TPU's F@H Team

[NastyHabits]

Has anyone been having any issues with getting work units again today at all?? Tried to download a few on a couple PCs but nothing downloaded at all... Must be needing a recharge or something I guess??

@phill It's a Sunday and a Holiday here in the US. Not an unusual occurrence.

 

[Space Lynx]

I might be over thinking it I guess @NastyHabits but when you click here.... you'll see what I mean... I wonder if the site is reading the results right... Sometimes things aren't quite what they seem to be and results and such get messed up... Any one else's experiences and such, please drop a line here Love to hear opinions about the points and results kicking about

It'll have been lovely to have such a nice warm temp to be outside with, The UK sucks for weather and temps

I love the UK weather. It helps my eyes for some reason. I have severe extreme dry eye disease. Most days I just lay in bed with moisteners on my eyes at least 30 mins every day.

I tried some really expensive eye drops doc gave me, but they made my face break out so I stopped taking those.

 

[debs3759]

I have a GTX 580 lying around doing nothing. Is it able to fold in Windows 10 x64? Don't want to waste my time setting up a system with it if it's no longer supported. If it's not supported, I'm wondering whether it'll work for wcg. The alternative is a less powerful GT 730 (Kepler), which I guess is not worth folding on.

 

[phill]

I'd guess @debs3759 the 580s will just be hot and not score so well... They can get quite toasty and they aren't the most efficient of cards... I'd try something newer to be honest if possible

 

[debs3759]

That's good to know, thanks. I'll leave it for benching and use the 730 (without folding) until I can afford a 3060 or whatever the next gen is when I save enough. Going to crunch on a couple more systems soon anyway Should have my collection better organised soon.

 

[phill]

I'd personally go with as new as possible without going daft with the price... From my 1080 TI's to my 3090 I have currently, the performance is pretty decent from an upscale point of view. The two TI's would give me between 3 to 3.5 million a day (if I left them on all the time that is) but the 3090 I have seen 5 million plus figured but again, the 3090 does suck more juice than the two 1080 TI's believe it or not...

I think the 3060/3070 range of card is a perfect choice and for one simple reason... You won't have to mess about with re-laying thermal pads down or anything with the higher models and they are generally not a million miles away from the 3080 performance anyways I think. The extra cost for the cards right now is a definite no no, but that's where I'd be headed to I think

Do let us know what you manage to get hold of and so on I'd also stick to the cards with the good aftermarket coolers. Strix XC3/FTW3 I think are some that come to mind, just watch the pricing as they tend to shoot up with those names on the side which is heart breaking....

 

[debs3759]

FTW3 is probably what I'll opt for. Mainly because I know EVGA use reference design cards, so waterblocks will be easier to find (although not sure if there are blocks for a 3060, I haven't checked that yet). It'll probably be the end of the year before I can afford a decent card anyway, all my money currently is being spent on shelving for my large collection, which has been very disorganised

 

[phill]

I'm finding as well as long as you can trim the power that the card pulls, temps are very reasonable and they work fine, you don't normally need to have extra cooling and such I think the 3080/3090's aside with their mega hot GDDR6X, they can become more of a problem than they should do....

 

[dogwitch]

I'm finding as well as long as you can trim the power that the card pulls, temps are very reasonable and they work fine, you don't normally need to have extra cooling and such I think the 3080/3090's aside with their mega hot GDDR6X, they can become more of a problem than they should do....

if you want good read on power etc. prime grid. they put warning on some of there work units

 

[phill]

if you want good read on power etc. prime grid. they put warning on some of there work units

Do you have a easy link so we could have a look please @dogwitch ?

 

[NastyHabits]

Do you have a easy link so we could have a look please @dogwitch ?

if you want good read on power etc. prime grid. they put warning on some of there work units

Yes please @dogwitch

 

[dogwitch]

PrimeGrid Message boards

it per type of work unit.
like
DO YOU FEEL LUCKY? wu.
2 to 3 day work unit for a 2080 or above. also you need very very good cooling on gpus .
i ran 2 wu and was pulling north of 600 on dual 2080s. and 280 on my cpu for a cpu base wu.
32 core 64 thread wus. those where 7 day ones

 

[mstenholm]

@NastyHabits and @phill misunderstood you I think. They thought that you (Primegrid) have any information about F@H WUs which of course they don’t. I could of course be the one that misunderstood it all

 

[phill]

PrimeGrid Message boards

it per type of work unit.
like
DO YOU FEEL LUCKY? wu.
2 to 3 day work unit for a 2080 or above. also you need very very good cooling on gpus .
i ran 2 wu and was pulling north of 600 on dual 2080s. and 280 on my cpu for a cpu base wu.
32 core 64 thread wus. those where 7 day ones

I think I'm going to need a bit of time with that link to make some sense from it and at 12:40am nearly, I'm not sure that'll be happening tonight!

