A year of nVidia

[Quitch]

Time and time again people come onto this board, or others, and say that nVidia isn't too bad, or you can't really notice the loss of image quality, or Half-Life 2 is an isolated inccident... time and time again forum posters are mustered, sent forth, and post rational and reasoned argument.

Yet, this is a huge duplication of effort. Examples are stated, restated and stated again, across many boards. Wouldn't it be better to have some article (or just a post) detailing everything that has happened in regards to nVidia for the last year. Hell, I followed it all and I still don't remember everything, yet I see so many misinformed articles who I want to correct, but I don't have the time or will to continually type out this information.

I really think something needs to be documented, and who better than Beyond3D, a site which brings some real technical know-how to the table.

[Nite_Hawk]

Hi Quitch,

I've been thinking about doing something like this for both ATI and Nvidia, but I haven't had the time recently, and am involved in working in another project at the moment. Might happen at some point though! :)

Nite_Hawk

[digitalwanderer]

Everytime I've tried to do something like that I keep getting comments like; "it's a nice recap, but nothing new", or "it's the same thing you've been reading about, but organized" and sort of gave up the effort on it.

I do it occasionally still in EB's staff lounge just so I have a place to reference all the links at though. (I'm incredibly lazy that way, private boards are great notepads. ;) )

[CI]

Hi!
Apparently we are on the same wavelength. I've done a record of the (negative) events surrounding nVidia since the launch of NV30, completed with relevant links and quotes. It's still not complete yet but I'll post it later (when I reached home) so that everyone can fill in the parts that's missed out. Cheers.
:)

[cthellis42]

Sounds good to me. ^_^ Information organization always makes me randy.

[Hanners]

Well, I've writen editorials that are basically summaries of some of the more pertinent goings on with nVidia of late, nut certainly no one article has covered every single aspect. Nor, to be honest, do I think it needs to - Most of the topics have been done to death, and those that don't 'get it' by now never will.

[CI]

This is not an article but rather a summary of references to the issues arising with nVidia since the NV30 preview. I've tried to link to the original source of information or official news releases if possible. Relevant quotes are included. Links without details are either too long to be included as quotes or play a more supportive role. Links to DriverHeaven thread's are not verified yet.

Some other events that are not included (yet):
- ShaderMark optimizations
- nVidia's article/brochure that explicitly claims NV30's 8-pipe architecture?
- Any article on trilinear issues with UT2003 other than the B3D thread?
- Any others?


Ok, here it goes:

15 Nov 2002: ExtremeTech previews the GeForceFX architecture
Article Title: Inside the GeForceFX Architecture

...
Pipes don't mean as much as they used to. . . . There are really 32 functional units that can do things in various multiples. We don't have the ability in NV30 to actually draw more than eight pixels per cycle. It's going to be a less meaningful question as we move forward...[GeForceFX] isn't really a texture lookup and blending pipeline with stages and maybe loop back anymore. It's a processor, and texture lookups are decoupled from this hard-wired pipe.
...

http://www.extremetech.com/print_article/0,3998,a=33705,00.asp
NOTE: nVidia avoids directly commenting on number of pipelines.


18 Nov 2002: nVidia officially announces the GeForceFX
http://www.nvidia.com/object/IO_20021117_7182.html


10 Feb 2003: Beyond3d raises issue of GeForceFX's 8 pipeline claim in its forums.
Thread Title: GeForce FX: 8x1 or 4x2?
http://www.beyond3d.com/forum/viewtopic.php?t=4252


11 February 2003: FutureMark releases 3DMark03
http://www.futuremark.com/pressroom/pressreleases/?021103


11 February 2003: nVidia executives criticizes 3DMark03
Article Title: Nvidia Decries 3DMark03
http://www.extremetech.com/article2/0,3973,884239,00.asp


14 February 2003: ExtremeTech analysis on legitimacy of 3DMark03
Article Title: Vendors Squabble Over 3DMark Benchmark

...
We spoke with company officials from ATI and Dell, and both companies believe that 3DMark03's methodology and implementation are essentially sound. Dell, in particular, will be using the benchmark as one of its metrics for qualifying 3D GPUs.
...

http://www.extremetech.com/article2/0,3973,888060,00.asp


15 February 2003:
Article Title: ATi and Futuremark's Response to NVIDIA's Report
http://hwextreme.com/reviews/misc/3dresponse/

~15 February 2003: FutureMark releases counterpoints to nVidia's 3DMark03 claims.
http://www.futuremark.com/companyinfo/Response_to_3DMark03_discussion.pdf


21 February 2003: TheInquirer clarifies with nVidia on GeForceFX architecture:

...
"GeForce FX 5800 and 5800 Ultra run at 8 pixels per clock for all of the following:
a) z-rendering
b) stencil operations
c) texture operations
d) shader operations"

and

"Only color+Z rendering is done at 4 pixels per clock"
...

http://www.theinquirer.net/?article=7920
Other references:
http://www.tech-report.com/news_archive_overview.x/2003/2/21
http://www.tech-report.com/news_archive_overview.x/2003/2/23


27 March 2003: nVidia officially releases Detonator 43.45
http://www.nvidia.com/object/winxp-2k_archive.html
http://www.nvidia.com/object/win9x_archive.html


