
>>>>

I like this footnote on page 1:

  "[3] Futuremark's BETA program is an open, fee based cooperation program
[...]"
                                       ----  --------- -------------------

No wonder nVidia left the BETA program!

That article could benefit from a little more science.  For example, the
claims
regarding pixel shader "detection" are implied, not proven.  If Futuremark
quoted byte scans of the driver binary (e.g., nvidia.sys or whatever it
might
be) that showed that the pixel shader code or name was explicitly a search
pattern used by the driver, it would be convincing, and it's a test anyone
could reproduce.

It *is* possible, if perhaps improbable, the drivers are detecting general
scenarios that can be optimized, so it's possible that the Futuremark shader
scenarios are actually "detected" by pure coincidence.

The irony is that the existence of an artificial benchmark makes optimizing
for artificial scenarios tempting, whereas if there were only major
applications
and games, consumers might see more optimizations that make a practical
difference in their actual lives!  I could just imagine if nVidia crammed an
optimized Quake 3 clone in to their drivers!  Or if nVidia changed the
microchip itself to have MD3 model buffers and animation paths.  All assets
for all levels would be in firmware.  Drawing the weapon crosshairs would
be a hardwired stage of the rendering pipeline, along with the health meter
and "Major's" face icon.  Similarly, the audio clip "Quad Damage" would be
part of nVidia's sound chip.  The nVidia bridge chip would have the Quake 3
application in silicon, so the request to run "quake3.exe" would be
"detected"
and serviced without ever touching the disk!  (Making minor changes to
"quake3.exe" might lead to a 24.1% drop in performance, raising questions
about cheating.)

--- Colin

----- 
FAQ and OpenGL Resources at:
  http://www.geocities.com/SiliconValley/Hills/9956/OpenGL

--

-- 
Author: Colin Fahey
  INET: cpfahey <at> earthlink.net

Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
San Diego, California        -- Mailing list and web hosting services
---------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru <at> fatcity.com (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB OPENGL-GAMEDEV-L
(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing).

Matt Campbell | 2 Jun 2003 08:59
Picon
Favicon

Procedural shaders

Hi,
    I'm currently trying to build some procedural shaders.  I'm getting some results, but getting very dissapointed overall.  I have a random noise texture (32x32x32).  Can someone give me some pointers or some sort of direction as to how I should be approaching procedural shaders (using fragment shaders).  My results just are not looking too good.
 
Thanks.
 
-Matt
Jimb Esser | 2 Jun 2003 09:29
Picon

Re: nvidia cheated in 3dmark2003?

Actually, I think you're pretty close on the mark at the end there.  I remember back in 3Dfx's days, their
driver would do certain optimizations specifically when they detected Quake was running, because it was
the primary benchmarking program (and it made the game run faster!).

Although ATI admitted to also fudging on the 3DMark tests!

Here's a snipped from C|Net's news.com:

"CNET | Digital Dispatch
Do speed tests lie?
May 29, 2003
Vol. 9, No. 22

There's nothing like a little scandal to spice up the 
otherwise boring world of hardware. Last week, we got more 
than a little excitement when Futuremark, maker of the 
industry's most common graphics-card benchmark, 3DMark03, 
accused Nvidia of manipulating test results. Nvidia, of 
course, denied cheating. Meanwhile, Nvidia's chief competitor, 
ATI, admitted that it, too, could be accused of fudging a 
little. So we went back to CNET Labs with Futuremark's 
updates and reran our tests on the current best cards from 
both makers. In the end, scores for Nvidia's GeForce FX 5900 
Ultra dropped as much as 22 percent, while scores for ATI's 
Radeon 9800 Pro 256MB fell only between 2 and 4 percent. Sounds 
shocking, but factoring in other tests, neither is a clear 
speed winner. Which will gamers buy? It's likely to boil 
down to word of mouth as these cards hit the streets and are 
used in real gameplay. "

----- 
FAQ and OpenGL Resources at:
  http://www.geocities.com/SiliconValley/Hills/9956/OpenGL

--

-- 
Author: Jimb Esser
  INET: esse0027 <at> tc.umn.edu

Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
San Diego, California        -- Mailing list and web hosting services
---------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru <at> fatcity.com (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB OPENGL-GAMEDEV-L
(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing).

mitea | 2 Jun 2003 23:34
Picon

RE: nvidia cheated in 3dmark2003?

