Folding Community Board

[GOTSpectrum]

Has anyone tried a mining card for folding?

[azmariadei]

1 hour ago, GOTSpectrum said:

Has anyone tried a mining card for folding?

If I could get one of those headerless 1060s they used for mining, I'd love to drop it into my server and give it a try.

[RollinLower]

8 hours ago, GOTSpectrum said:

Has anyone tried a mining card for folding?

there's some guy selling a mining RX 470 for cheap. pretty tempted to pick it up now, see how it compares.

still, a 470 isn't really a PPD beast.

[Guest]

 

Whoa... 6.6M from my dual 2080 Ti rig.

I like this Core 22.

[AthlonS2K]

So this winter instead of running 4 electric wall heaters in our apartment, we have decided to fold@home with our Steam/Minecraft Server.  Its really just a dell precision T5600 with a second xeon e5-2667 added and 64gb of ram. The appartment is warmer not chilled at the edges any more so much more cozy. Our skin isnt as dry and we get to help look for cures and be part of something bigger.  

 

Today I decided to put all that cycle time to a team, and felt what better community to join than this one.  

[justpoet]

5 hours ago, AthlonS2K said:

So this winter instead of running 4 electric wall heaters in our apartment, we have decided to fold@home with our Steam/Minecraft Server.  Its really just a dell precision T5600 with a second xeon e5-2667 added and 64gb of ram. The appartment is warmer not chilled at the edges any more so much more cozy. Our skin isnt as dry and we get to help look for cures and be part of something bigger.  

 

Today I decided to put all that cycle time to a team, and felt what better community to join than this one.  

Awesome!  Welcome!  

[RollinLower]

12 hours ago, AthlonS2K said:

So this winter instead of running 4 electric wall heaters in our apartment, we have decided to fold@home with our Steam/Minecraft Server.  Its really just a dell precision T5600 with a second xeon e5-2667 added and 64gb of ram. The appartment is warmer not chilled at the edges any more so much more cozy. Our skin isnt as dry and we get to help look for cures and be part of something bigger.  

 

Today I decided to put all that cycle time to a team, and felt what better community to join than this one.  

welcome to the team, glad to have ya!

[sanya567xxx]

Need help fixing this on a testbench system.. nvidia's gpu is running fine, this is lubuntu 19.10 with kernel 5.4.0, dkms fails when I try to install regular amd driver so I tried doing this and they did install fine but I can't really.. test the card. Also nvidia-settings don't work because I have that amdgpu driver installed [partially]. The latest script ( https://gitlab.freedesktop.org/drm/amd/issues/26 ) worked but I've no opengl/opencl stuff installed to even test the capabilities. As you can see, the gpu is detected [lspci does see it too].

[Metallus97]

BOOOOOOOI the C22 high Atom count WUs push my VIIs like 50% more it pulls close to 2,3mPPD on stock speed . But the low Atome Count ones have like 10% degradation in PPD

[Metallus97]

Sorry for linking myself but didn't want to crospost!

 

Update About Core22 moving to Advanced and new (testing) projects for Core22:

 

 

[GOTSpectrum]

28 minutes ago, Metallus97 said:

Sorry for linking myself but didn't want to crospost!

 

Don't worry about it man, glad to see more info on C22 floating about. 

[sanya567xxx]

also apparently this linux system running core 22 11738 only gets 1.2-1.3m, a windows system I have with core 22 on 11379 gets 1.4-1.5m, and my "unlocked card" - eg best case scenario - is getting 1.7 and on different projects I've screenshot 2.2m, although all 3 cards are 1080tis that don't thermal throttle and run at 1978, 2012, and 2038MHz (testbench, other system, my unlocked one respectively), with memory being about the same freqs too for each of them. It looks like either the cpu is too weak on the linux testbench (it's an i3 2120 running like 3.2GHz only, this would partially explain why the second system's card isn't doing as good as the unlocked one - the difference is too big to be just a couple MHz of the card, the unlocked has a 5GHz/4.6GHz AVX i5-9600k to work with, while the second system only has a 3.4GHz i5-6500) or there's something to do with how projects are different and there's power limits enabled/disabled on different systems..