10 April 2003: NVIDIA and Electronic Arts Form Exclusive Strategic Alliance
http://www.nvidia.com/object/IO_20030409_7420.html


12 May 2003: nVida officially announces the NV35
http://www.nvidia.com/object/IO_20030512_2396.html


14 May 2003: ExtremeTech releases article on cheating in 3DMark03 by 44.03 DetonatorFX
Article Title: Driver Irregularities May Inflate nVidia Benchmarks
http://www.extremetech.com/article2/0,3973,1088795,00.asp


14 May 2003: nVidia officially releases DetonatorFX 44.03
http://www.nvidia.com/object/winxp-2k_archive.html
http://www.nvidia.com/object/win9x_archive.html


23 May 2003: Extremetech releases article on "cheats"
Article Title: ExtremeTech Testing Confirms Futuremark Discovery of nVidia, ATI "Cheats"
http://www.extremetech.com/article2/0,3973,1105324,00.asp


23 May 2003: 3DMark03 Patch 330 Released (with Audit Report)
http://www.futuremark.com/news/?newsarticle=200305/2003052308#200305/2003052308
http://www.futuremark.com/companyinfo/3dmark03_audit_report.pdf

nVidia's response:

Since NVIDIA is not part in the FutureMark beta program (a program which costs of hundreds of thousands of dollars to participate in) we do not get a chance to work with Futuremark on writing the shaders like we would with a real applications developer. We don't know what they did but it looks like they have intentionally tried to create a scenario that makes our products look bad. This is obvious since our relative performance on games like Unreal Tournament 2003 and Doom3 shows that The GeForce FX 5900 is by far the fastest graphics on the market today.


ATi's response:

The 1.9% performance gain comes from optimization of the two DX9 shaders (water and sky) in Game Test 4 . We render the scene exactly as intended by Futuremark, in full-precision floating point. Our shaders are mathematically and functionally identical to Futuremark's and there are no visual artifacts; we simply shuffle instructions to take advantage of our architecture. These are exactly the sort of optimizations that work in games to improve frame rates without reducing image quality and as such, are a realistic approach to a benchmark intended to measure in-game performance. However, we recognize that these can be used by some people to call into question the legitimacy of benchmark results, and so we are removing them from our driver as soon as is physically possible. We expect them to be gone by the next release of CATALYST.

http://www17.tomshardware.com/technews/20030523_192553.html


03 June 2003: Futuremark and NVIDIA Statement
http://www.futuremark.com/news/?newsarticle=200306/2003060305#200306/2003060305

...
Futuremark now has a deeper understanding of the situation and NVIDIA's optimization strategy. In the light of this, Futuremark now states that NVIDIA's driver design is an application specific optimization and not a cheat.
...
However, recent developments in the graphics industry and game development suggest that a different approach for game performance benchmarking might be needed, where manufacturer-specific code path optimization is directly in the code source. Futuremark will consider whether this approach is needed in its future benchmarks.
...



16 June 2003: Firingsquad compares FX5900Ultra vs Radeon9800Pro and others
Article Title: MSI GeForce FX5900-TD128 Review
http://firingsquad.gamers.com/hardware/msi_geforce_fx5900-td128_review/default.asp
NOTE: Use of non-standard benchmarks on non-DX9 level games


30 June 2003: Firingsquad compares FX5900Ultra vs Radeon9800Pro and others
Article Title: 3D Performance with Unreal Tournament 2003
http://firingsquad.gamers.com/hardware/ut2003perf/default.asp
NOTE: Use of non-standard benchmarks


03 July 2003: Beyond3d thread on NV3x texturing in UT2003 issue created
Thread Title: GeForce FX & UT2003: "Performance" Texturing A
http://www.beyond3d.com/forum/viewtopic.php?t=6719


03 July 2003: digit-life (Unwinder) releases article on benchmark cheat analysis:
Article Title: Battle of ATI and NVIDIA: where are fair duels? or dishonest treating of 3DMark
http://www.digit-life.com/articles2/antidetect/index.html
NOTE: Article uses AntiDetector scripts.


~07 July 2003: First news of nVidia to rejoin FutureMark's beta program:
http://www.beyond3d.com/forum/viewtopic.php?t=6769


18 July 2003: nvidia (through HardOCP) defines guidelines for optimizations:
NOTE: Related images:
http://www.hardocp.com/images/news/1058512731oj8wsooRuO_1_1_l.gif
http://www.hardocp.com/images/news/1058512731oj8wsooRuO_1_2_l.gif
http://www.hardocp.com/images/news/1058512731oj8wsooRuO_1_3_l.gif


~19 July 2003: nVidia mades up with DriverHeaven and Omegadrive
http://www.driverheaven.net/showthread.php?s=&threadid=20967
http://www.driverheaven.net/showthread.php?s=&threadid=20963
NOTE: Forum requires free registration


21 July 2003: amdmb examines analysis on 30% performance increase for UT2K3
Article Title: UT2K3 Image Quality: Detonator 44.03 Examined
http://www.amdmb.com/article-display.php?ArticleID=247&PageID=1


22 July 2003: ExtremeTech releases article:
Article Title: nVidia's Risky 3D Optimization Gamble
http://www.extremetech.com/article2/0,3973,1200962,00.asp


~22 July 2003: DriverHeaven reports on legal means to end (nVidia) Omega drivers
http://www.driverheaven.net/showthread.php?threadid=18877&s=
NOTE: Forum requires free registration


26 July 2003:
Article Title: ATi & nVidia in driver "optimizations" race
http://www.3dcenter.org/artikel/ati_nvidia_treiberoptimierungen/index_e.php
NOTE: Article includes use of DigitLife's anti-cheat script.


