Folding Forum

Andre_Ti wrote: ↑Thu Sep 07, 2023 4:28 am It is expected to see your negative reactions at the slightest delay in issuing work units and creating conspiracy theories of researchers)

Note exactly sure what this means, but without folks here bringing up the concern the problem would NOT have been fixed.

Also as I have stated, IN MY OPINION, that the project has become poorly managed and lacks much feedback or transparency for something that is as important as it is. Not at all sure if that was what you are referring to, but I still stand by it, and I don't think their are any conspirancy theories here just observations on what has been going on recently with issuers (this is not the only one).

Not looking to ruffle any feathers, but rather to get some actions for issues that will go on unnoticed unless someone speaks up.

Sandy

SandyG wrote: ↑Thu Sep 07, 2023 5:24 am Not looking to ruffle any feathers, but rather to get some actions for issues that will go on unnoticed unless someone speaks up.
Sandy

It's just a joke.
I understand your concern, but in the absence of information, much can be assumed, of course it is better that we are told the reasons. But in fact, the operation of the servers is quickly corrected when they are completely turned off. The absence of a large number of working units, in my opinion, can be explained by a temporary lull after the summer period, especially since there is positive news about the end of testing of the new core23.
We are waiting for new, preferably large, working units.

Andre_Ti wrote: ↑Thu Sep 07, 2023 6:05 am

SandyG wrote: ↑Thu Sep 07, 2023 5:24 am Not looking to ruffle any feathers, but rather to get some actions for issues that will go on unnoticed unless someone speaks up.
Sandy

It's just a joke.
I understand your concern, but in the absence of information, much can be assumed, of course it is better that we are told the reasons. But in fact, the operation of the servers is quickly corrected when they are completely turned off. The absence of a large number of working units, in my opinion, can be explained by a temporary lull after the summer period, especially since there is positive news about the end of testing of the new core23.
We are waiting for new, preferably large, working units.

Ahh, good! Thank you for the clarification

Like to keep it fun here, and didn't catch the humor. Let's hope for better support and happier days

Andre, you clearly care a lot more than I do about watts, MHz, and the like. I just donate cards and power to this purpose is all. I do not worry about the details of how the cards are setup. For example, I tried using Linux to fold with to be able to donate more of the cards/power, but did not have time to optimize it so just went back to Windows as that was faster/easier.

Six of the seven folders I have do nothing but fold. Only my 4090 does anything else and that is my personal desktop, that plays the very occasional game when I find an hour to myself. I do at least let MSI Afterburner run a performance curve, but other than limiting to 80-degrees and using the curve, I do not pay attention to the results or try and tweak them beyond.

I have no “conspiracy theories” about what is going on, would just like to know. Joe and another answered, so great… hopefully the researchers are still interested in using the donated cards/power. If they have found new/better ways to do their research than F@H, then we can move onto something else for our cards/power.

Anyway, I am glad you are so into tweaking your voltages and monitoring your power consumed. It isn’t important to me, but if you will doubt numbers posted that I thought might be useful for folks, then I won’t post them was all I was saying.

Have a good day!

Let's close this topic out, I think we are back on track with WU's.

I feel like we could have a great discussion on a new Topic for many things here.

I do want to thank ALL that chimed in to help, had feedback and finally pushed through all of our chatter here.

Thanks everyone!!

Sandy

Sorry for the necrothread, but it's still happening, almost 30% of the time on Nvidia RTX 3000 and AMD RX 6000 cards. The GPUs will go in Waiting mode (i.e. They got assigned to a server, but no work unit is sent over x hours after receiving the "Could not get assignment" error) and it could stay there for days until I close F@H and open it again.

It's happening on multiple computers with varied hardware, the only constant being that these are under Windows (10 or 11, doesn't matter). It might be something with my ISP, but it was fine in the past.

