Hi guys,
I'm a visual effects artist and currently running a Windows 10 Pro system off a Ryzen 3900X, with 128 GB RAM, SSD and NVme drives and an RTX 2070 Super (I'd have gone for the RTX 2080Ti, but my money had run out at this stage! Looking to upgrade later this year though...)
I'm fairly new to this game, having only started about 3 weeks ago, and have been reading every resource that I can find with regards to Facewswap in particular, and this is by far the best site and best info I have found yet - congrats on an awesome site!
I've been running various tests with various models with varying results - it's all a trial and error learning curve, and I'm loving it! I have a few questions though...
I have a spare GTX Titan X lying dormant at the moment - would my current set up benefit from me adding this to it, for machine learning? Or would the Titan X effectively bottleneck the set up?
Or would I be better off upgrading later?
I'm not concerned with extraction times or render times, cos they are fast on my system anyway. But I would like to run larger models and have faster training times....(wouldn't we all??)
As for the planned future upgrade, it is worth me getting another RTX 2070 Super? And if so, should I run it with an NVLink, or just seperate?
Or should I wait and upgrade to a single GPU set up only (say the RTX 2080 Ti - or whatever new GPUs NVidia unveil later this year)?
I've heard conflicting views about running dual GPUs, and one article says that a second GPU would benefit, whilst another says not...! One article says another seperate - even different - GPU would be good, if it has more VRAM - whilst another says if you go the dual GPU route, they should be linked....
So many conflicting articles, it's confusing...!
Any help or advice regarding this would be awesome - in particular with my currently possible RTX 2070 Super/ GTX Titan X combo...
Many thanks in advance!