Folding Forum

Jorgeminator wrote:KTH Royal Institute of Technology seems to have fired up one, too. Great seeing some engagement in the Nordic countries as well

Saw that too! With SweClockers being a populars Swedish tech site, I can only imagine that bandwidth and latency would be lower for European contributors if that came online.

This got me thinking about the FAH network setup in general:

• Why isn't the "Collection Server network" redundant? (today always a single ip for CS delivery). And where a local server could be preferred? (for CS:s not cloud based). Especially because WU:s have an expiration timer, and a point system is in place... all this non-uploaded work need a confirmation it has been done, or those WU:s will be sent out again, right?
• Why can't users of the Client not only share their GPU and CPU, but also their hard drive and network card?

Sitting here on 700mbit upload capacity, and some 3+ TB disc...
...and where all this idling waiting for WU:s could be used to cater say a "Peer-2-peer network", where CS:s are supernodes..
... and having WU:s that been waiting for upload for more than 2days

The idea of a P2P storage system for FAH has already been tested ... I don't know why it has never been chosen, but here are some readings on the subject : http://en.fah-addict.net/articles/6-storage-home/

toTOW wrote:The idea of a P2P storage system for FAH has already been tested ... I don't know why it has never been chosen, but here are some readings on the subject : http://en.fah-addict.net/articles/6-storage-home/

Thanks for that info.
Although the detailed pdf about the "Storag@home" is now gone with the removal of "Adam Beberg":s blog.


But maybe that is not a big loss. Of what I managed to read on the above link, they started from scratch, and that is a big hurdle if you ask me.
Since then a lot of different storage networks have been created, and some of them are professional networks used by for example Google, LinkedIn and Facebook.

Don't know the performance on all the newcomers since 2006, but personally I like the specs on GNUnet.
FAH usercase in that network could be file sharing using "F2F topology with anonymity level set to zero" (non-anonymous file sharing). It would not be able to give "points" in FAH-network, but the GNUnet network has a "currency system" that translates to trust. So the more you give the network, the more trust you gain. And in a way that fits the user profile for most here, that are hunting "points".

Also: GNUnet has its own DNS-system, meaning it is pretty resilient if the internet for some reason would get in trouble, or start setting up borders.


Maybe I should start a completely new thread about this.
Not impossible there are finished projects that have a compatible point system... and still a high availability and performance curve that increases with user load (like torrent). Or maybe something like a "torrent-lib extension to the code" here could be a non-complicated solution, where a central tracker (that can give points) always "sees" status on availability, and warns if the amount of distributed copies are getting to low.

To answer the title of the thread: No, it's not really getting better. I just came back to find my laptop has been idle for over 8 hours, unable to get a WU

So it's still annoying that I have to remember to check once in a while. It's not yet a 'set-and-forget' thing

iceman1992 wrote:...It's not yet a 'set-and-forget' thing

One could say that it is indeed a set-and-forget setup but not yet a set-and-fold setup

Ever since I enabled advanced projects I've had a near-constant stream of WUs.

Azmodes wrote:Ever since I enabled advanced projects I've had a near-constant stream of WUs.

While that might work for you, please note that using that flag, there's a possibility of problematic WUs as this is the last stage of the project before being moved to full F@H. Thus, it would be recommended to keep an eye out in case there's any issue with the WU.

Thanks, I'm aware. Since I opted in last Saturday I did have one out of my seven FAH crunchers crash once, so maybe that's related. Still, this seems vastly preferable to being without work for hours on end at worst and never getting all slots filled at best, as it was before. I do monitor all my machines too.

I am still doing reasonably well on four GPUs and one CPU. The CPU always has work, while maybe one GPU will be without work for a half-hour or so, but I can live with it.

By the way, Welcome Home PantherX! I see you have responded to the call of duty. Your expert advice is always needed.

This was the first weekend that I almost constantly getting new work. I'm doing only gpu work by the way.

I've been getting pretty consistent CPU work for the past week, running a MacBook Pro (15-inch, 2017).

After a week of issues today looking more like normal only small gaps now with no work and the backlog of completed workunits finally sent.

I've not noticed any improvement here. Still mostly "WARNING:WU01:FS00:Failed to get assignment" unless I give it a good kicking.

@pachydermus +1

I can't say I have felt any rise either these last days, rather a down trend. But today is looking good, I have had a few long stretchs at 100% capacity on my two machines.
On the servers' side, the assign rate is steady at around 140.000/hour, which is as high as it ever was. A couple new servers started working, too.