12 August 2003: nVidia rejoins FutureMark benchmark development program
http://www.futuremark.com/pressroom/pressreleases/?081203


13 August 2003: nVidia officially releases Detonator 45.23
http://www.nvidia.com/object/winxp-2k_45.23
http://www.nvidia.com/object/win9x_45.23


24 August 2003: Beyond3D compares DX9 performance in TombRaider benchmarks
Article Title: Tomb Raider: Angel of Darkness DX9 Performance
http://www.beyond3d.com/misc/traod_dx9perf/


03 Sep 2003: Tomb Raider Patch v.49 released
http://www.eidosinteractive.co.uk/support/patchinfo.html?ptid=67


05 Sep 2003: Gamers Depot emailed John Carmack on NV3x pixel shading performance:

GD: John, we've found that NVIDIA hardware seems to come to a crawl whenever Pixel Shader's are involved, namely PS 2.0..

Have you witnessed any of this while testing under the Doom3 environment?

"Yes. NV30 class hardware can run the ARB2 path that uses ARB_fragment_program, but it is very slow, which is why I have a separate NV30 back end that uses NV_fragment_program to specify most of the operations as 12 or 16 bit instead of 32 bit."

John Carmack

http://www.gamersdepot.com/hardware/video_cards/ati_vs_nvidia/dx9_desktop/001.htm


10 Sep 2003: Valve showcases HalfLife2 benchmark performances on ATi Shader Day presentation:
Article Title: Valve lets off steam over NVIDIA performance... and offers a new graphics benchmark
http://www.techreport.com/etc/2003q3/valve


11 Sep 2003: Microsoft endorses Half Life2 benchmark:

We see ‘Half-Life 2’ as a new benchmark for the type of amazing experiences that can be delivered on the Windows platform, and DirectX 9.0 is clearly serving as the catalyst for the development of these state-of-the-art games," said Dean Lester, general manager of Windows Gaming and Graphics at Microsoft Corp. "‘Half-Life 2’ emphasizes the trend we are already seeing: Games for Windows now deliver the most cutting-edge technology and immersive entertainment available anywhere.



11 Sep 2003: nVidia responds to Half Life2 performance issues:

...
In addition to the developer efforts, our driver team has developed a next-generation automatic shader optimizer that vastly improves GeForce FX pixel shader performance across the board. The fruits of these efforts will be seen in our Rel.50 driver release. Many other improvements have also been included in Rel.50, and these were all created either in response to, or in anticipation of the first wave of shipping DirectX 9 titles, such as Half Life 2.
...



12 Sep 2003: ExtremeTech confirms Half Life2 benchmark results released by Valve
Article Title: Valve's Half-Life 2 Test Results Confirmed
http://www.extremetech.com/article2/0,3973,1265063,00.asp
NOTE: Analysis for individual test demos explained


15 Sep 2003: Beta Win2k & WinXP nVidia 51.75 Detonator Driver
(dated 28 Aug 03) available from http://www.3dchipset.com/drivers/beta/nvidia/nt5/5175.php


13 Sep 2003: Analysis reveals performance improvements of 51.75 driver at expense of image quality:
Article Title: AQUAMARK 3 PREVIEW & PERFORMANCE COMPARISON
http://www.driverheaven.net/articles/aquamark3/index.htm


14 Sep 2003: Comparison of 5900Ultra vs Radeon9800Pro AIW in PixelShader2 games:
Detonator 50 Drivers
http://www.gamersdepot.com/hardware/video_cards/ati_vs_nvidia/dx9_rel50/001.htm


15 Sep 2003: AquaMark3 officially released
http://www.aquamark3.com/archive.htm


15 Sep 2003: nVidia claims Detonator 51.75 not ready:

After receiving some emails from Andrew (European PR Manager for nVidia), we decided to pull down the Detonator 51.75 driver because this driver is not ready to go and is absolutely not ready to be installed on any machine yet. Andrew doesnt know exactly where its coming from, but nVidia released this driver to some hard- and software firms just for testing purposes, since theyve changed some interesting things for the 50.xx driver series.

Some of the release notes for this driver that were floating around the net are just a fake and have nothing to offer current nVidia card based owners.

Andrew also told us that nVidia has made some improvements in the new Detonators 50.xx series, including some enhancements to nView.

All in all, we can say that after a little talk we agree with nVidia to remove the driver and to make our visitors aware of the fact that this release is not official and also not allowed yet by nVidia.

