Plotting Folding System Efficiency

[Gorgon]

I recently acquired a couple of Eaton dual conversion UPSes and Network Cards. I've managed to hook them up to a couple of folding Rigs and I have achieved Stats Nirvana.
GTX 1070Ti and RTX2070

2 x GTX 1060 6GB

I query the PPD from the FAHClient on the system and take that value and divide it by the UPS output wattage via SNMP from it's Network Interface and then divide by my average cost of $0.13Cdn per kW.
So the two higher end cards are much more efficient and I'm averaging about $1/day per system in Electrical usage.

[kb5zue]

Is there a question in there somewhere?

 

[RollinLower]

1 hour ago, kb5zue said:

Is there a question in there somewhere?

 

not really, it's just juicy stats showing that eventough the higher wattage cards consume more power, they also produce more PPD per watt, so they are actually more efficient in F@H than the two low power cards.

[Gorgon]

1 hour ago, RollinLower said:

not really, it's just juicy stats showing that eventough the higher wattage cards consume more power, they also produce more PPD per watt, so they are actually more efficient in F@H than the two low power cards.

No, it’s just Stats p0rn  

 

FaH is kind of difficult to benchmark because of the length and variability of the Work Units.

Where I’m going with this to be able to break this out for a single card and run that card at different power targets to see if and where their sweet spots might exist for PPD/$.

 

In terms of just raw performance per Watt most Pascal cards are at the or highest efficiency at or slightly above the minimum power limit and after that point each incremental increase to the power limit results in a proportionately smaller decrease in the increase in performance to the point that most of the cards perform better in terms of PPD/W one or two steps below the maximum power limit.

 

but the Quick Return bonuses skew the data due to rewarding more points for WUs completed sooner.

 

My next steps will be to decrease the power limits further and see the effect on the system eifficiency. But in order to do that I need to split the PPDs out for each card and compare that to just that GPU’s power draw ignoring the system draw from the CPU, PCH, VRMs Case Fans etc. as well as where on the efficiency curve the power supply is running.