DOOM 3 & DX9 [Sep 17, 2003, 7:08 pm ET] – Share – Viewing Comments
With all the discussion lately about how Half-Life 2 will run on ATI hardware
compared with NVIDIA accelerators, BonusWeb.cz
shot off an email to id Software's John Carmack to ask how DOOM 3 and other
games are likely to be impacted by the difference between the two graphics
platforms running DirectX 9. Here's how he responded:
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.
Just to clarify all this precision stuff that seems to be confusing people:
Pixel Shader 2.0 spec is 24 bit minimum - with support for up to 128 bits (that technology's a year or two off, at least). This precision is a hardware issue; it goes beyond DirectX vs. OpenGL. Its more evident on DirectX because there are other things that DirectX does that exposes the disparity more than OpenGL; partly because DirectX evolves a LOT faster these days than OpenGL, so it has standard features that take advantage of new hardware a lot sooner (MS makes all the decisions it wants, which is good and bad; but OpenGL is a consortium, which means committees have to debate and compromise over months or years to get new features or code approved as "standard").
Anyways, you can run ATi at the 24bit spec. From what I understand, nVidia either runs at 16bit, OR 32bit (16 "doubled-up"). If you want to run PS 2.0 / DirectX9 stuff, you have to run 24bit or higher. Therefore, the nVidia cards suffer from a performance hit because of their inability to run the 24bit stuff natively. They have to "double-up" the 16 bit functionality and deal with the performance hit that this entails.
What Carmack has said, is that if you code things especially for "lower precision" (i.e. the 16bit mode that nVidia does by default), then the card works just as well as the ATI - but that running in "full precision" (24 or 32 bit mode), the nVidia will suffer a performance penalty.
In a nutshell, Carmack is basically saying that to get equal performance out of the cards, he's had to go out of his way to create a special set of code for the nVidia hardware; but the ATI can run "standard code" at a decent level. The amount of performance disparity isn't specified - but if it was 1 or 2 frames per second, I'd bet money that Carmack wouldn't have taken the time to write a custom graphics routine for it!
Glad I've got a little time to sit back and watch before I update my aging GeForce2!!
Take care,
--Noel "HB" Wade
Pixel Shader 2.0 spec is 24 bit minimum - with support for up to 128 bits (that technology's a year or two off, at least). This precision is a hardware issue; it goes beyond DirectX vs. OpenGL. Its more evident on DirectX because there are other things that DirectX does that exposes the disparity more than OpenGL; partly because DirectX evolves a LOT faster these days than OpenGL, so it has standard features that take advantage of new hardware a lot sooner (MS makes all the decisions it wants, which is good and bad; but OpenGL is a consortium, which means committees have to debate and compromise over months or years to get new features or code approved as "standard").
Anyways, you can run ATi at the 24bit spec. From what I understand, nVidia either runs at 16bit, OR 32bit (16 "doubled-up"). If you want to run PS 2.0 / DirectX9 stuff, you have to run 24bit or higher. Therefore, the nVidia cards suffer from a performance hit because of their inability to run the 24bit stuff natively. They have to "double-up" the 16 bit functionality and deal with the performance hit that this entails.
What Carmack has said, is that if you code things especially for "lower precision" (i.e. the 16bit mode that nVidia does by default), then the card works just as well as the ATI - but that running in "full precision" (24 or 32 bit mode), the nVidia will suffer a performance penalty.
In a nutshell, Carmack is basically saying that to get equal performance out of the cards, he's had to go out of his way to create a special set of code for the nVidia hardware; but the ATI can run "standard code" at a decent level. The amount of performance disparity isn't specified - but if it was 1 or 2 frames per second, I'd bet money that Carmack wouldn't have taken the time to write a custom graphics routine for it!
Glad I've got a little time to sit back and watch before I update my aging GeForce2!!
Take care,
--Noel "HB" Wade
Date
Subject
Author
Copyright © 1996-2025 Stephen Heaslip. All rights reserved.
All trademarks are properties of their respective owners.
Privacy Policy. Manage Cookie Settings.
News CGI copyright © 1999-2025 James "furn" Furness & Blue's News. All rights reserved.
Chatbear v1.5.0/blue++: Page generated 20 June 2025, 17:57.