I am starting to reach the limit of what I can cool with my mining farm and am wondering the best approach to take to sustain further expansion. I am currently 14 "rigs" most with 5 GPUs per box. I will be expanding further next week and was hoping to put off this until after the winter, but that is looking less and less likely.
I am in a 800sqft space with 20ft ceilings and a 3 ton HVAC unit that now seems to run 24/7.
So the big question is: What do you do to cool your 10Gh+ mining farm? Do you have pictures and or best practices?
|
|
|
Great spreadsheet! This will be very useful going forward.
|
|
|
Thanks!
Need more people to jump on it, should help improve the stale rate for everyone. I also need to work out a load balancing arrangement.
I'll switch over to server2 in a few minutes. What were you thinking about for the load balancing solution? Nginx or a specific proxy solution?
|
|
|
Would you guys be willing to show the buffer amount you have for the pool within your stats page? Is there a reason for not having this transparency?
Tbh it doesn't matter in pure PPS as long as it keeps paying.. Interesting suggestion gigasvps. We don't actually have a fixed buffer; we rebalance the system wallet periodically to align it with outstanding balances and future member earnings. Your posts have me thinking about adding some kind of solvency indicator and a failsafe that shuts down the pool in case wallet balance threatens to creep below the member balances. That way miners can be confident that they can withdraw their earnings; even if the entire pool management should die in an air disaster while on corporate retreat. Regardless of any promises made by us or any other pool, we advise our miners to withdraw their earnings regularly. For reasons mentioned in another thread, https://bitcointalk.org/index.php?topic=34282.0, ABCPool does not intend to be a bank. We don't want to be holding on to large amounts of BTC for our members. Also, I don't see a found blocks page. Do you not show when you find blocks?
I guess these things are not absolutely necessary to run a pps pool...
We know that our stats are currently pretty basic. We plan to introduce extra stats on the individual level, so users have more insight into their performance. We expect to work on that after this week. Any updates on these requests? I would like to mine at your pool but feel the transparency is necessary to grow. Also, from a pure vanity standpoint, i would also like to see some type of leader board.
|
|
|
Thanks, kirax. my desktop machine is dual purpose: Win7 for mundane work & gaming, mining with BAMT when it isn't being used for those tasks. I understand that Putty is needed for Win7, but how to I access/control the two other miners when my desktop is also mining with BAMT? Is there a version of Putty for Debian linux? I've read about people using "ssh" from a linux box to control other linux boxes, but have no idea how to make that work.
From any linux box run the command ssh root@192.168.1.111This command will allow you to ssh to any other linux box. If you have created another user on the system, replace root with that user's name.
|
|
|
Thank you for all of the feedback. I really appreciate you guys willing to be so helpful. I am only having these problems on 2 machines out of 13 so the % that is dropping out is quite low. I will definitely take the clocks down and make sure the cards are spaced as far apart as possible.
I now have a couple weeks full time on this so I will be making the appropriate adjustments from your recommendations.
|
|
|
You can still use top to monitor the processes, or screen -r gpuX to see the individual miners. But it still sounds like a card crashing, mine sometimes takes hours or days.
If one card crashes it takes the rest of them down with it or this a driver issue where a card crashes and none of the other cards can function?
|
|
|
Sounds like one of your cards is hanging. gpumon and atitweak don't work for me if one hangs (or crashes?? IDK the correct term). type top and look for a phoenix instace at 100% cpu. I have been able to edit the delay times in the .conf file and run gpumon quick enough to see which adapter is causing the issue.
I should have given more info. The computer ends up hanging randomly, hours or days after it was started. With the 5970s, i did need to up the wait time between starting gpus from 3 to 6 as the computer would sometimes hang when starting the mine process.
|
|
|
I am having an issue where mining on a rig will hault for no apparent reason. I can still ssh into the box and if I reboot, everything starts up fine again. Also, when I try to access gpumon after ssh-ing into the box, the process seem to be hung.
Is there a way to monitor the phoenix processes and if they become hung, restart them or the box itself?
|
|
|
Too bad I already got five of these ![Sad](https://bitcointalk.org/Smileys/default/sad.gif) should I get 3 more? Be greedy when others are fearful. So yes, buy three more, I did. ![Shocked](https://bitcointalk.org/Smileys/default/shocked.gif)
|
|
|
Looks like they are out of stock now.
|
|
|
Has anyone had a similar or the same problem while migrating to BAMT? I already found this one: http://blog.zorinaq.com/?e=46, but there's said that 8 GPU work fine whereas 10 GPU are currently the limiting number. I am glad it isn't just me. Unfortunately I am not a rocket surgeon when it comes to linux, so I think I'm just going to do 3 cards per box. If this is fixed, awesome, if not, it definitely won't derail my plans. Thanks again for BAMT, it's awesome.
|
|
|
What is your mining set up like gigasvps?
I'll post some pics soon. Basically a bunch of 5830s, 6950s and 5970s in aluminum and wood open air cases. Still building computers....
|
|
|
hi, I just purchased a 6950 and was wondering how to get this baby to go over 400 mhashs LOL
If you can unlock the extra shaders on the card, you could potentially get to 400Mh/s. Otherwise, you'll probably end up at 900/500 doing about 370Mh/s. YMMV
|
|
|
I read thru this whole post because I just bot 3 of these cards from amazon. I have a nice 1200 psu and I am hoping to get all 6 GPUs up to 420mhash. with linuxcoin 2.5G with one PSU. but I am a little worried about the heat on the 2nd gpu someone was talking about...
anyone still running this thing?
I have my first 3 running at 825/300 doing 380Mh/s per GPU. Anything above this seems to be unstable since you are pulling so much juice through the motherboard. You would probably need 16x extenders with the extra molex to be able to do higher Mh/s per GPU.
|
|
|
Anyone running 8 GPUs with bamt? I have 4 5970s i am trying to setup on a single machine, here's the setup: 2 - thermaltake 850w single rail PSUs 1 - ASRock Extreme4 MB -> http://www.newegg.com/Product/Product.aspx?Item=N82E168131572623 - 16x extenders, no extra molex 1 - 1x extender, no extra molex 4 - OEM 5970 that just came in from newegg.com So, when i had 8 GPUs installed, BAMT died when it tried to display the GUI. So it made it through the BAMT startup screen and the initial startup. I removed the GPU on the 1x extender and all is well. Is this a problem of me needing a MB with 4 16x PCIe slot or is it a problem with BAMT?
|
|
|
Ahhh, perfect. Exactly what I was looking for. BAMT FTW!
|
|
|
Check for a blank line after second pool. I use nano to edit the file and every time I open a pool file it will insert a blank line - even if I don't change anything (open the file, hit down arrow 3 times and see where you land). You can test failover manually by hitting ctrl C in the miner screen to see if it will loop back.
Thanks for the tip. I too use nano, but didn't think it would create a new line when you just opened the file. ![Huh](https://bitcointalk.org/Smileys/default/huh.gif) Maybe BAMT should ignore empty lines in the pool file?
|
|
|
|