LTT Official Folding Month 2019!!!

[SwedBertha]

Some people on Ubuntu are currently experiencing issues after "sudo upgrade". This is due to Nvidia drivers being changed and there the machine needs to be rebooted for F@H to detect proper GPU-lettering. I had this yesterday for a brief moment - Rest in pieces my 68 days uptime

[Gerowen]

On 9/18/2019 at 8:33 PM, Corrupt_Liberty said:

NOOOoooo!!!

 

That's gonna knock a couple k PPD off my folding for the day.

Because it might just take the entire day and 14 restarts for them to finish,

[mattheginger]

6 hours ago, SwedBertha said:

Some people on Ubuntu are currently experiencing issues after "sudo upgrade". This is due to Nvidia drivers being changed and there the machine needs to be rebooted for F@H to detect proper GPU-lettering. I had this yesterday for a brief moment - Rest in pieces my 68 days uptime

I'm on windows 10, and I've had multiple occasions in the last few days where my client stops folding and discards the WU, then sits there doing nothing (until a restart).

 

Always a 14250 WU, but not sure if that has anything to do with it.

[Rusted]

8 hours ago, SwedBertha said:

Some people on Ubuntu are currently experiencing issues after "sudo upgrade". This is due to Nvidia drivers being changed and there the machine needs to be rebooted for F@H to detect proper GPU-lettering. I had this yesterday for a brief moment - Rest in pieces my 68 days uptime

There is nothing odd in that, you need to reboot Linux when you update drivers and/or kernel. The distro I use tells you if you need to reboot for the updates to work. There is a notification in console after the update.

[Gorgon]

11 hours ago, shaz2sxy said:

It'll be moving it back down after the event when I remove the 1070ti

 

Anything else I can check to see how I can get more out of it? 

The question is which card is under performing? The 2070 Super, which should be averaging 1.5MPPD or the 1070 Ti, which should be averaging 850kPPD?

 

If it’s just one card then you need to look at that card’s Power Limit, overclock(s) and temperatures.

 

Also check the log files to see if your failing WUs or erroring due to an overclock and rolling back to the last checkpoint. If the second scenario is occurring less than three times then you will still complete and get credit for the WU. On the third error the WU will fail and be aborted.

 

It could also be a slow download bug where it’s taking more than 1/2 hour to download a WU.  I’ve been seeing those over the last few weeks.

[shaz2sxy]

39 minutes ago, Gorgon said:

The question is which card is under performing? The 2070 Super, which should be averaging 1.5MPPD or the 1070 Ti, which should be averaging 850kPPD?

 

If it’s just one card then you need to look at that card’s Power Limit, overclock(s) and temperatures.

 

Also check the log files to see if your failing WUs or erroring due to an overclock and rolling back to the last checkpoint. If the second scenario is occurring less than three times then you will still complete and get credit for the WU. On the third error the WU will fail and be aborted.

 

It could also be a slow download bug where it’s taking more than 1/2 hour to download a WU.  I’ve been seeing those over the last few weeks.

Thanks for the help. will try some things out over the weekend

 

doesn't help when my internet goes in the middle of the night so no new WU done

[SwedBertha]

3 hours ago, Rusted said:

There is nothing odd in that, you need to reboot Linux when you update drivers and/or kernel. The distro I use tells you if you need to reboot for the updates to work. There is a notification in console after the update.

Ah yeah I know. But there was little information when searching the error-codes so I thought I pop it in here as informational.

Never had issues on windows though.

[Stormseeker9]

00:10:49:WARNING:WU01:FS02:Exception: Failed to send results to work server: Transfer failed
******************************* Date: 2019-09-20 *******************************

 

These are server related issues right - ea the program side of things?

[Rusted]

2 hours ago, SwedBertha said:

Ah yeah I know. But there was little information when searching the error-codes so I thought I pop it in here as informational.

Never had issues on windows though.

On Windows driver installer usually reboots your computer after/or during the installation.

[GrockleTD]

So figured out why I suddenly stopped producing yesterday, internet seems to have been physically disconnected from the main lines of Comcast. Notedly, now that I’m at school and have internet, I did notice an email from them stating they were upgrading my speeds, (to an ever more obnoxious 100/5) so it might be related, but I’m not sure

[GrockleTD]

It’s just strange that they wouldn’t give advance notice if they were going to physically cut the cable, so that needs to be sorted out, (as in replacing the line between my house and the street) tbh, woulda been kinda nice If they had buried the line, instead of running it above ground

[RaiderZulu]

Ignore my request edit access on the form.  I was wanting to know if I had signed up successfully as a few others had some problems with it.

[Corrupt_Liberty]

If you guys get bored and want to game while your rigs are folding, go RETRO!

 

 

[Gorgon]

Finally installing 1903 on my Daily Driver. Figured I better do it now before it forces it during the Folding month. Had to use the media creation tool. Doubtlessly from hacking it to attempt to stop automatic updates in the past.

