eco, I was just going to post that. Aces picked up on it in aceshardware.com, I'll drop a few more quotes.
What I am really looking forward to is when the PC will be a truly high-performance
system with a lightning-fast bus and a unified memory architecture. Opening up the
3D-hardware to be really programmable and having the CPU / GPU and all other *PUs
work together with full access by your code. There are alot of really stupid things going
on right now, with the current PC-architecture, that needs to be solved.
That was Jim Malmros explaining why he thinks Nvidia and AMD are going in the right direction. Gotta get to nForce, but mercifully that seems to be pretty much on track. On the amusing side, there was this:
My main workstation is a dual P3/933 w/ PC600 512MB (ugh) RDRAM. Why the lame
config? Bought it 7 months ago when RDRAM was stupid expensive but I still wanted
dual processors, so I compromised by getting cheap RAM. In another few months I'll
probably get an Athlon 1.5Ghz w/ nForce motherboard and 1GB of RAM as my main dev
machine.
That was Brian Hook, apologetic AMD shareholder and presumably one of Elmer's AMDolts.
In developing games, we want to make the games as widely accessable
to as many game players as possible, while still staying on the leading edge of
technology. So, the base game should be developed WITHOUT any of these special
instructions in mind and if time permits -- and the game warrents it -- special code can
be added to optimized the game specifically for those chipsets.
That was Chris Rhinehart, pointing out the somewhat obvious primary flaw in Intel's "Backward compatibility doesn't matter- just recompile" argument for the P4. Of course, there's the related flaw in that argument that there's very little to recommend the x86 ISA besides backward compatibility in the first place, but let's not get in to that. It's all NetB**s* to me. |
