Single Channel vs Dual channel on GPU folding?

[ProDigit]

For non-CPU folding meaning GPU folding),
Did anyone test out if folding performance is affected when running single vs Dual channel memory RAM (not VRAM)?


I'm in the process of optimizing my server for FAH, by trimming my server, trying to optimize it for FAH, while keeping the electric bill low.
The latest addition of another 1060 has topped power consumption to 360W; 60Watts above my $1 a day target on electric bill, dedicated for FAH.

I've been able to save a few watts (about 25Watts) by disabling unneeded cores (going down from 10/20 c/t to 5/5 c/t), and pulling 2 of the 4 ram sticks (32GB to 16GB); switching to an SSD, and optimizing CPU, GPU and case fan speeds to nearly inaudible levels, while keeping the CPU temperature within acceptable ranges, without any real noticeable performance hit.

Now I think of further trimming down the RAM, by pulling 1 more stick, and running single channel, vs dual channel.
While this would definitely affect CPU folding (which I won't do when the temperatures are above 60F; 90-95% of the year here), I'm just wondering if it will affect the PPD of GPU folding in any significant way?

[bruce]

You've confirmed what would be consistent with how FAHCore_21 / _22 should work. Whenever possible, work is off-loaded to the GPU. It cannot run without support from one CPU thread, which includes some use of main RAM but it's a relatively tiny amount of the work that needs to be done. Faster RAM (including dual channel RAM) would be an insignificant improvement to FAH, although it might help other applications that you're running concurrently.

You might see a slight improvement for CPU folding