Please REMOVE RX560

[Lamberto Vitali]

The system crashing is because the GPUs are old, and because I use it as a test bed for ones I have taken apart and resoldered etc. You have no clue about PCI Express speeds, Folding does not need more than 1 of PCI-E v2 lanes for a 280X. That is a blatant fact available by me looking at my machines here.

I'm not demanding changes to the client to work around an unstable system, I'm suggesting changes to bring it up to speed with Boinc which allows the user to choose what and when they do things.

Here is a simple example. A user has a machine they like to use on Folding, and sometimes they go away for the weekend. Let's say they see a task finishing in 2 hours, they set it to "finish" because they're going away for the weekend. But before they go, they reboot their machine. Oops! Folding downloads another task without asking the user. Blatant stupidity. That slot was not set to run! Even worse, when the user sees the task they cannot run because they won't be here and the machine will be off, they can't cancel it! So you r scientists are waiting a few days until it times out then waiting for the next person to do it!

[aetch]

I will repeat what has been said many times already

THIS IS NOT BOINC

If you love BOINC so much then devote all you resources to it.

[Lamberto Vitali]

It would appear you didn't listen to your own instruction. You said "I'm out".

It isn't Boinc, Boinc is rubbish, but at least it's better than this. So, being the helpful bloke I am, I made some suggestions so this could do some of the UTTERLY BASIC things Boinc does. But no, you don't listen.

[aetch]

Who's not listening?

FAH listens and reacts quickly when they have created issues or they have issues on the server back-end.

They have a dedicated but limited developer resource, which they pay for. They are not depending on free developers who "might" be able to do something in their free time.
The client does see changes and upgrades but just like any other piece of software it takes time to develop, test/validate and deploy them. They are developing for Windows, Linux and macOS and they need to ensure it is stable and all works together on the same platform and cross-platform.
The folding cores (the programs which process the actual work units on your computer) see the most regular development and updates.
In 2021 the major push was to enable CUDA on the NVidia cards, which had a very bumpy ride and took several attempts to get that folding core and its dependencies correct.
In 2020 a small thing happened called covid, that lead to an explosion of people wanting to donate their computing resources to FAH. Basically, we overloaded the server back-end and they spent quite a number of months upgrading their back-end infrastructure to recover from that.
This linus tech tips video -> https://www.youtube.com/watch?v=HaMjPs66cTs is both the reason I started folding and an example of how swamped FAH were at the time.
Those developers have their priorities set by the FAH science team, your singular unstable system is not one of them.

Your suggestions may, or may not, eventually find their way into a client at some point in the future but they absolutely will not find their way into the current release of the client. The client is not patched in any way shape or form. Its only upgrade path is to be replaced by a new release.

Back to you.
These forums are here to help plebs, like you and me, resolve issues with their individual folding systems and to highlight bigger issues with the FAH back-end.

Old does not automatically mean unstable, it means underpowered for today.
You can buy a brand new system that is unstable. It's unlikely but possible.

Your issues come from your unstable system, fix it.

[Lamberto Vitali]

They have a dedicated but limited developer resource, which they pay for. They are not depending on free developers who "might" be able to do something in their free time.

Then they should be able to improve the program more than Boinc can.

The client does see changes and upgrades but just like any other piece of software it takes time to develop, test/validate and deploy them.

I wasn't asking for it overnight.

Those developers have their priorities set by the FAH science team, your singular unstable system is not one of them.

No idea why you keep referring to this "unstable system" - I have 7 PCs and 10 GPUs and am constantly adding more and rearranging things and testing. I like repairing old kit. The improvements I suggested to the client are nothing to do with anything going wrong at my end, they're to do with basic options for the user to decide what to do with their system. As I said earlier, consider someone who sometimes runs folding and sometimes runs something else, or nothing, or goes away on holiday. The client can't even remember if he was folding or not after a reboot!

Old does not automatically mean unstable, it means underpowered for today.
You can buy a brand new system that is unstable. It's unlikely but possible.

My old systems work fine. The one you keep on about has lots of fast GPUs, and a CPU which is perfectly fine at feeding them. I don't use the CPU for computation. The MB has a lot of connectors on it, although I can multiplex a connection if needed.

[Lamberto Vitali]

Seems the scientists got fed up of Python:

[Joe_H]

Or they no longer needed to refer to the manual through experience, and it worked well to raise the height of the monitor.

[Lamberto Vitali]

My monitor has adjustable height. Are times so hard they had to get fixed ones?

[Joe_H]

Depends on the source, the price, and who is doing the purchasing. When I worked at the university I had to take what they bought. Sometimes they were height adjustable and other time not.

[Lamberto Vitali]

I was always the one who did the buying. I selected something sensible, then asked the user if they wanted anything different, then bent the rules to buy the more expensive one they wanted. A kettle is a water heater. A gaming card is a maths coprocessor.