Log in

View Full Version : CUDA acceleration, MADVR and video card bandwidth


myworldoftech
21st October 2011, 17:38
I know games don't suffer too much when going from pcie x16 to x8 or even lower, nor does version 1.1 vs 2.0 matter much.

Does the same apply for CUDA decoders?

What about MADVR resizers?

I was thinking of building a mini itx htpc. Some of the mainboards I was considering are x4 (electrical).

Mangix
22nd October 2011, 03:46
i think version 2.0 has lower latency and 3.0 even less. the number of lanes should not matter. i know lower latency can reduce microstutter used in multi-gpu sli/crossfire setups but am unsure if this applies to cuda and madvr.