Thanks to Andrew Humber from nVidia for the info!

http://www.techconnect.ws/modules.php?name=News&file=article&sid=2649


15 Sep 2003: Removal of true trilinear filtering
Detonator 51,75 with new "optimizations"

...Very amazingly it was however that from unreal Tournament the 2003 admitted optimization "off" now suddenly also to all Direct3D-Games applied
...
As is to be seen clear, the nVidia driver does not use 51,75 under Direct3D generally for all plays for the texture stages 1-7 no more full trilinear filter, but only a pseudotrilinear.
...
The results under zuschaltung of the 4x of anisotropic filter are still more amazing. Again the pseudotrilinear filter is to be recognized with the texture stages 1-7. At the same time it is to be seen however clear that those texture stages are filtered 1-7 no more with the actually stopped degree at anisotropic filtering, but only with a stage less, thus only with 2x anisotropic filter. This effect is constantly alike with 2x to 8x of anisotropic filtering: The texture stages 1-7 only pseudotrilinear filtered and gotten only 2x anisotropic filter off, all the same whether 2x, 4x or 8x are adjusted.
...

http://www.3dcenter.org/artikel/2003/09-15_a.php
NOTE: Article is in German, use Bablefish to translate.


16 Sep 2003: Eidos pulled patch 49 for Tomb Raider: Angel of Darkness, due to nVidia influence.

"It has come to the attention of Eidos that an unreleased patch for Tomb Raider: AOD has unfortunately been used as the basis for videocard benchmarking. While Eidos and Core appreciate the need for modern benchmarking software that utilizes the advanced shading capabilities of modern graphics hardware, Tomb Raider: AOD Patch 49 was never intended for public release and is not a basis for valid benchmarking comparisons. Core and Eidos believe that Tomb Raider: AOD performs exceptionally well on NVIDIA hardware." - Paul Baldwin, Eidos



17 Sep 2003: John Carmack about GeForce FX and DirectX 9

Hi John,

No doubt you heard about GeForce FX fiasco in Half-Life 2. In your opinion, are these results representative for future DX9 games (including Doom III) or is it just a special case of HL2 code preferring ATI features, as NVIDIA suggests?

"Unfortunately, it will probably be representative of most DX9 games. Doom has a custom back end that uses the lower precisions on the GF-FX, but when you run it with standard fragment programs just like ATI, it is a lot slower. The precision doesn't really matter to Doom, but that won't be a reasonable option in future games designed around DX9 level hardware as a minimum spec.

John Carmack"

http://english.bonusweb.cz/interviews/carmackgfx.html


23 Sep 2003: Futuremark Publishes Guidelines for Driver Optimizations

...
1) It is prohibited to change the rendering quality level that is requested by 3DMark.

2) It is prohibited to detect 3DMark directly or indirectly. In its sole discretion, Futuremark may approve detection in order to fix a specified hardware error.

3) Optimizations that utilize the empirical data of 3DMark are prohibited.

4) Generic optimizations that do not violate the above rules and benefit applications in general are acceptable only if the rendering is mathematically consistent with that of Microsoft® DirectX® reference rasterizer.

As a summary, all 3DMark specific optimizations are prohibited. Additionally, all generic optimizations that change the rendering quality requested by 3DMark are prohibited.
...

http://www.futuremark.com/pressroom/pressreleases/?092303


26 Sep 2003: Tomb Raider Patch v.52 released
http://www.eidosinteractive.co.uk/support/patchinfo.html?ptid=67
NOTE: Release date remains as that of v.49

According to the patch release there is "No bug fixes" and the benchmarking mode has been removed. To me that suggests that the results would be consistent with the 49 release, hence results from that would still be an accurate reflection pf performance withing the game when utilising eiterh the 49 or 52 release.

http://www.beyond3d.com/forum/viewtopic.php?t=8176


27 Sep 2003:
Article Title: AquaMark 3 Image Quality Comparison
http://www.3dgpu.com/modules/wfsection/article.php?articleid=76


29 Sep 2003: ShaderMark2.0 released

...
ANTI-DETECT-MODE fights against shader and texture specific optimisations
...

http://www.shadermark.de/start.html

:)

[jb]

HOLY Crap thats a great list CI, Thanks sooo much!

[mczak]

since this list begins at 15 nov, this even leaves soom room for some more events if you want to cover a year :twisted:

[Joe DeFuria]

Nice list!

I don't know though, if it's a good thing or a bad thing that [H] doesn't appear on it at all. ;)

Also, I would say that the ball started rolling not in November, but with the R300 launch....that's when web sites started (presumably with nVidia inside info) "pre-hyping" the NV30. "Yeah, Radeon 9700 is good...but wait until NV30 gets here soon....I have these specs, see...."

The list also does not include any of the Cg release and hype which, IIRC also started in the Summer of '03....

[digitalwanderer]

Nice list!

I don't know though, if it's a good thing or a bad thing that [H] doesn't appear on it at all. ;)
I want a seperate list for [H] and Kyle so I can point him at it next time he tells me he's been backing ATi all year... :roll:

[Quitch]

Well, I've writen editorials that are basically summaries of some of the more pertinent goings on with nVidia of late, nut certainly no one article has covered every single aspect. Nor, to be honest, do I think it needs to - Most of the topics have been done to death, and those that don't 'get it' by now never will.

Not every one is around every second of every day. Just because something was covered, doesn't mean everyone saw it, nor that they ever will. How can you track down what you don't know has happened?

CI, that is a brilliant list, you should start a topic for it, maybe someone will sticky it (or at the least we can add links as required).

[Laa-Yosh]

I have to agree too, it's a very nice compilation.

