Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : Advanced Micro Devices - Moderated (AMD) -- Ignore unavailable to you. Want to Upgrade?

To: jcholewa who wrote (19366)	11/16/2000 3:36:04 PM
From: Pravin Kamdar	Read Replies (2) \| Respond to of 275872

I have a friend who is a lead programmer for a large game company. I sent him an email yesterday to ask him if they were planning on doing in SSE2 optimizations for their games. This is his response: ---------------------------------------------------------- No. Currently we do very, very, little assembly language programming anymore. Most of our optimization efforts are geared towards optimizing algorithms rather than hand coded assembly. Mostly because hand coded assembly is machine specific. We are trying to write our code to work on multiple platforms. PC, Mac, PS2, Linux, X-box. If the compiler takes advantage of it, and it produces a speed improvement, we "may" use it if there is some co-promotional deal with Intel. We have done assembly x87 instructions for our math library in the past but it hasn't changed performance all that much. I think we did a version for 3DNow and it improved our matrix operations much more significantly. We did not do an MMX version. We could not find any useful purpose for it. Currently we place most of our efforts on our algorithms, and doing special modifications for various video cards. Just by changing the way we do our fogging on some of the low end video cards, we can at least double our frame rate. By taking advantage of multiple texture units in some of the newer video cards, we can really crank up the speed. So there is less emphasis on the math part of our game, it occupies a small portion of our time-slice anyway. Even if we doubled its performance, we would probably only pick up another 7-10% speed increase at best. This kind of improvement would only be helpful on the low-end 300mHz systems that need all the help they can get. As long as you've got a decent video card, anything over a 400mhz system is overkill. It's all dependent on the video card. That's the bottleneck. In the not too distant future, the game geometry will be moved off to the video cards. Graphics and geometry take up a large portion of our processing time. About 15-20% is taken up by geometry. With this part of our game off-loaded to the graphics card, our games could probably run really well on a 200mHz system. By the way. At first I was kind of bummed that AMD didn't get the contract with Microsoft to produce chips for the X-box but now I'm kind of thinking that was a good thing. I don't know how much Intel is selling the chips for but they can't be making any money off the deal. These chips will be of their low-end chips and they're just stealing production bandwidth away from their more profitable chip sets. I've heard they increased the processor speed to a 733mhz system but if they have 2 or 2.4G systems by the end of next year and the first part of 2002, then these are going to be obsolete before they hit the shelves. Take care.