HTTP/1.1 404 Object Not Found Server: Microsoft-IIS/5.0 Date: Thu, 04 Jun 2009 11:05:23 GMT Cluster-Server: WEB1 P3P: CP="NOI ADMa OUR STP" X-Powered-By: ASP.NET Connection: close Content-Type: text/html

404 Object Not Found

HTTP/1.1 404 Object Not Found Server: Microsoft-IIS/5.0 Date: Thu, 04 Jun 2009 11:05:23 GMT Cluster-Server: WEB1 P3P: CP="NOI ADMa OUR STP" X-Powered-By: ASP.NET Connection: close Content-Type: text/html

404 Object Not Found

HTTP/1.1 404 Object Not Found Server: Microsoft-IIS/5.0 Date: Thu, 04 Jun 2009 11:05:23 GMT Cluster-Server: WEB1 P3P: CP="NOI ADMa OUR STP" X-Powered-By: ASP.NET Connection: close Content-Type: text/html

404 Object Not Found

GameSpy:
14851 users in the last hour. HTTP/1.1 404 Object Not Found Server: Microsoft-IIS/5.0 Date: Thu, 04 Jun 2009 11:05:23 GMT Cluster-Server: WEB1 P3P: CP="NOI ADMa OUR STP" X-Powered-By: ASP.NET Connection: close Content-Type: text/html

404 Object Not Found

PlanetQuakeHTTP/1.1 404 Object Not Found Server: Microsoft-IIS/5.0 Date: Thu, 04 Jun 2009 11:05:23 GMT Cluster-Server: WEB1 P3P: CP="NOI ADMa OUR STP" X-Powered-By: ASP.NET Connection: close Content-Type: text/html

404 Object Not Found

HTTP/1.1 404 Object Not Found Server: Microsoft-IIS/5.0 Date: Thu, 04 Jun 2009 11:05:23 GMT Cluster-Server: WEB1 P3P: CP="NOI ADMa OUR STP" X-Powered-By: ASP.NET Connection: close Content-Type: text/html

404 Object Not Found

Editorial Index

Recent Editorials:

I WANT my CD key!
12/13 - id's decision to use a CD key is justified

Report Card to the NIMF
12/1 - A response to the NIMF's report on violence in video games

Violence and Gaming
11/16 - Quake responsible for youth violence?

A Purist's Rules for FPS Multi-Player Design
11/5 - Keeping FPS' clutter-free

Rebuttal to Essobie's Editorial
10/15 - Grapple Controversy-Part Deux
The Woes of Being a Multi-Gamer
10/12 - Game Loyalty?
CTF != The Grappling Hook
10/7 - Q3 Arena sans grappling hook?
Jailbreak and Free Lunches
10/4 - Do mod makers "owe" people anything?
Pixels and Texels
9/13 - A look at the future of video cards
Yes, Camping is Evil!
9/2 - A response to "The Evils of Camping"!
Give Me Cable or Give Me Death!
8/31 - Will we all be LPBs one day?
Does Age Equal Maturity?
8/25 - A look at the age factor in gaming.
HeadHunting
8/23 - Mods and intellectual property
To Smack or Not to Smack
8/12 - Trash talking and the FFF!
The Evils of Camping
8/9 - We love to complain!
Trends in the Gaming Industry
7/13 - A look at the shift to multiplayer only games
32-bit Graphics Shows 3dfx's True Colors
7/12 - A continuation about the Voodoo3...
Is She 7 or 17?
6/30 - About the Voodoo3...
Doom 2000 and Q3A
5/26 - Fragmaster speaks his mind
The L33T D00D Multiplayer Tutorial
5/11 - Addressing their needs
Sue 'em All...
4/15 - The id Software Lawsuit
(more)

HTTP/1.1 404 Object Not Found Server: Microsoft-IIS/5.0 Date: Thu, 04 Jun 2009 11:05:23 GMT Cluster-Server: WEB1 P3P: CP="NOI ADMa OUR STP" X-Powered-By: ASP.NET Connection: close Content-Type: text/html

404 Object Not Found

Comments or ideas? Feedback?

