Message boards :
Number crunching :
Bitcoin GPU-based Mining Machines good for BOINC / SETI?
Message board moderation
Previous · 1 . . . 36 · 37 · 38 · 39 · 40 · 41 · 42 . . . 66 · Next
Author | Message |
---|---|
Ian&Steve C. Send message Joined: 28 Sep 99 Posts: 4267 Credit: 1,282,604,591 RAC: 6,640 |
are you overclocking the cards? Seti@Home classic workunits: 29,492 CPU time: 134,419 hours |
eng4hire Send message Joined: 25 Dec 10 Posts: 490 Credit: 34,584,466 RAC: 0 |
No. I dont even know how to do that. |
Tom M Send message Joined: 28 Nov 02 Posts: 5124 Credit: 276,046,078 RAC: 462 |
Thanks for the acid test Tom. It will be nice to see it go. I am playing with a Amd 2400G and it becomes non-responsive to the keyboard/mouse after I leave it alone. But I am not convinced it has frozen, it is just non-responsive :( Is it possible yours is non-responsive? Rather than frozen? Tom A proud member of the OFA (Old Farts Association). |
eng4hire Send message Joined: 25 Dec 10 Posts: 490 Credit: 34,584,466 RAC: 0 |
Let me verify when I get home next time I see it "freeze". If the GPU's are cold, I say the system froze. At least two of my Dell farm units were cold this morning. How MUCH RAM DOES LINUX TAKE TO RUN WELL? Plus the 1 GB per work unit? Each Dell unit has 4G of RAM. Doesn't seem enough... |
eng4hire Send message Joined: 25 Dec 10 Posts: 490 Credit: 34,584,466 RAC: 0 |
See? I think the miner is down again... My AVG Credit is dropping... |
-= Vyper =- Send message Joined: 5 Sep 99 Posts: 1652 Credit: 1,065,191,981 RAC: 2,537 |
Thanks for the acid test Tom. It will be nice to see it go. You can, But you need to install it on another computer on the network , for instance a little Raspberry Pi. Get the mail going in linux and Smokeping and configure it to drop you a mail if u got no ping from a machine or so. There are even programmable outlets with a LAN port that can be configured for these purposes to monitor a machine on a specific outlet with its ip , and if it doesn't respond, trigger a power on/off to get it going again. https://www.snapav.com/shop/en/snapav/surge-conditioning/wattbox-reg%3B-ip-12-controlled-outlet-chassis-power-conditioner-with-auto-reboot-wb-700ch-ipv-12 _________________________________________________________________________ Addicted to SETI crunching! Founder of GPU Users Group |
Ian&Steve C. Send message Joined: 28 Sep 99 Posts: 4267 Credit: 1,282,604,591 RAC: 6,640 |
See? I think the miner is down again... My AVG Credit is dropping... dont worry about anything today. the project has been mostly down for everyone, it's been having problems ever since they took it down for maintenance yesterday. everyone's RAC is dropping today. a better indication to see if the system froze, is to simply look at the screen and see what it is doing, could be frozen, could be network issues. or check the stats on your profile page. check when it last contacted the servers (the rightmost field in your list of computers section) Seti@Home classic workunits: 29,492 CPU time: 134,419 hours |
Ian&Steve C. Send message Joined: 28 Sep 99 Posts: 4267 Credit: 1,282,604,591 RAC: 6,640 |
Let me verify when I get home next time I see it "freeze". 1GB per GPU work unit is enough, the special app only reserves like 650MB per instance. what you need to do if running on low system memory though, is to change the RAM usage that BOINC is allowed to use. by default it's set to 50%. so BOINC wont ever use more, even if you have it. setting it higher will prevent you from running to those issues where system memory is preventing more work from being done. Seti@Home classic workunits: 29,492 CPU time: 134,419 hours |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
