gregorygarver.com

[MildredS]

hi .. i have a simple question about graphics cards . let me illustrate .. suppose i am buying a nvidia 6XXX graphics card and i have a configuration in mind ( eg - PCI - E , 256 DDR2 RAM , xyz clock speed etc ) . and in the store i find the same config from EVGA and XFX and others . can any one company change the basic architecture of the chipset such as number of pipelines . i know some makers factory overclock it but what about the base level design ? if the chipset is the same eg nvidia 6200 LE .. can one company sell it with 4 pixel pipelines and another with 2 ? please answer only if you know not to earn points and refer me to evga and xfx sites . any authentic answer is most welcome .. thank you ...

[ShadeSlayer]

everything other than the GPU itself for a given card series is generally fair game, though most card makers will use the reference design.. though you will generally find the less expensive cards in the same series will have 1) fewer pipelines (especially in a LE series card line).. 2) lower bit memory bandwidth as far as the memory bandwidth goes. as you look at different cards,your also looking at the bit width of the memory interface (be it 32, 64, 128, 256, or i've even seen some 8800gtx series cards with 384 bit memory bandwidth)I'll use the 6200's for reference.. just about all the 6200 series cards use either a 32 or 64 bit memory interface, with a rare few using 128 bit..as a general rule, the higher the bit number on the memory bandwidth, the better.. higher memory speed is great, until it gets bottlenecked by the lower bit memory interface..if your down to 2 possible cards of the same chipset GPU, and they both have the same number of pipelines, core and memory speed.. and are both within say $2 of each other, but one has a 32-bit memory interface and the other 128-bit.. I know i'd pick the 128-bit card every time