OC-Master
10-20-01, 07:45 PM
A good friend of mine who used to work over in silicon valley a few months back but moved back to e-town told me how graphics technology would change in the next two years in a BIG way.
First off, remember when the GPU (Graphics Processing Unit) was introduced by nVidia? This was of course the GeForce 256. It wasnt the massive increase in framerates that ohhed everyone but the alot less CPU usage that was used. nVidia was able to cut the CPU usage in half by completely channeling the floating point calculations to the AGP card. And this was why we got the GPU.
The next generation of graphics by ATI and nVidia will have GPU2 technology. This will once again strengthen the CPU by offloading even more floating point calculations and now wire frame calculations to the GPU.
To explain a quick demostration of this technology, John Carmack, (CEO) of ID Software had the following machine,
AMD K6-2 550MHz
128MB Ram (PC100)
nVidia Eclispe prototype (NV30) & ATI R300 prototype (10000) AGP
Sound Blaster Audigy
...
With this system, he was able to attain 60~85fps steady at 1600x1200x32bit @ 85Hz with Quake 3 Arena using the latest build. Using the current GeForce 3/R200 architexture, these results are far from possible! due to the CPU being strained to much and not being fast enough to calculate with a higher floating point that is needed. To tell you how powerful these new graphics cards are, the above system would replace the CPU of an Athlon @ 1GHz. That means that the GPU is accelerating over 450MHz extra, purely impressive!
Gamers with less than a GHz CPU will have a very nice alternative path for upgrading and rather than throwing out there current system, they just need to replace the current graphics card with a next generation card.
Finally, the only bad thing about this high amount of floating point speed is that it takes alot of bandwidth, something the AGP bus 2X and even 4X will begin to strain. To take full advantage of GPU2, one will have to have the AGP transfer rate equal to the performance of the Ram on the motherboard. This will even the performance between the GPU and Ram and therefore, no strain will be there. But almost all of todays computers come equiped with at least 4X so this all shouldnt be a problem.
(OK, i dunno any specs on Eclipse or R300 except for that there both 6 pipe engine GPUs!)
I hope this helps you understand graphics technology of the future, I tell you, I'm impressed! but I bet Intel and AMD Arnt!
Cheers All!:p
First off, remember when the GPU (Graphics Processing Unit) was introduced by nVidia? This was of course the GeForce 256. It wasnt the massive increase in framerates that ohhed everyone but the alot less CPU usage that was used. nVidia was able to cut the CPU usage in half by completely channeling the floating point calculations to the AGP card. And this was why we got the GPU.
The next generation of graphics by ATI and nVidia will have GPU2 technology. This will once again strengthen the CPU by offloading even more floating point calculations and now wire frame calculations to the GPU.
To explain a quick demostration of this technology, John Carmack, (CEO) of ID Software had the following machine,
AMD K6-2 550MHz
128MB Ram (PC100)
nVidia Eclispe prototype (NV30) & ATI R300 prototype (10000) AGP
Sound Blaster Audigy
...
With this system, he was able to attain 60~85fps steady at 1600x1200x32bit @ 85Hz with Quake 3 Arena using the latest build. Using the current GeForce 3/R200 architexture, these results are far from possible! due to the CPU being strained to much and not being fast enough to calculate with a higher floating point that is needed. To tell you how powerful these new graphics cards are, the above system would replace the CPU of an Athlon @ 1GHz. That means that the GPU is accelerating over 450MHz extra, purely impressive!
Gamers with less than a GHz CPU will have a very nice alternative path for upgrading and rather than throwing out there current system, they just need to replace the current graphics card with a next generation card.
Finally, the only bad thing about this high amount of floating point speed is that it takes alot of bandwidth, something the AGP bus 2X and even 4X will begin to strain. To take full advantage of GPU2, one will have to have the AGP transfer rate equal to the performance of the Ram on the motherboard. This will even the performance between the GPU and Ram and therefore, no strain will be there. But almost all of todays computers come equiped with at least 4X so this all shouldnt be a problem.
(OK, i dunno any specs on Eclipse or R300 except for that there both 6 pipe engine GPUs!)
I hope this helps you understand graphics technology of the future, I tell you, I'm impressed! but I bet Intel and AMD Arnt!
Cheers All!:p