Hmmm... I also thought it to be a hoax, but what about the clipping
planes? The shaders aren't conclusive because we can't see what the
driver actually processes, but a clipping plane. And it isn't an
optimisation (if that is the case, I'm not sure), it's cheating when you
'optimise' a video card's performance tester :))

----------------------------------
Peace and love,
Tweety
mitea <at> bbs.ro - tweety_04_01 <at> users.sourceforge.net
YahooID: tweety_04_01

> -----Original Message-----
> From: root <at> fatcity.com [mailto:root <at> fatcity.com] On Behalf Of 
> Colin Fahey
> Sent: Monday, June 02, 2003 5:50 AM
> To: Multiple recipients of list OPENGL-GAMEDEV-L
> Subject: Re: nvidia cheated in 3dmark2003?
> 
> 
> >>>Subject: nvidia cheated in 3dmark2003?
> >>>> http://www.futuremark.com/companyinfo/3dmark03_audit_report.pdf
> >>>>
> 
> I like this footnote on page 1:
> 
>   "[3] Futuremark's BETA program is an open, fee based 
> cooperation program
> [...]"
>                                        ----  --------- 
> -------------------
> 
> No wonder nVidia left the BETA program!
> 
> That article could benefit from a little more science.  For 
> example, the
> claims
> regarding pixel shader "detection" are implied, not proven.  
> If Futuremark
> quoted byte scans of the driver binary (e.g., nvidia.sys or 
> whatever it
> might
> be) that showed that the pixel shader code or name was 
> explicitly a search
> pattern used by the driver, it would be convincing, and it's 
> a test anyone
> could reproduce.
> 
> It *is* possible, if perhaps improbable, the drivers are 
> detecting general
> scenarios that can be optimized, so it's possible that the 
> Futuremark shader
> scenarios are actually "detected" by pure coincidence.
> 
> The irony is that the existence of an artificial benchmark 
> makes optimizing
> for artificial scenarios tempting, whereas if there were only major
> applications
> and games, consumers might see more optimizations that make a 
> practical
> difference in their actual lives!  I could just imagine if 
> nVidia crammed an
> optimized Quake 3 clone in to their drivers!  Or if nVidia changed the
> microchip itself to have MD3 model buffers and animation 
> paths.  All assets
> for all levels would be in firmware.  Drawing the weapon 
> crosshairs would
> be a hardwired stage of the rendering pipeline, along with 
> the health meter
> and "Major's" face icon.  Similarly, the audio clip "Quad 
> Damage" would be
> part of nVidia's sound chip.  The nVidia bridge chip would 
> have the Quake 3
> application in silicon, so the request to run "quake3.exe" would be
> "detected"
> and serviced without ever touching the disk!  (Making minor changes to
> "quake3.exe" might lead to a 24.1% drop in performance, 
> raising questions
> about cheating.)
> 
> --- Colin
> 
> 
> 
> ----- 
> FAQ and OpenGL Resources at:
>   http://www.geocities.com/SiliconValley/Hills/9956/OpenGL
> 
> -- 
> Author: Colin Fahey
>   INET: cpfahey <at> earthlink.net
> 
> Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
> San Diego, California        -- Mailing list and web hosting services
> ---------------------------------------------------------------------
> To REMOVE yourself from this mailing list, send an E-Mail message
> to: ListGuru <at> fatcity.com (note EXACT spelling of 'ListGuru') and in
> the message BODY, include a line containing: UNSUB OPENGL-GAMEDEV-L
> (or the name of mailing list you want to be removed from).  You may
> also send the HELP command for other information (like subscribing).
> 
> 

----- 
FAQ and OpenGL Resources at:
  http://www.geocities.com/SiliconValley/Hills/9956/OpenGL

--

-- 
Author: mitea
  INET: mitea <at> bbs.ro

Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
San Diego, California        -- Mailing list and web hosting services
---------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru <at> fatcity.com (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB OPENGL-GAMEDEV-L
(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing).

mitea | 3 Jun 2003 01:34
Picon

Mesa

Can I make an already compiled application to use Mesa if it was originally made to use OpenGL (in windows, no source)? How?
 