However, I think that you should add links to John Carmack's .plan updates as well, from spring '03, as he has talked extensively about the performance he got with the NV30 (ie. 50% as fast as the R300 when using the same full precision codepath).

[Laa-Yosh]

29 January 2003: John Carmack talks about NV30 and R300 performance
.plan update

The NV30 runs the ARB2 path MUCH slower than the NV30 path. Half the speed at the moment. This is unfortunate, because when you do an exact, apples-to-apples comparison using exactly the same API, the R300 looks twice as fast, but when you use the vendor-specific paths, the NV30 wins.

http://www.gamefinger.com/plan.asp?userid=johnc&id=16154


Edit: formatting

[euan]

You missed the comment from Nvidia around the time the DX9 non-standard benchmarking started, where they stated that FRAPS was not a reliable measuring tool for FX cards. :D

Other than that it's pretty good.

It would be made complete if you included all the marketing and PR FUD that was being spewed by NV before xmas. Such as the break-in video, the cinematic rendering, how it will beat the competition etc. Also where is the disapearance of the 5800 from Nvidia's website. And lastly the time they ridiculed there own 5800 product at a developers conference.

[euan]

Nice list!

I don't know though, if it's a good thing or a bad thing that [H] doesn't appear on it at all. ;)
I want a seperate list for [H] and Kyle so I can point him at it next time he tells me he's been backing ATi all year... :roll:

even better we could have all parties in a vertical or horizontal time line. hehe

[Sabastian]

Good compilation CI.

[cthellis42]

And lastly the time they ridiculed there own 5800 product at a developers conference.

That part actually impressed me, and I was like "ah, ok, they recognize their own flaws" and such. But, um... a lot of random crap has come along since then. -_-

At any rate, great compilation! Took me a bit to pick up my jaw there, but hey. Have fun keeping up. ;)

[Bouncing Zabaglione Bros.]

And lastly the time they ridiculed there own 5800 product at a developers conference.

That part actually impressed me, and I was like "ah, ok, they recognize their own flaws" and such. But, um... a lot of random crap has come along since then. -_-



To me, that was a kick in the teeth to their customers. Either they were stupid and built NV30 badly, only to suddenly realise it was trash, or they knew NV30 was rubbish all along, but they spent six months lying to their customers about it.

How do I know they are not doing the same thing right now in their offices about NV35? They are already lying and saying it's faster and better than the benchmarks and games show it to be. How do I know there won't be a video after NV40 is announced "impressing me" with how Nvidia think the NV35 is an underpowered DX8-and-a-half design they they decided to lie about for the last year too?

If that's the kind of respect Nvidia have for their customers, I'll take my money elsewhere thanks...

[CI]

Nice list!

I don't know though, if it's a good thing or a bad thing that [H] doesn't appear on it at all. ;)

Also, I would say that the ball started rolling not in November, but with the R300 launch....that's when web sites started (presumably with nVidia inside info) "pre-hyping" the NV30. "Yeah, Radeon 9700 is good...but wait until NV30 gets here soon....I have these specs, see...."

The list also does not include any of the Cg release and hype which, IIRC also started in the Summer of '03....


Well, I thought of starting out this list fairly recently when I realised that I'm losing track of the various (mostly negative) events that arise since NV30's launch. It's more of a reminder of what was wrong with nVidia. Hyping up a product before launch is fine by me since there's no indication of actual performance. I would prefer the product to speak for itself. It's the sheer amount of untruths, propaganda and damage that nVidia dealt out in its attempt to cover up/deny the GeForceFX weaknesses after third party evaluation that irks me. As such, I've tried to restrict the scope of the list to early article/threads whose contents pertain to such nature for various events. You could say that the focus of this list is a chronological documentation to nVidia's FUD attempts after NV30's launch in oppose to the truth.

Cg is primarily meant for developers. Articles/threads with statements from such folks inline with the focus of the list would be most useful. I haven't had the chance to encounter such documents yet. Like I said, this list is not complete and certainly far from perfect. I would welcome any other useful contributions in this area.

As for [H]ardOCP, I must also admit that I have not gone through their records yet. In the course of my search, apart from 3 slides, I have yet to come across any of the situations above in which they have been referenced for making an early investigation/statement/clarification in their headlines. I'll amend accordingly if I can find the time to dig into their site.


29 January 2003: John Carmack talks about NV30 and R300 performance
.plan update

The NV30 runs the ARB2 path MUCH slower than the NV30 path. Half the speed at the moment. This is unfortunate, because when you do an exact, apples-to-apples comparison using exactly the same API, the R300 looks twice as fast, but when you use the vendor-specific paths, the NV30 wins.

http://www.gamefinger.com/plan.asp?userid=johnc&id=16154


Edit: formatting

Dang! I've forgotten all about Carmack's .plan file. Thanks for the reminder. I'll include it for the update.


You missed the comment from Nvidia around the time the DX9 non-standard benchmarking started, where they stated that FRAPS was not a reliable measuring tool for FX cards. :D

Other than that it's pretty good.

It would be made complete if you included all the marketing and PR FUD that was being spewed by NV before xmas. Such as the break-in video, the cinematic rendering, how it will beat the competition etc. Also where is the disapearance of the 5800 from Nvidia's website. And lastly the time they ridiculed there own 5800 product at a developers conference.

