LTT Folding Team's Emergency Response to Covid-19

[Favebook]

9 hours ago, LordKitsune said:

I'm really wondering how my client is calculating me being able to do over 300k ppd when every WU I'm completing are worth less than 10k pts and taking 4+hours per WU after factoring in downtime.

 

Yes, I know it calculates based on having WUs constantly, but that is a bit of an oversight in software. (I've done computer programming, so I understand the potential difficulty of this) But I feel it may give the user a better experience if - for instance - the client logged (which it does have a log) the amount of time working, the amount of points for completing a WU, and the amount of time it takes to go from a completed WU to having a new one, then calculate a rough PPD that is actually close to the points you could actually be receiving. 

 

Granted, I do know that they normally don't have issues on this scale, so they likely never needed to re-evaluate how their program calculates points.

As of right now, my WU to power consumption looks horrible, 14 WUs completed over 4.5 days.

 

Anyways, little rant is over. It's not like this is a system breaking thing so I'm not too worried about it, I'm just tired of opening web client and seeing 1/10th of the points I'm expecting. (Honestly, at this point someone making 300k ppd would expect to be closing in on 1m points and I'm not even at 100k (EoC is saying 141k but not all of those are associated with my passkey).

QRB and selecting right WU to see PPD is your problem 99%. Tho, ignore those numbers for now. Give them 24-48h to fix their points distribution and then check EoC.

9 hours ago, ssgtob1 said:

What is this is? 

 

17:45:57:WARNING:WU00:FS00:FahCore returned: BAD_WORK_UNIT (114 = 0x72)
17:45:57:WU00:FS00:Sending unit results: id:00 state:SEND error:FAULTY project:14551 run:0 clone:1138 gen:0 core:0x22 unit:0x0000000180fccb025e725b383838b253

 

It also didnt count the points for this WU

This usually means too OC-ed machine, but there has been a hell of a lot of failed WUs on even stock GPU/CPUs, this is a known issue as far as I am aware.

Advanced client has a little bit more of those WUs and Beta is getting the most failed WUs as far as my experience goes (on 2x R9, 1070, 1080 Ti and 2080S)

9 hours ago, LordKitsune said:

Well, not just full. You can set it to medium or full and have both of them running but full basically says "Hey F@H, I'm not using any of this hardware so thrash it please."

 

No, I've only paused WUs and would resume them shortly after. I think the only 2 that I ditched were at the beginning when I broke the living dog turds out of the software lmao

Ahhh, didn't know there was an estimation issue in the first place, though the issue that hit you seems to be the opposite of what is happening to me. Currently my CPU is the only thing working and it's pulling in a whopping ~3k pts per day. 6hr work time (actually 8 or so but 2 have elapsed), base credit is 370 with estimated credit at 1000. Not even expecting to get 300 out of it though.

 

Actually, when you set it up at "full" you can use your computer freely, and you can even set your Folding Core Priority in Advanced Client Configuration.

 

Tho, you can learn more about power slider HERE:

Spoiler

The Power slider has three positions and two running options:

• Light
  • CPU is Folding at half speed, GPU Folding is off.
• Medium
  • CPU is Folding at three-quarter speed. GPU Folding is on. This is the default setting, and recommended for most users.
• Full
  • CPU is Folding at full speed. GPU Folding is on. This is the most productive setting. Will cause fans to run faster and may produce more heat.
• While I’m working
  • Folding is enabled to run at all times, based on the slider position.
• Only when idle
  • Folding starts only when the system has not been used for several minutes.

 

3 hours ago, GOTSpectrum said:

I've has 3 slots across two machines fail... Anyone else suffering high WU failure rate. These devices are not OC'd 

Yeah, known problem already. But it seems that it has been happening only on newer WUs. There is quite a few threads on Folding forum about those failed WUs.

Also, there is a new error (at least I haven't seen it before nor have a lot of other people):

"ERROR:Kinetic energy error of 1692.71, threshold of 10"

 

It does happen on Beta mostly, but it is quite interesting error.

 

1 hour ago, Ryan_Vickers said:

Ah I see, I was probably right then.  I have an R9 Fury which according to AMD has a stock clock of 1000 MHz, but my model shipped with a factory OC of 1050.  It runs flawlessly in other applications at 1150 MHz but this recent error happened at 1000 MHz so I'm not going to blame the card for it.  I guess I'll just wait for them to work this out.  No point clambering for WUs that will just fail anyway when there doesn't seem to be enough network bandwidth to serve everyone as it is.

I have two R9 too, running them above 1100 CC on folding usually causes bad WUs, but 1050-1070 CC is running really fine. But they are factory OC-ed at 1040CC. I'd blame a card if you are OC-ing it above factory OC, but if you are @ factory OC, then it's probably bad WU (nothing you can do about that).