 

oh wait I think I know the reason and it looks like it isn't the cpu affecting (although it might affect a little too)

That's god damn 270 watts of power being ate by this thing. On linux the default powerlimit is 250 and it only "eats" 235, with the "other" system getting about 250w consumption.

[justpoet]

3 hours ago, sanya567xxx said:

oh wait I think I know the reason and it looks like it isn't the cpu affecting (although it might affect a little too)

That's god damn 270 watts of power being ate by this thing. On linux the default powerlimit is 250 and it only "eats" 235, with the "other" system getting about 250w consumption.

Clearly you need a Binford upgrade.

[sanya567xxx]

5 hours ago, justpoet said:

Clearly you need a Binford upgrade.

you mean this? 

[justpoet]

11 minutes ago, sanya567xxx said:

you mean this? 

 

[GOTSpectrum]

1 minute ago, justpoet said:

 

Oh my god I forgot this even existed... I need to watch this... like now....

[TVwazhere]

On 1/4/2020 at 4:11 AM, yaboistar said:

Interestingly, pascal takes a hit when running P11738 and drops back down to C21 levels (1.3mppd) whereas P11737 runs at about 1.6mppd

 

i have not yet been able to test Maxwell on this high-atom-count C22 unit but i'll update when i have, i expect it to follow Pascal's trend

Damn, maybe I should have gotten 2060's instead of 1070ti's after all

[TVwazhere]

5 minutes ago, yaboistar said:

yeah, you should've. turing architecture is freakishly good in FAH's compute compared to pascal

I got my 1070ti's at a very good price so I'm not completely hating myself for it, but hearing they;re getting 50% increases on Core 22.... Damn. 

 

Think I should try moving to the beta core anyways? Or stick with Core 21?

[RollinLower]

1 minute ago, TVwazhere said:

I got my 1070ti's at a very good price so I'm not completely hating myself for it, but hearing they;re getting 50% increases on Core 22.... Damn. 

 

Think I should try moving to the beta core anyways? Or stick with Core 21?

i've been on beta with my 1070's for a little while now and PPD hasn't increased at all. 

i wouldn't recommend on Pascal (yet).

[RollinLower]

is there anything worse that WU's getting stuck on send?

if you're gonna fail a WU atleast don't make me run the whole thing first!

[MrFinn741]

On 1/3/2020 at 2:50 PM, GOTSpectrum said:

Has anyone tried a mining card for folding?

Yes, I have a p104-100 (something like a 1070 running at 2GHz with 4GB GDDR5X)  producing 650-677K PPD.

[sazrocks]

12 minutes ago, yaboistar said:

there is one server that seems to intermittently decide it doesn't want to accept uploads to it

Yeah sounds like the TempleU servers that have been having issues on and off for the last 5 months or so. They apparently ordered a new server at some point but I guess they haven’t gotten around to installing it yet. I’m not really sure why they don’t just spin up some cloud server to pick up the slack in the mean time considering how large F@H is.

[sazrocks]

Just now, yaboistar said:

if what's up thread is true and they can't afford to keep their net certificate up... i don't think more cloud storage is ideal. i know that some of their servers are actually hosted on the cloud (amazon AWS i think??)

Different orgs AFAIK. The TempleU servers are owned and run by TempleU staff, not whoever is in control of F@H after Stanford stopped managing it.

[mattheginger]

I’ve just returned from a break from folding due to moving, Christmas travel and a house that’s just a wee bit warm, and I come back to a bump of 30% ppd on the new core betas!

 

Im stoked. Are we expecting more gains and streamlining from this core? Or is it pretty much just the same performance from a core from creation to decommission?

[sanya567xxx]

Decided to test the difference of using different PCI-E slots on my B75 mobo i3-2120 in the linux machine, 1080ti. around 1.9m PPD for x16 slot (this is PCI-E Gen 2) and only 1.6-1.7m on the 4X slot. Big difference at around -15%, I suspect 2080ti would have even bigger though. 

core22 seems to be using pci-e more frequently or transfers more data? I think the added time for transfering data back and forth is what causes such slowdown. 

decided to test the difference in a mining application between linux and windows, same clocks, same temperature cards. difference is pretty impressive at same sol/s.