I assume a PCI-E 2.0 x4 slot is going to limit a GTX 960 FTW for folding, but does anyone know by how much? It'll be freed up as of tomorrow so I can test legitimately then, but if anyone knows for sure that it'll be a huge hit, it might save me some time.

I would think for Folding it really would not make that much of difference.
That is also if you can get the Card to Work.
Run some Tests and Post Back!
 
What Motherboard or Computer are you thinking about doing this on?

if the gpu was a 980 or higher i would say it would be limiting. but a 960, i dont think anyone would be able to say. would like to know myself if you can post your results after a dozen work units or so.

It's my folding machine with a Supermicro X10SLM motherboard and an E3-1275L cpu. I've got a new 1060 3gb in the main x16 slot (x8 electrical @ PCI-E 3.0 speed), but when the 960 is freed up from my gaming rig tomorrow I thought it might still be able to contribute to the folding effort.
 
I've noticed that there's some variation between the PPD estimates during normal folding, what's the best way to compare a few runs between slots? Just post the logs? Or is there a CUDA benchmarking tool? I assume normal gaming-focused benchmarks wouldn't be too useful.

I would not Mix the Two types of GPU on that Motherboard with that CPU.
But then you other two Slots are x8 >> So you may want to run x8 x8 being Gen3
But Still Test and Post what your returns are.
 

 
 

Well I ended up not able to find the x8 to x16 adapter that I swore I had, so I put the cards in my garage rig for now and sent that motherboard back. Will end up picking up another board with more I/O for my folding rig after the TZC is done, for now it's good to go.

I saw a 4-6% drop on a 970 going from PCIe 3 x16 to PCIe 1 x16 ( same as 2.0 x8), so I would say you'd be down 8-10% on x16 performance