LTT Folding Team's Emergency Response to Covid-19

[handymanshandle]

looks like I didn't get any COVID-19 related WUs right at this moment... Wished they would've added in a dedicated COVID-19 option lol

[FakeGamerGuy]

Same, but since they're already having a hard time keeping up with the amount of systems ready to work, it's probably for the best so otherwise-idle time gets used for good.

[Spotty]

21 minutes ago, pinksnowbirdie said:

looks like I didn't get any COVID-19 related WUs right at this moment... Wished they would've added in a dedicated COVID-19 option lol

Covid-19 work units are being given priority. If those aren't available you might get a work unit that contributes to a different research project, though that is still valuable. On the bright side you're at least getting work units. Lots of people stuck waiting to download unable to get anything at all.

 

The cause options are coded in to the program, no way for them to change those without releasing an update to the software.

[garydeppenbaugh]

1 hour ago, Ahaber said:

Why are you paused?

Probably cause it's doing a lot of the work when I'm idle.

[pducharme]

Running this F@H since I saw the callout from you guys.  I have a Ryzen 9 3900x, a Ryzen 7 2700x, a GTX 1060 6GB and a GTX 1080 ti all working on this

 

I have already 2.6 Millions points, ranked at 

49,231 of 2,155,628

 

 

[Spotty]

Another update from XOC.
Looks to be working on fixing some bugs (why the altest updates haven't worked) and possibly adding some new tracking statistics.

 

Quote

Before I get a ton of emails, I double checked the data files and the 6pm update's contents were the same as previous. A file freeze usually means they are doing something serious on the backend, hopefully it will be resolved soon.

I can't believe I haven't mentioned it yet, but there is an official F@H support forum that is probably going to be the most up-to-date on current issues, and where you can post your own problems if you need one of the F@H staff to see it.

I saw an article where CoreWeave recently donated 6,000 GPUs to folding right now!

I've received a couple emails from Operation Phoenix and they are getting ready to flip the switch on thousands of GPUs too!

I know the F@H team is working hard to get more infrastructure online to cope with all the demand, their facebook page says they are getting help from GitHub & Microsoft. I believe they have been getting some new servers setup on the Azure cloud.

You can also follow Dr. Bowman on Twitter if you aren't already. He's the current director of the F@H project in case you didn't know.

I spent most of the day re-coding some of the processing scripts, adding extra error-checking & re-downloading. Funny enough, it worked on the dev server perfectly last update I think it was and I never even realized it till I got the email for the EOC server that the file was corrupt. I've had to manually re-download the file and run the update for the individual stats a couple times today, so hopefully this will resolve that issue once and for all.

I also added in some more data validation, as I tried bumping up processing to the top 12,000 teams, but ran into some weird quirks. Come to find out there was one team that had a pretty nasty username. So I was going to let that run overnight to make sure everything was okay before pushing it out to the EOC server.

A big shout out to Raphael who gave the suggestion of adding processing for individuals based on their team's current production rate (instead of their team rank). I'm planning on implementing this tomorrow, and I think it will add individuals for around 50-100 teams that are currently very active.

Finally, if you've made it this far... Yes, I will try to get some aggregate statistics up SOON. People keep emailing me about it, and as soon I fix the big ticket bugs I'm going to get to work on it.

Bonus Throwback Thursday: If you remember when the Default team and Google were ranked separately because they had soooo many points and soooo much more production than all the other teams, you're probably a die-hard folder! LOL

https://folding.extremeoverclocking.com/

[Datsat7]

i don't think i can upload this project now because it has no ip

[Bitter]

I think I need to remount or repaste the cooler I just installed on the CPU so GPU folding only, I think I got it stable or it was just some bad WU's I was getting.

[Spotty]

6 minutes ago, Datsat7 said:

i don't think i can upload this project now because it has no ip

That's normal. It won't connect to the collection server until the work unit is complete.

 

Mine is saying the same.

[Datsat7]

Just now, Spotty said:

That's normal. It won't connect to the collection server until the work unit is complete.

 

Mine is saying the same.

thats because we are doing the same project 14533. i dont think our collection server is online

[vile1]

Hey guys, I have a WU that has been sitting for 3 days and not sending. Do I just let it time out on the 25th?