Thanks for the reminder on the FRAPS statement. Yep, found a thread here where Dave quoted nVidia to 3DVelocity's reply. Unfortunately I can't seem to locate it on 3DVelocity's site which would be better, unless it's in their forums...

As for the videos, it's just mostly common PR stuff which is not really the focus of the list. The NV30 is on their site:
http://www.nvidia.com/page/fx_5800.html
though it's unclear whether it's referring to the Ultra or non-Ultra version.
I'll include in the FlowFX spoof.


Ok, here's the updates for some of the suggestions as well as a few more:

9 August 2002: Tech-report covers DX9 cards
Article Title: ATI's Radeon 9700 versus NVIDIA's NV30
http://www.tech-report.com/etc/2002q3/nextgen-gpus/index.x?pg=5
NOTE: nVidia's non-disclosure of pipeline and texture sampling count


29 January 2003: John Carmack's .plan updates

...The NV30 runs the ARB2 path MUCH slower than the NV30 path.
Half the speed at the moment. This is unfortunate, because when you do an exact, apples-to-apples comparison using exactly the same API, the R300 looks twice as fast, but when you use the vendor-specific paths, the NV30 wins.
...
http://www.gamefinger.com/plan.asp?userid=johnc&id=16154
http://www.webdog.org/cgi-bin/finger.plm?id=1&time=20030208202555


24 February 2003: Tech-report summarizes the 3DMark03 controversy
Article Title: Dissecting the 3DMark03 controversy
http://www.tech-report.com/etc/2003q1/3dmark03-story/index.x?pg=1


19 May 2003: NV30 FlowFX spoof from nVidia
News Title: NV30 Is Handy!
http://www.nvnews.net/cgi-bin/archives.cgi?category=1&view=5-03
http://www.nvnews.net/files/NVIDIAFlowFX_video.zip
Alternative resources:
http://users.pandora.be/darkt/NVIDIAFlowFX_video.zip
(from: http://www.theinquirer.net/?article=9773)


2 June 2003: Dell weights in support for 3DMark03
Article Title: Dell voices support for 3DMark03

Dell uses many tools to evaluate system and graphics subsystem performance. We believe 3DMark03 is a solid synthetic graphics benchmark that covers a wide range of usage models and complements application-specific testing. Synthetic benchmarks like 3DMark03 help to differentiate graphics subsystem performance characteristics of both high end and lower end cards by utilizing sets of tests with varying degrees of graphics complexity. Entry level cards will be able to run at least one simple test to be used in comparisons for those interested in basic 3D functionality. Additionally, those interested in leading edge technology with be able to make graphics hardware comparisons with a range of tests using the new APIs, shaders, rendering techniques, etc. Dell believes 3DMark03 is a versatile tool that allows a fair comparison of today's wide range of 3D graphics solutions.

http://www.tech-report.com/onearticle.x/5208


5 June 2003: Tech-report reports on AF performance by renaming 3Dmark03
Article Title: Further NVIDIA optimizations for 3DMark03?
http://www.tech-report.com/etc/2003q2/3dmurk03/index.x?pg=1


13 June 2003: 3DVelocity reviews the NV35
Review Title: NVIDIA GeForceFX 5900 Ultra (Reference)

... In the words of author Thomas Bruckschlegel these 2.0 and 2.0 partial precision shaders deliver the same image as the old ones, but they have some shuffled instructions, which makes them invisible for current driver optimisations. Unlike in the review, these were run using DX8 Mip Filter Reflections due to some apparent problems with DX9 Mip Filter Reflections as reported by ToMMTi.

http://www.3dvelocity.com/reviews/5900u/5900u_16.htm
NOTE: The page in link above comes after the review as an update.


19 Jun 2003: nVidia raises doubts on reliability of FRAPS in response to 3DVelocity's review
Thread Title: NVIDIA Are the Industry Standard Shader Driving Force?

When I asked about the FX's relatively poor performance in our "real" game tests the reply wasn't entirely clear but they certainly claim to have doubts on the reliability of FRAPS and the reliability of those using it.
http://www.beyond3d.com/forum/viewtopic.php?t=6510


24 July 2003: Bjorn3D interviews nVidia driver team
Article Title: Bjorn3D NVIDIA Driver chat log
http://www.bjorn3d.com/_preview.php?articleID=313


Changes in nVidia's marketing for GeForceFX's architecture:
nVidia's claims of 8 pixels per clock for GeForceFX:

...
Table 1. Performance comparison of GeForce FX and previous-generation
GPUs.

Previous Generation | GeForceFX GPU | Benefits from GeForce FX
...
4 pixels per clock | 8 pixels per clock | Doubles the fill rate to power through new applications as well as the classic games
...

http://www.nvidia.com/docs/lo/2416/SUPP/Overview.pdf (PDF created ?, no longer available from nVidia)
File: TB-00653-001_v01_Overview_110402.pdf (PDF created 04/11/2002, no longer available from nVidia)

Key Features Title: Fifth-Generation Workstation Graphics Architecture

...and eight fully programmable pixel pipelines...

http://www.nvidia.com/object/IO_20030117_6779.html (Available as of 04 Oct 2003)

...
Up to 8 Pixels per Clock Rendering Pipeline
Makes all of your favorite games run faster.
...
http://www.nvidia.com/page/geforcefx_features.html (Available as of 04 Oct 2003)

...
Table 1. GeForceFX Family Comparison
| 5900 ...
Pixels per Clock | 8/4 ...
...

