Halflife 2 --> Officiell Tråd

My e-mail to Gabe:

quote:
--------------------------------------------------------------------------------
Hi Mr. Newell (or may I say Gabe?),

first I would like to say I'm a big fan of your work with Valve on HL & CS, I've been playing both from the very start back in 1998 (CS in1999), And I can't wait for HL2 (& CS2), keep up the good work..

My question is about videocards, you probably get alot of these but I wanted to ask you anyway because it's important to me, HL2 will probably my next primairy game for the next 5 years orso..

Here we go:

Is a ATi 9800pro card really alot better for HL2 then Nvidia's FX5900? Or is the difference not that big (quality & fps wise)?

Articles like this worry me:
http://www.beyond3d.com/misc/traod_...f/index.php?p=1

I used to go for Asus cards all the time because I really like their in-game OSD.. And Asus only makes Nvidia cards, so I was going to get a V9950 FX5900 especially for HL2, but if you say that's a mistake I will go for ATi this time.. Hope you find the time to answer my question..

Greetz & Thxs,

Menno de Kok
The Netherlands
http://www.cyberwarriors.nl
--------------------------------------------------------------------------------

Gabe's answer:

quote:
--------------------------------------------------------------------------------
I have been a long-time NVIDIA card user. Currently I have ATI 9800 Pro's in both my work and home machines.

The DX9 performance described by the Beyond3D article is consistent with what we've been seeing.
--------------------------------------------------------------------------------

So it really looks like Nvidia screwed up with DX9 performance, and I have to let my favorite combination Asus/Nvidia go.. But to be absolutly sure I'm gonna wait for the HL2 benchmark that was suppossed to be released before HL2.. We are talking more then €300,-- here, so I want to make 100% sure I get the right thing..

Beyond3d article
1) The NV35 has 4 pixel pipelines with 2 texture sampling units (typically called a "4x2" architecture). This is witnessed by the "FFP - Pure fillrate" (400MHz * 4 pixel-pipes) and "FFP - Dual texture" (not a great deal of loss from the "FFP - Pure fillrate" stats, due to the 2 texture sampling units) figures.
(2) The NV35 has to take 2 clock cycles when asked to execute every two PS 2.0 instructions (PS 2.0 - Simple result is about half of the PS 1.1 and 1.4 ones, and the "PS 2.0 - Longer" is about half of "PS 2.0 - Simple").

The above comes into play in actual PS 2.0-featured games. The main thing is that while the NV35's 4x2 architecture really shouldn't be at any disadvantage compared to a "8x1" architecture (like ATI's R3x0) when we have situations where pixel shader instructions are short and simple, there is a potentially significant impact when instructions increase... and the NV35 has to take extra cycles in order to execute them. Bear this in mind when you look at the Tomb Raider AOD benchmarks later in this review.

Ursprungligen inskrivet av deli
nej de måste man inte de är ju bara klicka på den så kan man läsa den..