It lives, Coprocessors: [11] NVIDIA GeForce GTX....again. I hope the third time's a charm. The first FinTech board died just short of three months, the second one lasted about an hour. NewEgg is going to be replacing the second one soon. The current one is an ASUS Expert which has dropped in price quite a bit lately. Seems the problem with the ASUS board was some conflict with the pair of 1050s...they kept stalling after a few hours. I replaced those two with a pair of 950s and the stalling seems to have stopped. I would try a 12th card, but all I have left is some 750 Ti which don't get recognized, and the 1050s. I do have some 1070s, but those will stay with the Hackintosh which also just got a 'new' motherboard and seems to be working better. Now I need to start bothering Gigabyte about that first FinTech board seeing as how the same ram & cpu work fine in the ASUS board. Then decide what to do with the second FinTech board when it returns. The problem with the Gigabyte boards was they just stopped booting, no lights, no fans, no signs of life whatsoever....dead. I did try jumping the PS so it would power up by itself, one board had lights for a few seconds and then nothing, the other board powered up but would never post. The price on these things just keep dropping. |
Tom M Send message Joined: 28 Nov 02 Posts: 5124 Credit: 276,046,078 RAC: 462 |
It lives, Coprocessors: [11] NVIDIA GeForce GTX....again. Ouch. I hope your experience is not my experience. I just ordered a Gigabyte AM4 board for another high gpu experiment I am intending to run. Tom A proud member of the OFA (Old Farts Association). |
Tom M Send message Joined: 28 Nov 02 Posts: 5124 Credit: 276,046,078 RAC: 462 |
@eng4hire, It looks like your 9 gpu rig hasn't contacted the server in a while. Is it down? Would it run with 8? Tom A proud member of the OFA (Old Farts Association). |
Tom M Send message Joined: 28 Nov 02 Posts: 5124 Credit: 276,046,078 RAC: 462 |
It lives, Coprocessors: [11] NVIDIA GeForce GTX....again. Hurray!!! +1 Tom A proud member of the OFA (Old Farts Association). |
Tom M Send message Joined: 28 Nov 02 Posts: 5124 Credit: 276,046,078 RAC: 462 |
It lives, Coprocessors: [11] NVIDIA GeForce GTX....again. It looks like you have both the gtx 1050's back in there as well as a couple 960's and a 950. Its amazing how good a production level you are getting with a bunch of gtx 1060's and the rest. Excellent bang for the buck. The Gtx 1060 3GB prices on eBay are approaching $100. They were at $130. And I bought most of mine for $150 :) Tom Tom A proud member of the OFA (Old Farts Association). |
Tom M Send message Joined: 28 Nov 02 Posts: 5124 Credit: 276,046,078 RAC: 462 |
So exactly what Mining Motherboards are all of you running? Tom A proud member of the OFA (Old Farts Association). |
Ian&Steve C. Send message Joined: 28 Sep 99 Posts: 4267 Credit: 1,282,604,591 RAC: 6,640 |
I'm not running a mining board, just a normal consumer level ATX board. but it'll run 7-8GPUs fine. good enough for 700,000RAC it seems lol. seeing the issues from the actual mining boards isnt surprising, and enough to make me shy away from them. they are designed to run mining, which doesnt care about PCIe speed. running 10 GPUs at PCIe gen 3.0 is a lot harder to maintain than running 10 GPUs at gen 2.0 or even 1.0. most boards that have a "mining mode" feature, one of the things it does is usually setting the PCIe gen to 1.0 for easier compatibility. and doing this doesn't cause any issues for mining since the bandwidth requirements are so low. the other problem with the mining boards is that you are limited in CPU capacity. all of the mining boards, because of the socket and CPUs they support, you will be limited to a 4core/8thread chip (all H110, B250, Z270, etc series boards). i like to have an entire core dedicated to GPU work and run -nobs, to maximize the performance loss from running on the x1 link. One can run without -nobs of course, and plenty do, I just personally prefer to keep the speed. there is just one (1) mining specific board that has the ability to run a higher thread count CPU, and that is the ASUS H370 Mining Master board, in which you can add up to an i9-9900k with 8 cores and 16 threads, and it can support 13-14 normal GPUs (from what i read). but then you're looking at a $500 CPU and a board with anemic VRMs and there's probably a better way... the last problem comes down to just build quality and build intent. none of these consumer level boards, mining boards or otherwise, were designed with this kind of 24/7 use, and certainly not with all the PCIe devices crammed into them. I haven't had any real problems with my z270 consumer board for the 7-GPU system at the hardware level, but I understand that