http://www.rage3d.com/index.php?cat=87#newsid33816077 look here, asus is readying to include a seprate PPU in the future
"We can't optimise our code so we've got to increase the power of the computers to make up for our complete n00b programing skills" :uhh: That is without a doubt one of the single most stupid ideas I have ever heard in all my life.
No really, its quite clever actually. SLI on the other hand is a stupid commercial idea..
There's enough power in the current generation of CPUs to take care of practically anything you might want to do with physics provided you program it properly, unless you're planning on screwing around with some really complicated stuff that only comes up in the sciences anyway.
Somewhat true (ever wonder why large simulations aren't run on UP desktops?), but software is always slower than a dedicated processing unit, because it has to be crunched by the CPU, which in turn introduces bus overheads. It's a very bright idea, and likely not much more expensive than nVIDIA's HSI chip to the end user. Besides, physical laws don't change much; once the chip is developed it needn't be updated unless to exploit advances in other technologies. Welcome to last month.
Everyone thought graphice processors were a STUPID idea until they saw the results they could bring... just wait until the next gen of physics engines are shown... they should make havoc-engine look like Atari. By the way, I dont think SLI is a stupid idea... assuming ofcourse you have an extra $600 laying around you dont need ;)
The question you have to ask yourself is this: How realistic can physics get? Once you get to a certain level, which I believe we are fast approaching, there will be no point to advance physics simulations any further because the end user will not notice the difference. Now with the current generation of CPU you have all the power you're likely to need, provided you optimise your code properly, to implement those physics engines.
Edit: And SLI is a stupid idea, you're lucky to top 140% performance for double the price.
140%!!!! IS THAT IT! that is bad, VERY bad
like i said... for some ppl an extra $500 means NOTHING, so for those ppl 140% is great
Well when you think that the method that 3dfx used back in the late 90s yielded much better results... No 140% is not great.