 

[vf1000ride]

It's back up and running again. It must have picked up a work unit right after my last post, never fails I have that kind of luck.  My stuck WU is still stuck but at least I am back to producing and I did see it finish and upload one.  Hopefully by morning I will have crossed the 500K threshold.  WooHoo

[Spotty]

36 minutes ago, Datsat7 said:

thats because we are doing the same project 14533. i dont think our collection server is online

Mine just finished. Uploaded fine.

 

04:20:57:WU02:FS01:0x22:Completed 980000 out of 1000000 steps (98%)
04:22:16:WU02:FS01:0x22:Completed 990000 out of 1000000 steps (99%)
04:23:36:WU02:FS01:0x22:Completed 1000000 out of 1000000 steps (100%)

04:23:41:WU02:FS01:0x22:Saving result file ..\logfile_01.txt
04:23:41:WU02:FS01:0x22:Saving result file checkpointState.xml
04:23:41:WU02:FS01:0x22:Saving result file checkpt.crc
04:23:41:WU02:FS01:0x22:Saving result file positions.xtc
04:23:41:WU02:FS01:0x22:Saving result file science.log
04:23:41:WU02:FS01:0x22:Folding@home Core Shutdown: FINISHED_UNIT
04:23:42:WU02:FS01:FahCore returned: FINISHED_UNIT (100 = 0x64)
04:23:42:WU02:FS01:Sending unit results: id:02 state:SEND error:NO_ERROR project:14533 run:0 clone:11240 gen:1 core:0x22 unit:0x0000000180fccb025e72f2b6ee02290d
04:23:42:WU02:FS01:Uploading 49.92MiB to 128.252.203.2
04:24:58:WU02:FS01:Upload 0.75%
04:25:04:WU02:FS01:Upload 11.77%
04:25:10:WU02:FS01:Upload 29.80%
04:25:16:WU02:FS01:Upload 47.58%
04:25:22:WU02:FS01:Upload 65.48%
04:25:28:WU02:FS01:Upload 83.38%
04:25:36:WU02:FS01:Upload complete
04:25:36:WU02:FS01:Server responded WORK_ACK (400)
04:25:36:WU02:FS01:Final credit estimate, 122920.00 points
04:25:36:WU02:FS01:Cleaning up

Edit: Though the next work unit I got does have collection server IP listed.

 

It uploaded fine for me so I wouldn't worry about it but if you finish the WU and it gets stuck uploading please let me know.

Edited March 20, 2020 by Spotty

[enterprise-psi]

Linus posted an ig story about an upcoming folding video !!!

 

Edit using the 6 gamers 1 pc machine

[Syaoran]

Would changing my RX 580's GPU workload to compute in AMD settings make folding faster or help during folding?

[vile1]

10 minutes ago, Syaoran said:

Would changing my RX 580's GPU workload to compute in AMD settings make folding faster or help during folding?

I think we went over this before. 

That is more of a Bitcoin Mining setting to down-volt and possibly down-clock the GPU for in preparation of the card running 24/7 365 churning bitcoin.

Since you're trying to crunch #'s as FAST as possible for a cure, just leave it alone and run that bitch as fast as she will go, as stable as you can.

I actually recently found that my 4.6GHz OC on my 6700K was not as stable as I had thought and had to bump it down to 4.5GHz

 

Pro-tip: if you have a multicore CPU with 8 threads or more, split the CPU into two, 4-core CPU slots (if you have an 8 core proc), or three 4-core slots if you have a 12core proc, etc. The WU's seem to be optimized for 4 cores.

[Running Raptors]

1 hour ago, vile1 said:

I think we went over this before. 

That is more of a Bitcoin Mining setting to down-volt and possibly down-clock the GPU for in preparation of the card running 24/7 365 churning bitcoin.

Since you're trying to crunch #'s as FAST as possible for a cure, just leave it alone and run that bitch as fast as she will go, as stable as you can.

I actually recently found that my 4.6GHz OC on my 6700K was not as stable as I had thought and had to bump it down to 4.5GHz

 

Pro-tip: if you have a multicore CPU with 8 threads or more, split the CPU into two, 4-core CPU slots (if you have an 8 core proc), or three 4-core slots if you have a 12core proc, etc. The WU's seem to be optimized for 4 cores.

Will this affect the estimated time drastically if you just have all cores working on one WU?

[Running Raptors]

I have a gt 625 that hasn't gotten a single WU since I have had the machine on and fiolding for about 4 days straight on the cpu. Is there like a logarithm that exists where it won't give WU to gpu's that are not common or good for folding?