Document Title: Technical Brief: NVIDIA GeForce FX GPUs
http://www.nvidia.com/page/techbriefs111802.html
http://www.nvidia.com/object/overview_tb.html
http://www.nvidia.com/attach/4109?type=support&primitive=0
File: NVIDIA_GeForce_FX_GPUs_041803_v2.pdf (PDF created 23/04/2003, available as of 04 Oct 2003)


Other information yet to be included:
- nVidia's recommendation for 2xAA and ?AF in benchmarking?
- nVidia (David Kirk interview?) claims that definitions not longer accurate to describe their architecture.


Keep those suggestions flowin' in! :wink:
Last of all, do pardon my grammar if it sounds funny in any way! Cheers again!
:)

[Sxotty]

To me, that was a kick in the teeth to their customers.
If that's the kind of respect Nvidia have for their customers, I'll take my money elsewhere thanks...

You'd take your money elsewhere anyway so it is no big deal. I was glad they did it, and the best part is everyone who knows they got burned will either
A) Buy ATI next time
B) do a little research and not buy on brand loyalty

So really you should be happy, the more informed the consumer the higher standards NVDA, and ATI have to adhere to, this really is in the end true.

[AJ]

Keep those suggestions flowin' in! :wink:


That's looking like a pretty comprehensive list. One thing that I recall was the 100k units of NV30's. Back in Feb the news leaked out from TSMC that the first printout was only 100k (where as a typical major launch normally consists from 1-2 million units). Quite a few sites picked this up and wrote that GFFX had been canned before it's launch. (INQ comes to mind)

NVidia vocally denied this rumor and later put out a press release saying that 1.5 million GFFX units had been ordered/shipped, but if you read the release closely you noticed that this was for the whole NV3X family not NV30. (Couldn't find the urls, sorry.)

Cheers,

AJ

BTW: A list like this should be hosted in it's own url (Uttar? How about NFI?) so that you could aways simply easily refer to the url. :) Also, in order to be fair there should be open pages "available" for all manufacturers so that you could compare notes easily...

[digitalwanderer]

Keep those suggestions flowin' in! :wink:


That's looking like a pretty comprehensive list. One thing that I recall was the 100k units of NV30's. Back in Feb the news leaked out from TSMC that the first printout was only 100k (where as a typical major launch normally consists from 1-2 million units). Quite a few sites picked this up and wrote that GFFX had been canned before it's launch. (INQ comes to mind)

NVidia vocally denied this rumor and later put out a press release saying that 1.5 million GFFX units had been ordered/shipped, but if you read the release closely you noticed that this was for the whole NV3X family not NV30. (Couldn't find the urls, sorry.)

Cheers,

AJ

BTW: A list like this should be hosted in it's own url (Uttar? How about NFI?) so that you could aways simply easily refer to the url. :) Also there should be open pages "available" for all manufacturers so that you could compare notes easily...
You've become an awful lot more interesting on the boards since your job change, and I mean that in a most complimentary way! :)

[AJ]

You've become an awful lot more interesting on the boards since your job change, and I mean that in a most complimentary way! :)

Thanks mate. 8) It's really refresing when you don't need to worry about mixing personal opinions with company statements. I like it too.

Cheers,

AJ

[CorwinB]

IIRC, some big wig at NV (perhaps DK, I'm not sure) made some really interesting comments last year, which are quite funny when you look back, something among the lines of :
"A 256 bits memory bus is overkill" or
"It is not possible to build a true next generation part on the 0.15 process"

Anyone got links to those ?

[THe_KELRaTH]

How about including info where it was found that the FX only does vertical offset AA, not horizontal offset axis AA. Yet another "feature" to create the illusion that their product is more efficient than their competitors in benchmarks.

www.jlmay.f2s.com/NV30-ATIAA.jpg

Anybody remember what the pre 3Dmark03 cheat drivers were?

[Pete]

44.03 was the last official "pre-cheat" release, IIRC. I think 44.10 were the same, but I believe those weren't released on nV's site.

[Neeyik]

44.03 was the last official "pre-cheat" release, IIRC. I think 44.10 were the same, but I believe those weren't released on nV's site.
I don't think you can label the 44.03s as being "pre-cheat" since they contain 3DMark03-specific opti-che-code. The 330 build catches them though.

[digitalwanderer]

I'm truly starting to think that the 30.82 set is the last non-cheat set of nVidia drivers :(

[Ali]

I didnt see it listed, but might have have missed it.

I think the nVidia sponsored Doom3 benchmarks showing the NV30 spanking the r300 is interesting. Also the quote from john Carmack saying he recorded a new demo because he wasnt hapy with the one nvidia prepared.

Ali

[Sabastian]

I'm truly starting to think that the 30.82 set is the last non-cheat set of nVidia drivers :(

I will add too that. Since the Det30's are a tad slower I often opt for the Det 29.XX series or lower. After nvidia started on the Det30.xx performance slowed significantly. heh, who knows maybe they were cheating with the Det 2X. series as well.