----------------------------------
Peace and love,
Tweety
YahooID: tweety_04_01
 
Eero Pajarre | 3 Jun 2003 09:09
Picon

Re: Mesa

mitea wrote:
> Can I make an already compiled application to use Mesa if it was 
> originally made to use OpenGL (in windows, no source)? How?
>

Copy the Mesa dll (named as opengl32.dll) to your application's
directory.

		Eero

----- 
FAQ and OpenGL Resources at:
  http://www.geocities.com/SiliconValley/Hills/9956/OpenGL

--

-- 
Author: Eero Pajarre
  INET: epajarre <at> koti.soon.fi

Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
San Diego, California        -- Mailing list and web hosting services
---------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru <at> fatcity.com (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB OPENGL-GAMEDEV-L
(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing).

MartinGBell@ntlworld.com | 3 Jun 2003 13:39

Models: loading, displaying...

Hi all,

Just wondering what the deal is with loading and drawing models in OpenGL?
I've got some code that loads .ASE files - all very nice, but I presume the
ASE file data isn't ordered into nice efficient triangle strips, it's just
separate poly's, normals, UV coords etc.

So... does this actually matter?  If I use e.g. a display list, will this
identify and optimise coincident vertices from the passed data, or will I
need to pre-process the data to get nice triangle strips etc.
Same goes for textures; will I need to sort the poly's by texture for max
efficiency, or again, will the all-clever list compilation do this?

Cheers,
Martin

--------------------------------------------------------------------
mail2web - Check your email from the web at
http://mail2web.com/ .

----- 
FAQ and OpenGL Resources at:
  http://www.geocities.com/SiliconValley/Hills/9956/OpenGL

--

-- 
Author: MartinGBell <at> ntlworld.com
  INET: MartinGBell <at> ntlworld.com

Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
San Diego, California        -- Mailing list and web hosting services
---------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru <at> fatcity.com (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB OPENGL-GAMEDEV-L
(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing).

Stephen J Baker | 3 Jun 2003 17:19
Favicon

Re: nvidia cheated in 3dmark2003?

mitea wrote:
> Hmmm... I also thought it to be a hoax, but what about the clipping
> planes? The shaders aren't conclusive because we can't see what the
> driver actually processes, but a clipping plane. And it isn't an
> optimisation (if that is the case, I'm not sure), it's cheating when you
> 'optimise' a video card's performance tester :))

Yes - the clip plane was clearly a cheat - because if you moved the demo
eyepoint off the pre-programmed 'track', the resulting image was incorrect.

Swapping the shader instructions might be valid - if it was performed for
ALL programs and if swapping them had no possible consequences for valid
programs...that would just be a regular optimisation.  But if swapping
those instructions was only done for that specific application or if doing
so could cause adverse consequences for other applications - then that too
would be cheating.   My understanding was that the latter was the case...but
I don't know for sure.

The problem here is that benchmarks are written SPECIFICALLY to be independent
of the hardware they run on and to NOT be specially optimised for individual
cards.  That's a noble goal - but it doesn't help the buying public to know
which widget to buy.  For the "Big Name" cards, games WILL be optimised to
their specific quirks - so benchmarks that aren't are just useless.

The temptation to cheat in order to make the benchmarks "more realistic"
must be huge.

----
Steve Baker                      (817)619-2657 (Vox/Vox-Mail)
L3Com/Link Simulation & Training (817)619-2466 (Fax)
Work: sjbaker <at> link.com           http://www.link.com
Home: sjbaker1 <at> airmail.net       http://www.sjbaker.org

----- 
FAQ and OpenGL Resources at:
  http://www.geocities.com/SiliconValley/Hills/9956/OpenGL

--

-- 
Author: Stephen J Baker
  INET: sjbaker <at> link.com

Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
San Diego, California        -- Mailing list and web hosting services
---------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru <at> fatcity.com (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB OPENGL-GAMEDEV-L
(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing).


Gmane