[vile1]

13 minutes ago, Running Raptors said:

Will this affect the estimated time drastically if you just have all cores working on one WU?

The way I look at it: if they designed the Units to work best on 4 cores, then I'll run 4 cores. Better time is being made processing information, then waiting for a WU that likes my specific configuration. These guys have been doing this for years, and Im sure they optimize everything to work on computers most folding users are using. Mind you, this is ALL speculation on my part. You do whatever you want, ALL folding is BEST folding!

 

13 minutes ago, Running Raptors said:

I have a gt 625 that hasn't gotten a single WU since I have had the machine on and fiolding for about 4 days straight on the cpu. Is there like a logarithm that exists where it won't give WU to gpu's that are not common or good for folding?

I think also the WU's do look for hardware that will be most useful. Again, Im speculating.

[Running Raptors]

2 minutes ago, vile1 said:

The way I look at it: if they designed the Units to work best on 4 cores, then I'll run 4 cores. Better time is being made processing information, then waiting for a WU that likes my specific configuration. These guys have been doing this for years, and Im sure they optimize everything to work on computers most folding users are using. Mind you, this is ALL speculation on my part. You do whatever you want, ALL folding is BEST folding!

 

Well if it's true or not it's fine cause now i split my 2700x into 4 groups and now all of them have WU so can't change it now lol but yes you are right all folding is best folding

[Naamari]

i've got nice upload queue goin on.

[Spotty]

15 minutes ago, Running Raptors said:

I have a gt 625 that hasn't gotten a single WU since I have had the machine on and fiolding for about 4 days straight on the cpu. Is there like a logarithm that exists where it won't give WU to gpu's that are not common or good for folding?

With that graphics card it's unlikely you would be able to finish a GPU work unit before it timed out.

 

The GT625 is not listed in this GPU PPD Chart, but the GT 630 which should be roughly twice as fast as the GT 625 is listed at an estimated 2,500 points per day - the lowest of any of the cards listed.

It would not be worth using for folding. Too old, too slow.

 

1 hour ago, vile1 said:

Pro-tip: if you have a multicore CPU with 8 threads or more, split the CPU into two, 4-core CPU slots (if you have an 8 core proc), or three 4-core slots if you have a 12core proc, etc. The WU's seem to be optimized for 4 cores.

 

8 minutes ago, vile1 said:

The way I look at it: if they designed the Units to work best on 4 cores, then I'll run 4 cores. Better time is being made processing information, then waiting for a WU that likes my specific configuration. These guys have been doing this for years, and Im sure they optimize everything to work on computers most folding users are using. Mind you, this is ALL speculation on my part. You do whatever you want, ALL folding is BEST folding!

Due to the way the points are awarded and the quick return bonus; the faster you complete a work unit the more points you receive.

In theory it should be best to have all of the cores in one slot working on one project at a time so it can complete it in the shortest possible time. I'm not aware of any optimisation for 4 core CPUs or any reason why limiting slots to 8 threads would improve performance.

[Muhlis]

I have two computers, and I run both of them with the same username and passkey. Am I doing good? Should I do something else to ensure they making points?

[vile1]

Just now, Muhlis Gursoy said:

I have two computers, and I run both of them with the same username and passkey. Am I doing good? Should I do something else to ensure they making points?

Nope! You good!

[Cable_Player]

10 hours ago, Alex Atkin UK said:

I lost one because it was done in 2 hours but never submitted and its literally been days (well past the timeout now).

 

9 hours ago, vf1000ride said:

Frustration level rising.  Managed to turn in 366k points in the last 4 days.  Today, can't get a single WU.  14hrs now with no luck, tried restarts to no avail.  Got one WU that refuses to send and is on the problem child list over at the F@H forums and both gpu/cpu idle since 3am.  Bugger.

 

9 hours ago, Valentyn said:

I've one of those units stuck trying to send for a day now, the longer it takes, the less points it's worth, and it keeps ticking down.

 

6 hours ago, kelvinhall05 said:

Hi guys, I noticed this evening that my GPU slot has gone from constantly trying to download a new WU, to being "failed". Any ideas what it is and how I can fix it?

 

51 minutes ago, Naamari said:

i've got nice upload queue goin on.

Hi all,

 This is frustrating and has happened to me a few times. Each time it has, pausing all clients, restarting my PC and unpausing the folding worked and they uploaded soon after.