If it is hanging for hours that is another problem, possibly involving your ISP or any firewalls involved. There has been a recurring problem that if any ACK packets are dropped, the client network connection may hang instead of timing out and retrying. Some ISPs set their routers to drop ACK packets for HTTP connections if they are busy under loaded conditions. In a browser you might see a page partially load and stop, then you just reload to get the full page. The F@h client attempts the same thing, but may hang if too much is dropped. Restarting the F@h client resets the network connection.

pyrocyborg wrote: ↑Fri Sep 15, 2023 4:19 pm Sorry for the necrothread, but it's still happening, almost 30% of the time on Nvidia RTX 3000 and AMD RX 6000 cards. The GPUs will go in Waiting mode (i.e. They got assigned to a server, but no work unit is sent over x hours after receiving the "Could not get assignment" error) and it could stay there for days until I close F@H and open it again.

It's happening on multiple computers with varied hardware, the only constant being that these are under Windows (10 or 11, doesn't matter). It might be something with my ISP, but it was fine in the past.

I have not seen many of these for a while, but did see one yesterday.

One thing I found was the back off on retry can get long. If you have the webUI up, stop processing, wait a bit for it to really stop, then restart. I think that causes the retry mechanism to start over. The down side is that you have to be watching it to know it stalled in the first place.

Still other issues with the system going on I fear...

Sandy

No issues here for a while now. Had a couple folders crash today. Think there might be something with older nVidia drivers and patch Tuesday. Upgraded nVidia (537 I think it is) to see if it stops.

SandyG wrote: ↑Sat Sep 16, 2023 12:18 am
I have not seen many of these for a while, but did see one yesterday.

One thing I found was the back off on retry can get long. If you have the webUI up, stop processing, wait a bit for it to really stop, then restart. I think that causes the retry mechanism to start over. The down side is that you have to be watching it to know it stalled in the first place.

Still other issues with the system going on I fear...

Sandy

Yeah, sometimes, hitting pause and letting it "wait" for a few hours then starting it again does work. Sometimes, it doesn't. Restarting F@H is usually a faster way, but it's not failproof as it's dependant on having avaiable WU by the time your PC tries looking for one.

As per Joe_H comment and seeing a suggestion there (or should a say, the possibility to test a theory), I moved two of my "problematic and easily movable" PC around and put them in another building with two different ISP. The exact same thing happens and they stop looking for a new WU after the servers tell there is no WU available.

It really happens once there are no WU available for a particular model range for a few minutes. If you happen to try to download a WU in that timeframe, and if no WU becomes available within the next 2-3 tries, it will hang there. Maybe this is something "Windows" related or all ISP do that around here, even those that worked just fine on that end over the past few years.

In the last few days, it's been way more common on my RX 6000 cards than on the Nvidia ones (all of my AMD rigs will stop at one point during the day). On the Nvidia side, I've seen it once or twice with a RTX 3090 and once with a RTX 3080.

I guess that the only reason I see that more often now is that there are less WU available for AMD RX 6000 cards. There's a lot of 0x22 jobs available at all times, but these must be for either Nvidia cards or other AMD models.

I'll just have to babysit the AMD rigs

Yep, no generated work units is the likely issue. Also I don't think pausing will cause an immediate retry, can't remember for sure, but if you STOP with with webUI it does, Fire up the FAH Control panel and watch the logs you can see what's what.

Sandy

Monitoring the total active users on EOC shows that about 2,000 more users became active over the last week. No idea on what equipment they are bringing to F@h, but that will have increased demand on the servers for WUs.

Joe_H wrote: ↑Sat Sep 16, 2023 8:00 pm Monitoring the total active users on EOC shows that about 2,000 more users became active over the last week. No idea on what equipment they are bringing to F@h, but that will have increased demand on the servers for WUs.

That is a good thing, but then I might ask how many left or stopped processing.

Also I can't imagine for new users that when the home screen for the client is broken that is an inviting experience with FAH. Going on a week now with no points totals. Will bump the other thread for this issue.

Sandy

Lazvon wrote: ↑Sat Sep 16, 2023 1:45 am No issues here for a while now. Had a couple folders crash today. Think there might be something with older nVidia drivers and patch Tuesday. Upgraded nVidia (537 I think it is) to see if it stops.

Still no “no work” issues for me, and looks like the nVidia driver update after the last Win10 updates fixed my crashes (after a day at least).

More WU have been added on September 15th, but you are folding them too fast !

I guess that more projects are finishing as the time goes by, but we didn't have a lot of new ones to replace them ...