[ShortRouter]

2 minutes ago, GOTSpectrum said:

No way, but it was at around 100 Tflops before, 

 

Dr Greg said they have had a 10x increase in the number of folders... 

 

So 500-1000 Tflops could be realistic when WUs get doled out properly 

Im pretty sure you mean Pflops Over 100 Pflops for sure I'd guess.

[GOTSpectrum]

2 minutes ago, ShortRouter said:

Im pretty sure you mean Pflops Over 100 Pflops for sure I'd guess.

Shush... it's 0348 here and I've been staring at a spreadsheet half the day!

[rkv_2401]

1 minute ago, ShortRouter said:

Im pretty sure you mean Pflops Over 100 Pflops for sure I'd guess.

Yep, he meant PFLOPS. It would be interesting to see just how much processing power is available, maybe they'll announce it eventually?

[ShortRouter]

04:49 here gna make breakfast soon

[GOTSpectrum]

2 minutes ago, ShortRouter said:

Im pretty sure you mean Pflops Over 100 Pflops for sure I'd guess.

Oct last year it was 90ish Pflops.... so 10x the number of active folders. Even if they are only 1/2 of the average folder due to WU issues and what not... the numbers are still big

[GOTSpectrum]

Just now, ShortRouter said:

04:49 here gna make breakfast soon

Have you slept yet?

 

[ShortRouter]

Just now, GOTSpectrum said:

Have you slept yet?

 

yeah, sleep schedules tend to mess up when you don't have to work in the office^^

[GOTSpectrum]

8 minutes ago, GOTSpectrum said:

No way, but it was at around 100 Pflops before, 

 

Dr Greg said they have had a 10x increase in the number of folders... 

 

So 500-1000 Pflops could be realistic when WUs get doled out properly 

@ShortRouter Nothing to see here.... Move along....

[GOTSpectrum]

Just now, ShortRouter said:

yeah, sleep schedules tend to mess up when you don't have to work in the office^^

I havent been to sleep yet...

[ShortRouter]

hahahaha

[Tika]

I feel like this giant surge of folding is as if everyone took 1 extra vitamin gummy today

[GOTSpectrum]

We are catching up, look at this!

 

 

Keep pushing guys... She's still got more in her! 

[Tika]

I got my brother's PC working on it now, I have never seen so many points per day... It's majestic..... (980,000 points per day compared to my laptop's puny 575,000 ppd)

[Tika]

Is it normal for task manager to say the GPU is only at 50 percent usage? It does it on both my laptop and desktop.

[GOTSpectrum]

1 minute ago, Tika said:

I got my brother's PC working on it now, I have never seen so many points per day... It's majestic..... (980,000 points per day compared to my laptop's puny 575,000 ppd)

The 2070 can pull 1.4-1.6M ppd. the 2080ti 3.5M PPD...

 

All together I can pull in about 4-4.5M PPD with my motley crew of GPUs

[GOTSpectrum]

Just now, Tika said:

Is it normal for task manager to say the GPU is only at 50 percent usage? It does it on both my laptop and desktop.

Yes, use GPUz instead

[LordKitsune]

4 minutes ago, GOTSpectrum said:

We are catching up, look at this!

 

 

Keep pushing guys... She's still got more in her! 

I'm thinking of buying a box fan and attaching it to the side of my case for maximum airflow. Any thoughts anybody?

[ShortRouter]

Not a contest or anything, but can we beat 300 PFlops that 2 of the most powerful supercomputers (Summit and Sierra) can push? Only a matter of WU availability imo!

[LordKitsune]

Wow, apparently F@H sent my CPU a big boi WU. 1.84 days, and that's after letting the thing run for an hour to figure out its time (started at over a week).

[Tika]

2 minutes ago, LordKitsune said:

I'm thinking of buying a box fan and attaching it to the side of my case for maximum airflow. Any thoughts anybody?

I just took off the side panel and set the GPU fans to full blast

[FakeGamerGuy]

I have to build/upgrade the rest of the machines at work, but one I was able to remote in to start crunching. LETS GO

 

[ShortRouter]

5 minutes ago, LordKitsune said:

I'm thinking of buying a box fan and attaching it to the side of my case for maximum airflow. Any thoughts anybody?

Don't bother for a single GPU system. Set your case fans to normal speed and bump your GPU fan speed a notch so the card doesn't thermothrottle. You can use MSI afterburner for it.

[LordKitsune]

Just now, Tika said:

I just took off the side panel and set the GPU fans to full blast

What's a side panel? 
CPU temps don't care, it has water on it. GPU though... she gets hot if I don't watch it.

[Tika]

5 minutes ago, GOTSpectrum said:

The 2070 can pull 1.4-1.6M ppd. the 2080ti 3.5M PPD...

 

All together I can pull in about 4-4.5M PPD with my motley crew of GPUs

After the pandemic I am going to sell my old PS4 slim gold and buy a 2070 FE