[Arun]

Keep those suggestions flowin' in! :wink:
BTW: A list like this should be hosted in it's own url (Uttar? How about NFI?) so that you could aways simply easily refer to the url. :) Also, in order to be fair there should be open pages "available" for all manufacturers so that you could compare notes easily...

That's not a bad idea, although I'd like to use NFI for other stuff than GPUs from now on ( okay, so that means nothing for hundreds of years :P ) - but if no one else proposed, sure. Could even do a small Perl script to make it easier to manage - I always loved how darn easy GPU:RW was to update. No useless crap, just the essantial.


Uttar

[Sabastian]

11 Nov 2003: FutureMark Patch 340 released

Patch Driver 3DMarks GT1 GT2 GT3 GT4
GeForce FX 5950 Ultra 330 52.16 6412 205.7 46.6 37 37.3
GeForce FX 5950 Ultra 333 52.16 5538 205 39.6 33.1 26.3
Radeon 9800 XT 330 3.8 6435 209.7 45.4 38.6 36.3
Radeon 9800 XT 333 3.8 6436 210.5 45.4 38.6 36.3

http://www.beyond3d.com/forum/viewtopic.php?t=8952&start=0&sid=da9e6dbd11e69e93de41f8604d6a29dd

Note: 333 is the release candidate of 340 according to DaveBaumann

EDIT: I thought that it would be good to add this to the list CI. ;)

[ByteMe]

11 Nov 2003: FutureMark Patch 340 released

Patch Driver 3DMarks GT1 GT2 GT3 GT4
GeForce FX 5950 Ultra 330 52.16 6412 205.7 46.6 37 37.3
GeForce FX 5950 Ultra 333 52.16 5538 205 39.6 33.1 26.3
Radeon 9800 XT 330 3.8 6435 209.7 45.4 38.6 36.3
Radeon 9800 XT 333 3.8 6436 210.5 45.4 38.6 36.3

http://www.beyond3d.com/forum/viewtopic.php?t=8952&start=0&sid=da9e6dbd11e69e93de41f8604d6a29dd

Note: 333 is the release candidate of 340 according to DaveBaumann

EDIT: I thought that it would be good to add this to the list CI. ;)



WTF??????? Nvidia STILL at it?

[digitalwanderer]

11 Nov 2003: FutureMark Patch 340 released

Patch Driver 3DMarks GT1 GT2 GT3 GT4
GeForce FX 5950 Ultra 330 52.16 6412 205.7 46.6 37 37.3
GeForce FX 5950 Ultra 333 52.16 5538 205 39.6 33.1 26.3
Radeon 9800 XT 330 3.8 6435 209.7 45.4 38.6 36.3
Radeon 9800 XT 333 3.8 6436 210.5 45.4 38.6 36.3

http://www.beyond3d.com/forum/viewtopic.php?t=8952&start=0&sid=da9e6dbd11e69e93d e41f8604d6a29dd

Note: 333 is the release candidate of 340 according to DaveBaumann

EDIT: I thought that it would be good to add this to the list CI. ;)



WTF??????? Nvidia STILL at it?
Good morning BiteMe, wants some coffee with that brief reality interlude? ;)

[cthellis42]

Better throw a donut in as well.

[digitalwanderer]

The Dig places a selection of fresh and tasty donuts alongside the coffee.

[ByteMe]

The Dig places a selection of fresh and tasty donuts alongside the coffee.

Maybe I am not calm enough, but if anyone in real life treated me the way nvidia has been treating the consumer... I'll punch him/it in the mouth. I would think that 90%+ of everyone here would be fuming.

How's the job? and the ticker?

[Quitch]

11 Nov 2003: FutureMark Patch 340 released

Patch Driver 3DMarks GT1 GT2 GT3 GT4
GeForce FX 5950 Ultra 330 52.16 6412 205.7 46.6 37 37.3
GeForce FX 5950 Ultra 333 52.16 5538 205 39.6 33.1 26.3
Radeon 9800 XT 330 3.8 6435 209.7 45.4 38.6 36.3
Radeon 9800 XT 333 3.8 6436 210.5 45.4 38.6 36.3

http://www.beyond3d.com/forum/viewtopic.php?t=8952&start=0&sid=da9e6dbd11e69e93d e41f8604d6a29dd

Note: 333 is the release candidate of 340 according to DaveBaumann

EDIT: I thought that it would be good to add this to the list CI. ;)

So... wait. The driver is listed as official because it doesn't breach rules, but this patch hits its FPS score, while not touching that of the Radeon? Uh...

[Xmas]

How about including info where it was found that the FX only does vertical offset AA, not horizontal offset axis AA. Yet another "feature" to create the illusion that their product is more efficient than their competitors in benchmarks.

www.jlmay.f2s.com/NV30-ATIAA.jpg
Apart from the image being poorly scaled, how do you get to that conclusion ("no horizontal offset AA")?

The 8xS patterns (there are two that could deserve that name) are well-known and do exactly what you'd expect from them. One is 2x2 OGMS with 2x1 OGSS, and the other is 2x RGMS with 2x2 OGSS.

Yes, it is worse than ATI's gamma-corrected sparse pattern (although I wish the gamma would be adjustable), but it does in no way create the illusion of being more efficient.