ps. The RAC maybe smoking but the system doesn't show a single puff (of smoke :)

And for my NEXT experiment...... ;)

I am waiting for a new 1 to 4 expander card. Several different tests appear to show one of my expander cards died.

I want to see how many gpus I can hang on this thing before I get the dreaded "out of pcie resources" message again.

I am also going to be PSU limited because I don't have a PSU that I can run "outboard". If I understand it right, I would have to buy an HP hot swapable PSU to use the external adapter I have. I think the only reason I would probably have to do that is if I started running a lot more higher draw Gpu's. This is one of the advantages of a lot of gtx 1060 3GB's. They appear to draw around 85 watts per gpu (mostly).

Its MONDAY. Eat, Drink and be Merry for tomorrow we do (gulp) MAINTENANCE!

Tom

I haven't had any success with those 1 to 4x expander switches. They always end with stalled tasks after a while, and some machines refuse to even use them.
It would probably be best to avoid the cheaper versions, there are more expensive versions using x4 adapters, but they start around $200. Probably not worth the cost.

I just uploaded a bunch of cpu errors. They or something apparently locked the system up.

I have switched the Pcie speeds down to Gen2 and just installed the latest Ubuntu base (security) updates.

This thing doesn't overheat so what should I be looking for?

Tom

This is the explanation for -194 error code

ERR_ALREADY_RUNNING -194 - Another instance of BOINC Client is already running.

So for some reason, you appeared to have had another instance of BOINC running which you tried to process a work unit already being worked on by the initial BOINC instance.
Probably a fluke in timing. Restart BOINC after making sure there is no left over instance running in memory. Use the System Monitor to check. AP work units tend to hang around for a long time after exiting the Manager. Same for GPUGrid tasks.

Thank you. In my haste I didn't read it closely.

Got up this morning and the box was "hung" cycling through Bio boot code. I have reflashed back down to version 1.08 and am switching down to 7 gpus (the MB I can afford to replace this one with has 7 gpu slots, but only one cpu).

Tom

---

A proud member of the OFA (Old Farts Association).

I haven't had any success with those 1 to 4x expander switches. They always end with stalled tasks after a while, and some machines refuse to even use them.
It would probably be best to avoid the cheaper versions, there are more expensive versions using x4 adapters, but they start around $200. Probably not worth the cost.

Ok, then I should also investigate the higher priced version(s). Below $200 though :)

I tried my current 1 to 4 expanders on my AMD AM4 motherboards and they promptly limited me to 3 gpus no matter how I switched the plugs around. :(

Tom

---

A proud member of the OFA (Old Farts Association).

i think you should take your own advice, and once you get it running, stop messing with it.

---

Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

i think you should take your own advice, and once you get it running, stop messing with it.

I agree. If it had not been cycling through the Bios and not getting any further (this morning when I got up) I would not have made any other changes.

I had promised myself I would leave it alone.

So I have back flashed the bios to "the original" and reduced the gpus to 7 and left the pcie at Gen2.

If it runs without crashing or something else for a few days, maybe to next Tuesday, I will consider adding the 2 gpus I took offline back in.

Tom

---

A proud member of the OFA (Old Farts Association).

help please
one of my fast computer screen savers for seti says no running tasks this is been going on for over 12 hours
thanks from
kx1 h

---

help please
one of my fast computer screen savers for seti says no running tasks this is been going on for over 12 hours
thanks from
kx1 h

I suspect you are being hit by the Seti server issue. https://setiathome.berkeley.edu/forum_thread.php?id=83895 is where we talk about this topic.

If you check your event log, you may see something about "http internal error" and if you look in your transfer tab you may see a bunch of downloads that are in "back off" mode.

As I write this, the log jam seems to have cleared. So try using a manual retry on the last download in the transfer tab. That may restart your downloads. Also a manual update under the projects tab may help.

Tom

---

A proud member of the OFA (Old Farts Association).

I have threadripper CPU and Nvidia RTX2080Ti's, also Antminer ASIC (SHA256 or Bitcoin) however, I am an ethical 'whale' from Gridcoin. I am happy to try some ideas for setiathome . I just started researching if people are managing to use ASIC miners with BOINC and if so, how they interface them. Normally, for mining using a pool, the ASIC miner has an entry for the pool such as 'sha256.eu.mine.zpool.ca:3333'. The pool manages swopping to the most profitable SHA256 algo coin and other stuff (takes a fee). If someone could invent a 'BOINC ASIC pool' such as the old set up back in 2014 ( see here https://boinc.berkeley.edu/wiki/Release_Notes_for_BOINC_7.4 ) it may help with plans for the future. It used the XML file with quote... 'need to add a "coproc" XML element to your cc_config.xml configuration file'.

I think it all ended because Bitcoin Utopia ended up being dodgy. Please be aware of not mixing up the expired & defunct BOINC project called 'Bitcoin Utopia' with the state of Utopia concerning Bitcoin.

Adios amigos

your Antminer can only do one thing (hence ASIC). and that is process SHA256 hashes.

since SETI does not use this function, your Antminer ASIC will never work for this project.

---

Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

Opps. Somebody decided to transplant this thread from Q&A gpus over to "numbercrunching".

Tom

---

A proud member of the OFA (Old Farts Association).

It does turn out there is a Mining MB for AMD's AM4 sockets. It has 6 gpu slots. It may run as high as 7 gpus.
They are the Biostar MB's TA320-BTC and TB350-BTC. I am running the TB350-BTC with a 2700 cpu.

Tom

---

A proud member of the OFA (Old Farts Association).

