nV News Forums

 
 

nV News Forums (http://www.nvnews.net/vbulletin/index.php)
-   NVIDIA Linux (http://www.nvnews.net/vbulletin/forumdisplay.php?f=14)
-   -   Wild Cats run faster (http://www.nvnews.net/vbulletin/showthread.php?t=23473)

SnapIT 01-12-04 09:00 PM

Wild Cats run faster
 
Oki, i realize this is inappropiate in one way, but in another way not... there are benefits for Nvidia Users in this too...

Quadro is left WAY behind, cheaper and up to 3x faster cards are 3d labs wildcats, i will not even discuss this as any real test will reveal just that... (actually i am being VERY generous towards Nvidia here)

So why am i posting this here? Well, the agp part of the driver is free and can be made to work with the Nvidia driver allowing the KT800 chipset full support instead of the limited support provided otherwise..

It is all open source, so happy hacking... no, it cannot be used by Nvidia, and if they use it without releasing the entire driver as open source there would be problems for Nvidia...

(tip, tar -zxvf do a pipe grep or a find) rename, copy...

Happy hacking...

Soul-Crusher 01-13-04 12:03 AM

Re: Wild Cats run faster
 
Quote:

Originally posted by SnapIT
Quadro is left WAY behind, cheaper and up to 3x faster cards are 3d labs wildcats, i will not even discuss this as any real test will reveal just that...
Both 3dlabs and nVIdia cover a pretty broad spectrum from cost-effective dual headed displays for workstations to high-end cards for CAD and stuff like that. Before making such a claim, it would be nice to see a benchmark showing an apples-to-apples comparison.

I'm not familiar with the KT800 (Isn't that an Athlon-64 chipset?), what is your problem, specifically?

As for the driver itself, it looks like it just uses agpgart.

Oh, and I found a funny comment in their driver:
// Copyright (C) 2000 3Dlabs Inc. Ltd
// All Rights Reserved.
//
// This is UNPUBLISHED PROPRIETARY SOURCE CODE of 3Dlabs Ltd.
//
// This software and its associated documentation contains proprietary,
// confidential and trade secret information of 3Dlabs Ltd and except
// as provided by written agreement with 3Dlabs Ltd.
//
// a) no part may be disclosed, distributed, reproduced, transmitted,
// transcribed, stored in a retieval system, adapted or translated
// in any form or by any means electronic, mechanical, magnetic,
// optical, chemical, manual or otherwise,
//
// and
//
// b) the recipient is not entitled to discover through reverse
// engineering or reverse compiling or other such techniques or
// processes the trade secrets contained therein or in the
// documentation.
//
// US Government Restricted Rights Legend
//
// Use, duplication or disclosure by the Government is subject to
// restrictions as set forth in subdivision (c)(1)(ii) of the Rights in
// Technical Data and Computer Software clause at DFARS 252.227-7013,
// and/or in similar or successor clauses in the FAR, DOD or NASA FAR
// Supplement. Unpublished rights reserved under the Copyright Laws of
// the United States.
//


Oops, not anymore :p

zander 01-13-04 06:04 AM

Looking at the Wildcat III and Wildcat4 Linux Drivers, release 3.02, it seems that the 3Dlabs wildcat drivers are proprietary and similar to those provided by NVIDIA and ATi in that no source to any of the interesting driver components is made available (correct me if this is not true).

The kernel component seems to interface with the DRM, source for which is shipped with the driver (which I consider a delicate licensing situation, paired with the 3Dlabs license), but also relies on a binary object file. The included Makefile mentions the source files this object file was built from, but these files are not included (if they're shipped separately, I'd be interested in a URL). Both the included DRM source and the object file seem to be highly kernel version dependent, the latter even references versioned Linux kernel symbols directly, defeating the purpose of the open source interface layer (and implying another delicate licensing situation). To support AGP operation, the driver requires Linux AGPGART, with which it interfaces using code similar to that employed by the NVIDIA and ATi drivers (which shouldn't come as a surprise, the API is given); 3DLabs seems to support Linux/x86 only.

SnapIT, while I'm sure that most frequenters of this forum greatly appreciate your selfless quest for the benefits of NVIDIA users and will gladly overlook the occasional lack of substance or hints of fud in your posts, they might loose faith in your credibility or begin to doubt your intentions, should they find that you make uninformed or deliberately false claims along with the mention of supposedly superior competitor products.

SuLinUX 01-13-04 06:27 AM

Where does it say Wildcats are 3 times faster?, last time I read the QuadroFX dominated the the other cards.

SnapIT 01-13-04 06:42 AM

Re: Re: Wild Cats run faster
 
Quote:

Originally posted by Soul-Crusher
Both 3dlabs and nVIdia cover a pretty broad spectrum from cost-effective dual headed displays for workstations to high-end cards for CAD and stuff like that. Before making such a claim, it would be nice to see a benchmark showing an apples-to-apples comparison.

I'm not familiar with the KT800 (Isn't that an Athlon-64 chipset?), what is your problem, specifically?

As for the driver itself, it looks like it just uses agpgart.

