| ||||||||||
| One thing I have not been able to find any data about is what level of performance reduction one could expect in apps like Premiere Pro, Resolve and others that constantly pump video to and from the GPU for buffering and processing. Just looking at the architecture, it appears 4 lanes is inadequate. Of course with anything computer bus related: There is generally a larger discrepancy in theoretical bandwidth than real world application tests. Unfortunately, we don't have any of those as the gear to do them isn't yet available. Maybe there would be a way to gang a couple of Thunderbolt busses together and get 8 lanes and that would get us over the hump (or close). But those busses would probably have to be completely dedicated to that task, not laden down with other devices vying for the bandwidth. All of this development would come at a very high cost though, when we have a working, cheap solution with pci express slots on desktop motherboards. Taking that external without bottlenecking it is the challenge. The portability premium strikes again. Carey On Jan 16, 2012, at 2:13 PM, Tsassoon wrote: > Get a Mini :-) I'd very much like it if nVidia committed to building a Thunderbolt-attached CUDA engine. > > > Tim Sassoon > SFD > Santa Monica, CA > > Sent from my iPhone > > > > On Jan 16, 2012, at 11:45 AM, Satya G Meka <lists@rowbyte.com> wrote: > >> I hate those Glossy screens that come with iMac > > +---End of message---+ > To unsubscribe send any message to <ae-list-off@media-motion.tv> |
