LTT Official Folding Month V (2022)

[Alex Atkin UK]

4 minutes ago, Lightwreather JfromN said:

One can always just build their own VPN server. Just a single script to set up and then install a client.

This is a script that I use quite often: http://github.com/Nyr/wireguard-install

iirc, there's also a setting to change what gets logged (I've forgotten how), so that's there if you don't trust the provider.

Plus when it comes to region unblocking, using a VPS gives you better odds of not being blacklisted compared to a dedicated VPN.

[Gorgon]

2 hours ago, GuardianAngel_ said:

Update:

 

Actually, I've now got a second card doing the exact same thing  - another 2080 ti.

These types of errors can sometimes be caused by a bad batch of WUs but I've folded many hundreds of 18601 Alzheimer's WUs and have only had 4 WUs fail on one specific GPU so I don't think this is the case.

 

You have 2 different errors:

An exception occurred at step 29174: Error invoking kernel: CUDA_ERROR_LAUNCH_TIMEOUT (702)

This is the client not coming back in time. i.e. not seeing a frame completed (1/100th of a WU) in the expected time period.

2 hours ago, GuardianAngel_ said:

04:04:15:WU06:FS07:0x22:An exception occurred at step 29174: Error invoking kernel: CUDA_ERROR_LAUNCH_TIMEOUT (702)
04:04:15:WU06:FS07:0x22:ERROR:98: Attempting to restart from last good checkpoint by restarting core.
04:04:15:WU06:FS07:0x22:Folding@home Core Shutdown: CORE_RESTART
04:04:19:WARNING:WU06:FS07:FahCore returned an unknown error code which probably indicates that it crashed
04:04:19:WARNING:WU06:FS07:FahCore returned: UNKNOWN_ENUM (-1073740791 = 0xc0000409)

and this is the core crashing.

 

At this point I'd suggest.

1. Delete the Folding@Home Client and it's Data and Program directories (corrupt files?)

2. DDU and re-install your NVidia drivers. (Bad Driver)

3. Reinstall F@H Client

[GOTSpectrum]

Go team Go! 

[Alex Atkin UK]

Nice, Corsair finally got their Type-4 12VHPWR cables in stock!

[podkall]

51 minutes ago, GOTSpectrum said:

Go team Go! 

That's all you?

[BondiBlue]

9 minutes ago, podkall said:

That's all you?

No, that's the LTT team (click for full resolution):

 

[podkall]

2 minutes ago, BondiBlue said:

No, that's the LTT team (click for full resolution):

 

go team go

[GOTSpectrum]

19 minutes ago, podkall said:

That's all you?

If I were doing 2 bil a day I think I would be starting my own team lmfao 

[BondiBlue]

I'm not 100% certain, but I have a strong feeling that the issues with my Dell Precision T3500 are hardware related, possibly due to thermal expansion. The machine crashed again this morning, and just like the other times it's crashed it happened shortly after I paused F@h. I'm only folding on the GPUs, but there are four GPUs installed so the CPU is kept somewhat busy. At this point I don't really know what else to do. I've been very busy lately, and I don't really have time to move all of these GPUs over to another machine. I have a second Precision T3500, but that one has issues with the memory slots and will crash at random (really at random, not from any sort of thermal cycling or anything), so I don't want to use it right now. 

 

Based on my findings I might classify this Precision as a piece of hardware that died during folding month. It never had these issues before. I can try to reseat the CPU, but I'm not sure how effective it'll be. 

 

Maybe I'll use this as justification for upgrading to a newer Precision workstation. Dual CPU models like the T7610 and 7910 aren't too expensive on eBay...

 

Edit: just to add why I'm so sure it's a hardware issue - I finally managed to catch the machine showing a blue screen this morning. The error was a machine-check exception, and the computer was locked up while showing the BS. It's fully powered down right now, and when I get home later I'll fire it back up and just leave at least one GPU folding 24/7 until the end of the event. Hopefully that'll help it limp along to the end. 

[RollinLower]

2 hours ago, Gorgon said:

These types of errors can sometimes be caused by a bad batch of WUs but I've folded many hundreds of 18601 Alzheimer's WUs and have only had 4 WUs fail on one specific GPU so I don't think this is the case.

 

You have 2 different errors:

An exception occurred at step 29174: Error invoking kernel: CUDA_ERROR_LAUNCH_TIMEOUT (702)

This is the client not coming back in time. i.e. not seeing a frame completed (1/100th of a WU) in the expected time period.

and this is the core crashing.