 

Once that’s done I’ll be pulling the Radeon Pro W4100 out and installing my 1060 6GB Short board and that will be the end of the GPU rearranging in preparation.

 

The new gaming system hung overnight running stress tests. I noticed that the CPU (2600X) was hitting 94C with mprime running on all cores and a synthetic load on the GPU so I dialled it back to 8 cores.  This was even after I replaced the front stock Fan with 2 Noctua 140mm 1500rpm redux and the rear with a 120mm Noctua Redux 1700rpm. I suspect It’s either a combination of too much voltage or a bad contact between the CPU and heatsink.

[Mattias Edeslatt]

How can one verify that one is signed up and all details are correct for the event?

[GOTSpectrum]

On 9/18/2019 at 4:20 PM, GOTSpectrum said:

Been tuning my clocks, dialling them right in ready for the event, 

 

1080ti - 2075mhz, 10188‬mhz(effective)

 

1080 - 2100mhz, 9486(effective)

 

Bot bad, leaving the high density system with max fans and a 80% power target as they boost higher this way.

Si just an update, these clocks are fully stable for folding... Quite surprised with the 1080 ti tbh. 

[GOTSpectrum]

2 hours ago, Mattias Edeslatt said:

How can one verify that one is signed up and all details are correct for the event?

Just message me your folding ID, will probably get back to you tomorrow evening as I have work in 8 hours.

[GOTSpectrum]

Couldn't sleep so did a couple superchats and got us a WAN shout out!!!!

[Gorgon]

2 hours ago, GOTSpectrum said:

Si just an update, these clocks are fully stable for folding... Quite surprised with the 1080 ti tbh. 

That's impressive - My 2070 maxxes out at 2055. It can go to 2070 but will fail the odd WU.

[GOTSpectrum]

UFD TECH also gave us a shout out.

[GOTSpectrum]

6 minutes ago, Gorgon said:

That's impressive - My 2070 maxxes out at 2055. It can go to 2070 but will fail the odd WU.

It failed one WU, but after upping fan speed and voltage, dropped mem clock by a hundred megahertz it seems to have gone fully stable, but only time will tell

[sanya567xxx]

2 hours ago, GOTSpectrum said:

Si just an update, these clocks are fully stable for folding... Quite surprised with the 1080 ti tbh. 

am surprised by that too. Also I noticed that it's not the power target but the temperature target being lowered is what helps it boost higher. So you can move temp target to like 84 or even 75C (that's what RoG Tweak II sets it when power and temp targets are linked), but then you can actually increase power target by quite a bit, leaving temp target low, and the gpu will still boost as high.

for several gpus you can also just flash the XOC bios which removes the PL/TL completely.. But my strix 1080ti (not OC model) failed to go through stress testing while using that bios for some reason. Was using Kombustor w/ asus mods for it, app/driver just crashed at stable clocks and temps at like 70 degrees.

Also, what's your expected PPD for 1 1080ti?

I wonder if playing like this (boost +167 - effective 2025), voltage +30, mem +100 (11110), fans maxed out, PL 120%, temp 76 is possible without the need to toggle profiles. I know xonotic won't care much, gpu is barely used in that game, but overwatch? with new drivers heating the system up while on character selection screen and just chilling in fights - idk if it won't care about the temp going close to the target.

[GOTSpectrum]

2 minutes ago, sanya567xxx said:

am surprised by that too. Also I noticed that it's not the power target but the temperature target being lowered is what helps it boost higher. So you can move temp target to like 84 or even 75C (that's what RoG Tweak II sets it when power and temp targets are linked), but then you can actually increase power target by quite a bit, leaving temp target low, and the gpu will still boost as high.

I think I'm just going to leave it as is, seems to be working out fine. My temps are around 50-55 when folding on both cards.

[sanya567xxx]

6 minutes ago, GOTSpectrum said:

I think I'm just going to leave it as is, seems to be working out fine. My temps are around 50-55 when folding on both cards.

50-55? are they in a custom loop? mine stays at 65 with fans blasting 100% and an added external fan on the side of the case.

[Gorgon]

6 hours ago, Gorgon said:

The new gaming system hung overnight running stress tests. I noticed that the CPU (2600X) was hitting 94C with mprime running on all cores and a synthetic load on the GPU so I dialled it back to 8 cores.  This was even after I replaced the front stock Fan with 2 Noctua 140mm 1500rpm redux and the rear with a 120mm Noctua Redux 1700rpm. I suspect It’s either a combination of too much voltage or a bad contact between the CPU and heatsink.

Looks like it was a combination of a bad paste job and the thermal sensors in Linux still having the +10C offset from the early Ryzen days. Booted into Windows and ran HWinfo64 and Prime95 and saw the idle temps of 46C and low 90s under the small FFTs.

 

Replaced the cooler with one from a 2700 though it's the same version with RGB, which will make it go faster  Applied some new paste and it was maxxing out at 88C now.