32-bit Graphics Shows 3dfx's True Colors
A continuation of the previous editorial, "Is She 7 or 17?"
  — by
Roscoe Sincero

I was originally going to write this editorial for Inside3D. However, Inside3D is a primarily a programming site and not a hardware site and it would seem out of place. A hardware site like PlanetHardware would be a good place but I already submitted a similar editorial at PlanetQuake. This editorial is a continuation to the previous editorial. My focus with this editorial as with the previous is to enlighten people on how misleading 3dfx and their zealots are.

The ultimate purpose of editorials such as this is to provide some balance between reality and the claims made by 3dfx and their supporters. These supporters made a number of claims that are questionable. I have already discussed 3dfx's questionable leadership role regarding performance in OpenGL games in my previous editorial, Is she 7 or 17? The topic for this editorial involves the claim made by 3dfx and their supporters that 32-bit color support has a performance hit of about 20 to 40 percent. Is that really true or is there more to this story? What is the real story? And what are some of these tactics that they use to distract you from the real story?

In his discussion on 3d Accelerators, John Reynolds made this bold claim:

On the downside, to achieve V3-performance in multitextured titles usually requires the more expensive TnT2 Ultra, and some of its features, such as 32-bit output, have an undesirable impact on framerates-the Q3test, for example, takes upwards of a 40% performance hit when moving from 16- to 32-bit color depth on a regular TnT2.

In a recent interview by PCVelocity, Brian Burke of 3dfx had this to say:

If you look at numbers of cards that are overclocked and are not shipping, the 32-bit numbers look OK on a TNT-2. But at shipping clock speeds they are not as impressive and drop between 20 and 40%. These are significant performance hits.

Be honest with yourselves. How many of you believed them?

The scary part in all of this is that I do not see anyone actually verifying the claims made by these people. The data that is freely available for everyone to view clearly contradicts their claims. When you look at Brian Hook's Benchmarks, for example, you will arrive at a completely different conclusion from the 3dfx supporters. The Elsa Erazor III is a non-Ultra TNT2. This is not an "overclocked" TNT2. At 640x480x16 on a Celeron 400MHz, the frame rate Mr. Hook reported was 36.3. At 640x480x32, Mr. Hook reported 35.3. When you look at PlanetRiva's Kingpin Demo Benchmarks using the Diamond Viper V770U, you do not see any performance hit greater than 10% even at 1280x960. Either we have smart individuals not being able to do simple math or we have individuals in our mist as well in the industry spoon feeding us horse manure. I'm pretty sure smart people can do simply math.

The funny part is the excuses that these 3dfx supporters conjure up to explain the contradiction. You will be bombarded with information involving Z-buffer, memory bandwidth, and all kinds of good stuff. Most if not all of what they say is true. However, it does not address the question. They give you a long, detailed, and knowledgeable-looking response but it does not answer the question. They are actually answering a different question. If they were asked, "why do we not see a 20% performance hit at 640x480 with 32-bit color support?" then their answers would be fine. However, they were not asked such a question. The real question is "why are people claiming a 20 to 40% performance hit when the data clearly shows otherwise?"

It is true, however, that you can have a large performance hit with 32-bit color. In general, this is limited to high resolutions. You often do not see this over-exaggerated performance hits at lower resolutions. What we have here folks are examples of individuals with an agenda looking at only a small part of the data to make a broad conclusion. They are giving you worse case scenarios; they are essentially trying to scare you. It is like a reviewer benchmarking the Voodoo3 and notes that the V3 only gets 25 FPS at 640x480x16 using the mon2.dm2 benchmark and 16-bit textures. The reviewer then benchmarks the V3 using 8-bit palletized textures and notes that the V3 gets 75 FPS. He then proudly proclaims that 16-bit textures are a big fat waste. There is a 66% performance hit with 16-bit and that is way unacceptable. We should all stick with 8-bit palletized textures until we have the hardware to use 16-bit with little loss in performance. Do you think that is right? This is exactly what they are doing with 32-bit color performance.


HTTP/1.1 404 Object Not Found Server: Microsoft-IIS/5.0 Date: Thu, 04 Jun 2009 11:05:23 GMT Cluster-Server: WEB1 P3P: CP="NOI ADMa OUR STP" X-Powered-By: ASP.NET Connection: close Content-Type: text/html

404 Object Not Found