- Jun 8, 2005
Have you tried taking the watts even lower? At what point do you start losing ppd?Hey, I still have all my cards in. BUT. I decided to try underwatting them to save money Every card I have is running at 200 watts except the 3080TI is running at 250 (instead of 400) The 3070TI default is 350 ! The 3080TI is still getting 9 million ppd, and the 3070 is getting 5.3m. The 2080TI's default to 260, and they are all at 200. My total at the moment is still 46m (the same its been all race). So I am saving over 600 un-needed watts !
When I have time I may, but we are in a contest, and I did say I would be all in, so later maybe.Have you tried taking the watts even lower? At what point do you start losing ppd?
So that's a 20% cut in power (nice!), what was the drop in ppd?My newest rig, a 5950x + 3080TI was taking 625.Now its taking 500. This is at the wall kill-a-watt numbers. 125 watt savings on ONE box.
Nothing (statistically) I am still at 46-47 million ppd.So that's a 20% cut in power (nice!), what was the drop in ppd?
(Trying to find the thread where I tweaked my old RX 580 down for less power, but can't remember the thread now ).
Just realized we're now at less than 1/2 of [H] (and the official racers at roughly 1/3), so more than 1 lap behind...It seems to have settled for the moment, with the output of our entire team staying at 1/2 of [H]... Wow Does this count as being 1 lap behind, or is this more already?
Because in gaming (at default stock), they want the customer to see high fps and gaming stability and not so much about the watts consumed.Its been days now, and at 23% to 38% less power, my ppd has remained the same. Why do they put the initial power so high ?
Right now, my 3070TI is doing 5.4m on a 18201 at 200 instead of 325 watts.Because in gaming (at default stock), they want the customer to see high fps and gaming stability and not so much about the watts consumed.
For DC community, crunching efficiency matters to most of us unless you pay very little electricity cost. Here is an old report that I did back in Dec 2016 to investigate gpu power limit versus crunching efficiency. Back then you cannot do gpu voltage-frequency tuning with Maxwell cards. It could still give you additional savings in power. I use this in Windows system only. Can't find linux where I can set the voltage-frequency curve
For folding, because of QRB, I'm guessing the optimum folding efficiency lies somewhere around 70%+ range. If you see my other post here. on my 3070Ti, I set the power limit to 205W or (205/290 = 70.7% power limit) and my PPD is about -2.81% lower than the reported average.
View attachment 56668
The same card 3070 Ti and the same project (18201) that you posted in our forum, I assumed you run at stock (probably closer to 290W?) and your PPD is 10% higher than mine but consuming almost 30% more power.
View attachment 56669
That site is in California. (F@H) . EOC stats are in Houston Texas, both won't be touched by the noreaster.Stay safe!
Maybe the FAH website will go down at the same time, lol.
Not me! lol, tanking aside, my rig's output is the same as 2yrs ago (I'd not long upgraded CPU then, and haven't been able to buy a sensibly priced GPU since).Pretty amazing how much more points we all have managed!
|Thread starter||Similar threads||Forum||Replies||Date|
|Folding@home real PPD for 4090 for Windows and linux. ** edit ** 29m PPD Yes !||Distributed Computing||9|
|Folding@Home bigadv EOL 4th Anniversary Challenge: [H] vs TAAT||Distributed Computing||99|
|17th Folding@Home Holiday Race: The race is over and Mark's Marauders win.||Distributed Computing||251|
|S||boincgames.com Sprint, Si-Dock@Home||Distributed Computing||3|
|I||Question Any optimization on E@H or F@H? PG user.||Distributed Computing||4|
|Folding@home real PPD for 4090 for Windows and linux. ** edit ** 29m PPD Yes !|
|Folding@Home bigadv EOL 4th Anniversary Challenge: [H] vs TAAT|
|17th Folding@Home Holiday Race: The race is over and Mark's Marauders win.|
|boincgames.com Sprint, Si-Dock@Home|
|Question Any optimization on E@H or F@H? PG user.|