Oh, and I found a funny comment in their driver:
// Copyright (C) 2000 3Dlabs Inc. Ltd
// All Rights Reserved.
//
// This is UNPUBLISHED PROPRIETARY SOURCE CODE of 3Dlabs Ltd.
//
// This software and its associated documentation contains proprietary,
// confidential and trade secret information of 3Dlabs Ltd and except
// as provided by written agreement with 3Dlabs Ltd.
//
// a) no part may be disclosed, distributed, reproduced, transmitted,
// transcribed, stored in a retieval system, adapted or translated
// in any form or by any means electronic, mechanical, magnetic,
// optical, chemical, manual or otherwise,
//
// and
//
// b) the recipient is not entitled to discover through reverse
// engineering or reverse compiling or other such techniques or
// processes the trade secrets contained therein or in the
// documentation.
//
// US Government Restricted Rights Legend
//
// Use, duplication or disclosure by the Government is subject to
// restrictions as set forth in subdivision (c)(1)(ii) of the Rights in
// Technical Data and Computer Software clause at DFARS 252.227-7013,
// and/or in similar or successor clauses in the FAR, DOD or NASA FAR
// Supplement. Unpublished rights reserved under the Copyright Laws of
// the United States.
//


Oops, not anymore :p

I said it it not even something to discuss, there is not one benchmark out there, do a search for yourself, that has the quadro ahead of the Wildcat, nada, zilch, zero, ingen... ok, the benches prove that the wildcat is WAY ahead and a LOT cheaper... www.google.com there is your link... I don't work for 3dlabs and right now i am no fan of them either...

Sorry, but the the open source part is still open source, and therefore you will find it bundled with the driver... the licence... well, i have to talk to them about that...

What is the problem with the kt800? do a search on any site you can find, and find out...

The agp driver part of the wildcat is actually mine and i used OSS licence to supply it thinking they would know what that means... obviously not...

I am removing my code from the wildcat driver release, i'll host the code elsewhere, and let you know...

See you around...

SnapIT 01-13-04 06:48 AM

Quote:

Originally posted by SuLinUX
Where does it say Wildcats are 3 times faster?, last time I read the QuadroFX dominated the the other cards.
Where did you look? Nvidias PDF?

Ehhh... no, the wildcat is so much faster, actually, i never benched it but i can tell you that the wildcat owns the quadro in every setup i have compared, real world setups, that is...

And it costs 2/3 of the card it whips...

Ok, i am out of this discussion now, there is no discussion about this, it is just the way it is...

QuadroFX did not dominate anything even when it was new... it definently does not dominate anything now, even the 9800 Pro with DRI whips it...

I do however see driver problems in 3dlabs immediate future...

zander 01-13-04 06:50 AM

http://www.spec.org/gpc/opc.data/vp711/summary.html

SnapIT 01-13-04 07:02 AM

Quote:

Originally posted by zander
http://www.spec.org/gpc/opc.data/vp711/summary.html
Interesting... you may have proved me wrong there...

Unless it was for one small thing, NONE of these benches were made on a Linux system with OGL2.0...

So in THIS forum they are totally irrelevant...

I see the performance in windows with OGL 1.2 and sure, it is all good, FX owns all there...

Who uses that spec? Nobody i know of...

But ok, sure, if you want to stretch it that far, you are correct...

nutball 01-13-04 07:25 AM

Quote:

Originally posted by SnapIT
Where did you look? Nvidias PDF?
... actually, i never benched it but i can tell you that the wildcat owns the quadro in every setup i have compared,

You need to read that sentence back to yourself.

I never benched it either (in fact I don't even own a Wildcat), but I can tell you that the Quadro 0wnZ the Wildcat. In fact I can say anything I like! It's easy to talk. That's the problem with people, they talk too much and say too little.

SuLinUX 01-13-04 07:53 AM

Thought as much, Tomshardware could'nt of been wrong and next time post proof instead of claiming it :rolleyes:

zander 01-13-04 08:01 AM

SnapIT, don't take this personally, but I'm sure most people value the results of one of the most commonly used workstation/OpenGL graphics performance benchmarks more than your subjective impression with setups you provide no detail on whatsoever. The posted SPECviewperf® results were indeed generated on Windows systems, but do you honestly expect them to be reversed on comparable Linux configurations?

LaNcom 01-13-04 10:32 AM

You know, my old SPEA V7 PCI owns 'em all, it's at least twice as fast as the QuadroFX, Radeon9800 and the Wildcat 4 alltogether...

Oops, sorry wrong reallity... I tried the Wildcat4, FireGL and QuadroFX 3000, and the QuadroFX is the fastest by a large margin, using XSI on Linux. The FireGL was nice, the Wildcat was really disappointing. Plus, IIRC, only the Quadro supports Linux for AMD64... Maybe SnapIT compared the Wildcat4 to the QuadroFX 500, then it would win; but if I have to choose between the 7210 and the 3000G, I would go Nvidia.
Then again, if there really is a way to improve it's performance any further... just go ahead!


All times are GMT -5. The time now is 07:18 PM.

Powered by vBulletin® Version 3.7.1
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Copyright ©1998 - 2014, nV News.