 

[dogwitch]

I think I'm going to need a bit of time with that link to make some sense from it and at 12:40am nearly, I'm not sure that'll be happening tonight!

lol if you want to cook your card. that the software to do it with.
or to make sure it works from a online buy.
btw i was not kidding on power draw ether.
seeing said item. make a gpu bleed. if it can pass that. folding a breeze!!!!

 

[phill]

lol if you want to cook your card. that the software to do it with.
or to make sure it works from a online buy.
btw i was not kidding on power draw ether.
seeing said item. make a gpu bleed. if it can pass that. folding a breeze!!!!

Sounds like another Furmark test lol!! Do you run them on your rig/s @dogwitch ??

 

[dogwitch]

Sounds like another Furmark test lol!! Do you run them on your rig/s @dogwitch ??

fur mark ok. but run the multi day wu that are double precision both on cpu and gp. normal fill 8gb per wu. cpu floats around 8 to 10 gb for cpu wu.

 

[mstenholm]

Could we keep F@H separated from other GPU supported distributed computing? If @dogwitch want to discuss PrimeGrid then start a thread about that.

 

[dogwitch]

Could we keep F@H separated from other GPU supported distributed computing? If @dogwitch want to discuss PrimeGrid then start a thread about that.

it was relating to testing temps memory. folding does not push the memory as hard.
so i was ref other thing that does. that all. nothing more.

 

[Space Lynx]

Google details its protein-folding software, academics offer an alternative

Once computationally impossible, AIs now translate protein sequence to structure.

arstechnica.com

just wondering, but does this kind of stuff make folding at home irrelevant? now that Google is getting into folding, I mean there is there anything the little guy can do to contribute or no?

 

[80-watt Hamster]

Google details its protein-folding software, academics offer an alternative

Once computationally impossible, AIs now translate protein sequence to structure.

arstechnica.com

just wondering, but does this kind of stuff make folding at home irrelevant? now that Google is getting into folding, I mean there is there anything the little guy can do to contribute or no?

One wouldn't think that DeepMind would completely obviate distributed folding projects. I read the article, but neither source paper, and my limited understanding of these things suggests that even Google/Alphabet won't have the resources (or the willingness to commit them) to scale DeepMind to take on every existing and future folding problem. It also seems like protein folding simulation is a bit of a goldfish problem: No matter how many you solve, there are always more waiting. Not to mention that, presumably, not all projects will get approved for DeepMind time, and the question of whether a particular project is compatible.

TL;DR: Less relevant perhaps, but probably not irrelevant.

 

[1freedude]

It was a good run, but I've shut down the rigs. Please hit me up directly for any challenges or competitions. My 24/7 contributions are done for now.

 

[dogwitch]

It was a good run, but I've shut down the rigs. Please hit me up directly for any challenges or competitions. My 24/7 contributions are done for now.

guessing power cost?

 

[debs3759]

Received an ROG Strix 1080 Ti OC this morning, that was gifted me by a folder in the folding forums. Finishing my current work in f@h and wcg, then going to test and bench it, hopefully getting it folding by tonight. I can't find the link to the site that lists what different cards can generate, but I expect to be getting over 2 million ppd by tomorrow (replacing the 1060 3GB, which has been averaging around 600K). Guess I need to put a few more ££ aside for the electric bill

 

[mstenholm]

Received an ROG Strix 1080 Ti OC this morning, that was gifted me by a folder in the folding forums. Finishing my current work in f@h and wcg, then going to test and bench it, hopefully getting it folding by tonight. I can't find the link to the site that lists what different cards can generate, but I expect to be getting over 2 million ppd by tomorrow (replacing the 1060 3GB, which has been averaging around 600K). Guess I need to put a few more ££ aside for the electric bill

Well you had to go back a whole page in this forum to find such a list GPU F@H list

If you are aiming for +2M then it seems like you have to switch to Linux but 1.9M isn't bad either. There are big variations, my own 2070/Win 7 does 1.6 to 2.4M. Adding the beta string (client-type, beta) under Config, Expert normally bring some of the more "lucrative" ones but ATM that is on the expense of used CPU memory. The PC mentioned above only have 8 GB and using 1.8 GB for one job is to much for me.