it's just a matter of time. the mining boards are made fairly bottom dollar at that, cheaper boards that "get the job done" = theoretically more profits for the miner, that is if you believe their marketing. so what's my next step? this guy: Supermicro X9DRX+-F 10x PCIe 3.0 8x slots 1x PCIe 2.0 4x (in 8x physical slot) supports 2x E5-2600(v2) chips with 40 PCIe lanes each (80 total for the 10 8x slots) This is a board/setup that's actually designed for constant use. I'll be likely combining my 7-GPU system with the [2] RTX2070 system and harvesting the CPUs (2x E5-2690 8c/16t) and 32 GB DDR3 Reg ECC ram out of it. I'll also likely be still using the 1x USB risers at first, but eventually switching it all over to using full 8-lane shielded PCIe 3.0 risers. Seti@Home classic workunits: 29,492 CPU time: 134,419 hours |
Tom M Send message Joined: 28 Nov 02 Posts: 5124 Credit: 276,046,078 RAC: 462 |
Seems like you mentioned that before and it is a solid $700 MB? But at least I could reuse my cpus and probably my ram :) Not to mention all my gpus :) What size? Extended-ATX or custom/larger? Tom A proud member of the OFA (Old Farts Association). |
rob smith Send message Joined: 7 Mar 03 Posts: 22199 Credit: 416,307,556 RAC: 380 |
Link to the spec: https://www.supermicro.com/products/motherboard/xeon/c600/x9drx_-f.cfm Five year old design, but still a half decent bit of kit. Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
Ian&Steve C. Send message Joined: 28 Sep 99 Posts: 4267 Credit: 1,282,604,591 RAC: 6,640 |
depends where you buy it. Cheapest i was able to find was on amazon for $575 incl. shipping. I had $50 Amazon gift cards, so they went to use here. The board is brand new. I would have preferred to buy a used board for some more savings, but this board is incredibly rare and I couldn't find any used boards available for sale (at a decent price, some on ebay, but asking for more than the new board from amazon). the board is a proprietary supermicro form factor, designed for use with some of their server cases, it will not fit in any standard case. I will be mounting this on an open air style frame for now, like this: Link to the spec: indeed. but the plus sides are worth it. older design = uses older CPU and memory, which is MUCH cheaper. I already have some CPUs and memory that I plan to use from another system of mine, so no additional costs there, but if I were looking to buy CPUs and ram as well, E5-2600 v1 and v2 chips are cheap on ebay, as is Registered ECC DDR3 server ram. Supermicro has an X10 version of this board also with the same naming scheme (X10DRX), same 11 8x PCIe slots, but uses v3/v4 chips and DDR4 ram, which are much more expensive and would increase the cost substantially. Seti@Home classic workunits: 29,492 CPU time: 134,419 hours |
Tom M Send message Joined: 28 Nov 02 Posts: 5124 Credit: 276,046,078 RAC: 462 |
@Ian&SteveC, I now see two of your systems with huge gpu #'s in the listings but in fact 2 gpus or the usual 9. Your RAC has gone up on both of them but I suspect the actual production may not have? If the actual production has gone up, exactly what did you do to tweak them? Tom A proud member of the OFA (Old Farts Association). |
Ian&Steve C. Send message Joined: 28 Sep 99 Posts: 4267 Credit: 1,282,604,591 RAC: 6,640 |
actual production hasn't really increased. everyone's RAC has been increasing over the last few weeks due to the new diversity in WU's being sent out, basically we are just getting a higher than "normal" credit per unit time. the rate at which it is working through WU's is the same. The RAC on the 2070 machine is increasing however because i recently added a second 2070 to it. the 2070 machine has 2 GPUs, the 1080ti machine has 7 GPUs the GPU count just keeps the size of the cache higher (the project servers only send you 100 WU per GPU, more GPUs = more work given), so there is little or no downtime for the system when the project either goes down randomly due to server issues, or during routine Tuesday maintenance. It helps dampen out the large dips in RAC that would otherwise occur due to these scenarios. Seti@Home classic workunits: 29,492 CPU time: 134,419 hours |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.