Who can explain video bandwidth and core clock speed?

[Dputiger]

Ok: I understand that on a GeForce 2 with 200 Mhz memory (400 Mhz effective) you have a maximum fill rate of 400 * 16. (400 Mhz * a 16-byte or 128 bit data channel).

How does the core speed play into this equation at all? What good does raising the core speed do, in theory--and is it necessary on today's video cards at all?

[Aleph1]

I think its core speed times the pipelines...a geforce2 has 4 pipelines, so 200*4 is 800, hence the 800 million pixels per second fill rate. =)

[Dputiger]

Ahh, ok....So, let's see...four pipelines, 200 Mhz, yeah, 800 million pixels per second.

This makes sense.

However why would Memory bandwidth than be a constrictor? The Memory bandwidth on a GeForce Pro is 6.4 GB/s!

Unless...how much memory bandwidth does a certain amount of pixels take up? I guess that could affect it.

[Win_98]

btw: what do they mean by 250mhz ram dac?

[Luser]

RAM DAC stands for Random Access Memory Digital to Analog Converter. Basically that's the part that reads your video memory and puts it on the screen. The higher the frequency of that chip, the faster it can throw pixels at the monitor, so the higher resolutions and refresh rates it supports.

[Luser]

The speed of the memory is a limiting factor because the GeForce chip has to read and write texture data, frame buffers and z buffer info along with other stuff to memory. That first calculation is just a theoritical maximum of that chip. In reality, you'd never get anywhere near that because of the latencies involved with the reading and writing to RAM.