Video Card Description

[Josh]

Okay first of all, if you have no idea what the heck I'm talking about don't reply...

I have a good graphics card in my computer. A Variant of the 6800.

Here is the specs:

256MB memory

256-bit memory

325mHz Core Clock

700mHz Memory Clock

12pp, 5vp

Okay now I know my way through a computer, but I wanted to get some things about the video card cleared up. Anyone who is willing to describe these specs and what each actually does for my gameplay are more then welcome.

I know the memory stores the video textures.

I don't know what the bit of the memory benifits.

The core i'm not sure completely.

The memory clock I believe affects performance greatly but i'm not sure how.

And the pp and vp, i dont know what they offer.

[galespurehoney]

Okay first of all, if you have no idea what the heck I'm talking about don't reply...

I have a good graphics card in my computer. A Variant of the 6800.

Here is the specs:

256MB memory

256-bit memory

325mHz Core Clock

700mHz Memory Clock

12pp, 5vp

Okay now I know my way through a computer, but I wanted to get some things about the video card cleared up. Anyone who is willing to describe these specs and what each actually does for my gameplay are more then welcome.

I know the memory stores the video textures.

I don't know what the bit of the memory benifits.

The core i'm not sure completely.

The memory clock I believe affects performance greatly but i'm not sure how.

And the pp and vp, i dont know what they offer.

the memory stores the video textures, 3D setup and vertex information (scene layout) , and the screen buffer (an in memory image of the final drawn screen that the RAMDAC reads and pumps to your monitor). so is used for quite a few bits and pieces.

the 256bit part of the memory is the width of the data channel the GPU has to the memory. 256/8 = 32bytes of data can be read or written on each tick of the memory clock. so 32bytes x 700,000,000ticks of the clock per second (700Mhz) = 22,400,000,000bytes per second, = 21875000kilobytes per second, = 21362Megabytes per second, = 20.8Gigabytes per second bandwidth of data between the GPU and the GFX memory.

so if you boost the memory clock, you boost how fast the data gets to and from the GPU. So when throwing lots of textures around, and 3D scene data, this can only be a good thing.

the GPU core clock is how fast the GPU runs. how fast it can process geometry and vertex data to set up the 3Dscene, how fast it can paint the 3D scene with textures, and how fast it can apply aftereffects like filtering, bilinear, trilinear, and antisotropic... or other things like antialiasing etc.

on a side note, antialiasing is very memory bandwidth dependent, provided the GPU is fast enough to generate the anitaliased content fast enough

cheap cards usually have a decent chip, but have alot of features sections diabled, plus have a crappy memory bus (128bit, or even 64bit) and slow (cheap) memory chips, and therefore a slow memory clock, throttling the chip down to "cheap card speed"

med range cards usually have a high end chip, but with less sections disabled, and have a better (256 or 128bit) memory bus, and faster memory chips - giving you med range speed.

high end cards, have all the features (usually pipes) of the GPU enabled, and coupled to the fastest memory over the widest memory bus (256bit ,320bit ,384bit , or even 512(!)bits wide)

i hope this helps.