 

At this point I'd suggest.

1. Delete the Folding@Home Client and it's Data and Program directories (corrupt files?)

2. DDU and re-install your NVidia drivers. (Bad Driver)

3. Reinstall F@H Client

might also be worth it to roll back a driver version depending on if the issues persist.

[RollinLower]

1 hour ago, GOTSpectrum said:

If I were doing 2 bil a day I think I would be starting my own team lmfao 

or a cloud computing business or something. 

[RollinLower]

gotta love the delay in the EOC PPD graphs by the way. my server was down last night, and just now it shows the dip. Hopefully tomorrow shows a brighter story.

 

i now just realised this is a triple post by me, i'm gonna stop the spam.

[Senzelian]

12 minutes ago, RollinLower said:

i now just realised this is a triple post by me, i'm gonna stop the spam.

rEpOrTeD!!!1!1!11!1!!

 

But you know you can edit your posts right?

[RollinLower]

4 minutes ago, Senzelian said:

rEpOrTeD!!!1!1!11!1!!

 

But you know you can edit your posts right?

yeah i know, but i can't merge them

[GuardianAngel_]

4 hours ago, Gorgon said:

These types of errors can sometimes be caused by a bad batch of WUs but I've folded many hundreds of 18601 Alzheimer's WUs and have only had 4 WUs fail on one specific GPU so I don't think this is the case.

 

You have 2 different errors:

An exception occurred at step 29174: Error invoking kernel: CUDA_ERROR_LAUNCH_TIMEOUT (702)

This is the client not coming back in time. i.e. not seeing a frame completed (1/100th of a WU) in the expected time period.

and this is the core crashing.

 

At this point I'd suggest.

1. Delete the Folding@Home Client and it's Data and Program directories (corrupt files?)

2. DDU and re-install your NVidia drivers. (Bad Driver)

3. Reinstall F@H Client

Cheers @Gorgon, I'll post an update soon.

[NetoriusNick]

10 hours ago, RollinLower said:

Reinforcements have arrived!

 

I'll deal with the cable mess later, for now we resume the fold!

This puts a drive on me to finish my projects and water cool my cabinet ( always on the never never even with 50 % of the parts already)

 

[Bianks84]

Climbing team ranks y'all... Just love seeing big numbers lol

[BiotechBen]

Just found a local guy who ran 1080 SLI and wants to dump the cards for $100 each, is willing to let me test them.

[cbigfoot]

Just now, BiotechBen said:

Just found a local guy who ran 1080 SLI and wants to dump the cards for $100 each, is willing to let me test them.

test without the bridge lol 

[BiotechBen]

1 minute ago, cbigfoot said:

test without the bridge lol 

If the 1080 DOES work and IS stable, I think I'm only getting one. As cool as 2 would be, I have no use for two and I don't think my PSU would be happy, and would require me to get more cables since I have misplaced the originals.

[cbigfoot]

54 minutes ago, BiotechBen said:

If the 1080 DOES work and IS stable, I think I'm only getting one. As cool as 2 would be, I have no use for two and I don't think my PSU would be happy, and would require me to get more cables since I have misplaced the originals.

what cards you have currently? for $ 100 ea if you have an older card i would try and sell the older card...

 

[BiotechBen]

12 minutes ago, cbigfoot said:

what cards you have currently? for $ 100 ea if you have an older card i would try and sell the older card...

 

I've got an rx480 8gb rn, if the 1080 is not defective, the 480 would become a dedicated folding card room heater, or it would go into the collection of parts that would become a gift computer as I upgrade.

[GOTSpectrum]

So it gets to 00:15 and I realise I haven't eaten, beans on toast to the rescue, blog post should be up around the usual time, maybe a slight delay while I eat and get myself sorted. Not been home very long from supporting a friend going through a breakup.

 

Remember due to my long stream attempt I will not be available for the usual blog post, instead the stats will most likely be added sometime the day after. Possibly even skipped but don't worry the stats will be collected and verified as usual by @cbigfoot

[cbigfoot]

OK Points look good...just a final Reminder TODAY is the last day for you to start folding and still hit 20 days...and you must fold for Team LTT (223518) folding for any other team does not count towards the count.

[GOTSpectrum]

Tfw you don't disable automatic updates and you get forced restarted on... Right as I'm writing a blog