Hey Tom, I just stumbled onto a mining system I hadn't seen mentioned before over at GPUGrid. The manufacturer is Onda and they have a 8 full slot and 12 full slot mining motherboard. So no need for risers.
Onda B250 D8P
https://imgur.com/a/e03wZMM
And here is a vendors page for the 8 slot version
https://www.banggood.com/ONDA-B250-D8P-D4-BTC-Mining-Motherboard-Main-Board-8-x-PCI-E-For-Intel-LGA1151-Platform-p-1302855.html?akmClientCountry=America&cur_warehouse=CN#customerQA
Seems the miners think highly of these models.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

pretty sure i've mentioned something like that in the Multi-GPU thread. If you're going to do it, that socketed Onda board is the one to get. Another company called Octominer makes a very similar product, but it comes with a 1.6GHz or so mobile CPU retro fitted to it, because crypto miners dont care about or need CPU resources. I did some testing in the past with SETI and very low power CPUs and it definitely negatively affected performance.

some caveats:
each slot, though full size, is still only 1 lane
unsure if they will sustain pcie 3.0, but they probably do. (miners dont care)
strange form factor that wont fit anything other than their own chassis, tends to be expensive for what you get
b250 chipset only supports up to 4c/8t CPUs, keep that in mind if you want to run -nobs

---

Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

I saw discussion on this vendor board in the mining fora and there were a few comments that the cpu is the hamstring with not enough lanes directly out of the cpu and every one of the slots basically being fed out of the PCH except for the first slot next to the cpu socket.

I still think that being able to plug a full length board into the native mobo socket is an advantage even if only at X1 speeds. With all the recent traffic on our threads about risers, cables and splitters being the weak link in the chain to using a multi-gpu platform, I still can see the appeal of this form factor. Even if it means having to come up with a special case or mounting. Seems the open air mining racks that you or Tom are using would not be a problem.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

you're definitely not wrong about cleaning up the setup. less cables and mess. and eliminates the problem of bad USB cables, which has been the weak link every time in my experience, and new cables even seems to have fixed Tom's issues with his cheap splitter, since his 9 GPU system is still rocking. not having to deal with that would be welcomed for sure.

---

Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

I saw discussion on this vendor board in the mining fora and there were a few comments that the cpu is the hamstring with not enough lanes directly out of the cpu and every one of the slots basically being fed out of the PCH except for the first slot next to the cpu socket.

I still think that being able to plug a full length board into the native mobo socket is an advantage even if only at X1 speeds. With all the recent traffic on our threads about risers, cables and splitters being the weak link in the chain to using a multi-gpu platform, I still can see the appeal of this form factor. Even if it means having to come up with a special case or mounting. Seems the open air mining racks that you or Tom are using would not be a problem.

Using the open rack of a mining case would not be a problem. The biggest single problem I can see is not being able to -nobs because there are not enough cores/threads in the available CPU (assuming you are shooting for more than 8 GPUs).

Some of the readme files for AMD gpu's appear to recommend 2 cpus / gpu task for Seti.

I like everything about what you have pointed to except it really needs a middle range cost 8t/16c CPU.

I am not sure that the 1151 socket supports any such thing.

Oh, well.
My AMD box crashed once after running nearly 5 days so I raised the CPU voltage a little more while leaving it at 3.7GHz. I also confirmed I have my DDR voltage at 1.35 and "everything" seems to be humming along.

Now if I could just get my pesky RAC up where "theoretical" calculations think it could go.... :)

Tom

---

A proud member of the OFA (Old Farts Association).

And here is a vendors page for the 8 slot version
https://www.banggood.com/Onda-B250-D8P-D3-8-PCI-E-Slots-Mining-Motherboard-For-Intel-LGA1151-Platform-p-1249172.html?rmmds=detail-left-hotproducts__1&cur_warehouse=USA
Seems the miners think highly of these models.

You might want to look at the RAM that board is setup to use.... I don't think that will work very well in a 8 GPU Linux system.
Meanwhile ASUS released a new BIOS a couple months ago aimed at stability, seems to work well.
The last slot diagram shows the bottom slots running normal AMD cards, I might try running nVidias in those last 6 slots soon.

Worth a try I guess, but I wouldn’t hold my breath on it working. I’ve never seen ANYONE get more than 13 normal GPUs working on that board. There’s a huge thread on bitcointalk about that board specifically.

https://bitcointalk.org/index.php?topic=2110421.620

It seems to be a hardware limitation. Can only pass 13 GPUs if you use some of the mining cards, like the P106 cards.

---

Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

the spoofed client only works with the number of GPUs it was built for. if you need a 14-GPU client, just ask Juan to build you one.

earlier versions of his client were using a simple multiplier on the cards detected. which may or may not be what you have done now. that might be a better approach with a high card count and if you anticipate changing your config around a lot between running different numbers of GPUs for testing or whatever.

---

Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

I picked up a used Biostar TB80 with G1840 celeron for under $80 total on eBay. It has 6 slots and I had 5 RX-560 working fine on Seti & Milkyway. with a 430watt supply (since upgraded) CPU has only two threads but that was OK for the RX-560 for those two apps. last time I looked on eBay I noticed the price has dropped down to $60 and ram is now included. It is running Ubuntu 18.04 with latest AMD driver and used ddr3 ram from retired systems.

If I had it do over I would have just bought the board by itself, under $40, and add the cpu. I ended up getting an i7-4790S later as I have other plans in the future for it. That chip was not cheap and I spent a lot of time looking for alternatives.

Unfortunate, the CPU prices for 1150 are not anywhere as cheap as the 1366 CPUs . If anyone knows of 1366 miners, let me know.