Bitcoin Forum
September 22, 2024, 05:45:51 PM *
News: Latest Bitcoin Core release: 27.1 [Torrent]
 
   Home   Help Search Login Register More  
Pages: [1]
  Print  
Author Topic: Bitfury rig seems a bit more temperature-sensitive than most  (Read 1687 times)
salfter (OP)
Hero Member
*****
Offline Offline

Activity: 651
Merit: 501


My PGP Key: 92C7689C


View Profile WWW
July 11, 2014, 03:56:19 PM
 #1

Summer has well and truly arrived in Las Vegas.  I had moved all my mining hardware out into the garage a while back to get the heat and noise out of the house.  Most of my gear is working OK in triple-digit temperatures...fans speed up a bit, but they keep on moving.

The exception, though, is my Bitfury rig (nine H-boards with the appropriate backplane...the one that looks like a bunch of PCIe 1x slots).  In the cooler months, I was getting about 270 GH/s from it easily.  Every once in a while, it'd lose communication with some of the chips; restarting bfgminer would get it back on track.  As the weather got warmer, though, it started getting stuck more and more often.  I backed off the speed a bit.  It worked a while longer, then started getting wedged again.  I found that cgminer had added support for these systems (--enable-bab) and would adjust clock speeds automatically to optimize performance, so I built it and switched over from bfgminer.  It'll keep running now, but as the temperature outside pushed past 110 last week, hashrate fell down to about 120 GH/s!

I have heatsinks on the backside of the boards opposite each mining ASIC and voltage regulator.  I have it in a Spotswood frame with three fans good for over 270 cfm between them.  The other miners (two BFL Jalapeños, two Antminer S1s, a Gridseed blade miner, and a handful of Gridseed orb miners) are stock, except for some of the Gridseed orbs which I stacked together with standoffs so they'd take less space.  These other miners haven't skipped a beat.

Since the Raspberry Pi in the Bitfury rig was also controlling the BFL and Gridseed miners, I needed to set up another to take over for them when I moved the Bitfury into the laundry room.  Mining speed picked up to 200 GH/s...still not back to 270.  One of the fans had worn out, so I replaced it. (The fan was working before I moved the rig, but it had gotten noisy. Finding high-speed fans that aren't riced out with LEDs is harder than it should be, but that's fodder for another rant.) With the new fan in place and the rig still in the laundry room, it's finally back to mining at 270 GH/s.

Tipjars: BTC 1TipsGocnz2N5qgAm9f7JLrsMqkb3oXe2 LTC LTipsVC7XaFy9M6Zaf1aGGe8w8xVUeWFvR | My Bitcoin Note Generator | Pool Auto-Switchers: zpool MiningPoolHub NiceHash
Bitgem Resources: Pool Explorer Paper Wallet
lowbander80
Legendary
*
Offline Offline

Activity: 1036
Merit: 1000


View Profile
July 13, 2014, 10:04:04 AM
 #2

i DONThave experience with bitfury but antminers and dragons are not affected with temps reaching 40c
they run the same overheat off-course but run Ok
tarmi
Legendary
*
Offline Offline

Activity: 1232
Merit: 1011


View Profile
July 13, 2014, 10:16:48 AM
 #3

not only that bitfury rigs are temperature sensitive, but I found out that they are also vibration sensitive.

I had a lot of troubles with one of my rigs because of that. I had performance dropping spontaneously, for like 50 %, only because the H-boards were not sitting firmly on those ISA style connectors and the air pushed by fans was moving the boards a bit, causing a lot of MISO and other errors.

xstr8guy
Hero Member
*****
Offline Offline

Activity: 784
Merit: 1004


Glow Stick Dance!


View Profile
July 13, 2014, 10:20:16 AM
 #4

Bitfury rigs are sensitive. Period.

I have 4 rigs. None have ever worked with all 16 boards. So my rigs each have 12 to 15 cards each. Some run pretty stable but others need to be nursed back to health virtually every day. Nursing them back to health involves restarting, wiggling or swapping boards around and general voodoo practices. Mine are in air conditioning though.

I've been running mine nonstop since last autumn and love them to death but they are slowly dying. Both the H boards and the M boards seem to be losing stability and croaking one by one.

By the way, I'm in Vegas too. I have 3 S1's and 2 Super Jupiter modules in my oven (aka garage). And they don't seem to be bothered by the heat at all. In fact I just fired up (bad choice of words?!) my S1's a week after I shut them down to run them at bitsolo.net for shits-n-giggles.
tarmi
Legendary
*
Offline Offline

Activity: 1232
Merit: 1011


View Profile
July 13, 2014, 10:26:21 AM
 #5

yeah, I had also issues when the rigs were fully populated.

but I obtained some additional m boards, and they are working perfectly now. more then 500 GH for 16 h-boards.

no sign of dead hardware, or anything. running since october last year.
wttbs
Legendary
*
Offline Offline

Activity: 2210
Merit: 1109



View Profile
July 13, 2014, 11:56:40 AM
 #6

If you have issues with heat emission from the H-boards V1, you should try to cool the back of the PCB. If you put a heatsink on top the bitfury chip it won't do very much, it only helps a little. Most heat generated by the bitfury chip comes from the bottom of the chip -> back of the PCB. Try to fit a heatsink on the back of PCB but I know there is little space for that.

We designed a new version of the M + H board concept and made little holes in the PCB where the Bitfury (rev2) chips are mounted, so we have better heat emission. unfortunatley you can't make those holes in the PCB afterwards..  Undecided



An other option is to undervolt the boards, if you lower the input voltage to 11.8, 11.7, 11.5 or maybe even 11V the boards should still work, the hashrate will be at a lower rate but it won't get as hot as on 12V and this way the total hashrate might not be optimal but higher than de 200GHs I guess. Just try it out.
salfter (OP)
Hero Member
*****
Offline Offline

Activity: 651
Merit: 501


My PGP Key: 92C7689C


View Profile WWW
July 15, 2014, 06:11:05 PM
 #7

If you have issues with heat emission from the H-boards V1, you should try to cool the back of the PCB. If you put a heatsink on top the bitfury chip it won't do very much, it only helps a little. Most heat generated by the bitfury chip comes from the bottom of the chip -> back of the PCB. Try to fit a heatsink on the back of PCB but I know there is little space for that.

I have heatsinks on the backs of my boards:



(It's an old picture from when I only had two H-boards, but all my H-boards are set up like this.)

Tipjars: BTC 1TipsGocnz2N5qgAm9f7JLrsMqkb3oXe2 LTC LTipsVC7XaFy9M6Zaf1aGGe8w8xVUeWFvR | My Bitcoin Note Generator | Pool Auto-Switchers: zpool MiningPoolHub NiceHash
Bitgem Resources: Pool Explorer Paper Wallet
klondike_bar
Legendary
*
Offline Offline

Activity: 2128
Merit: 1005

ASIC Wannabe


View Profile
July 16, 2014, 12:19:04 PM
 #8

If you have issues with heat emission from the H-boards V1, you should try to cool the back of the PCB. If you put a heatsink on top the bitfury chip it won't do very much, it only helps a little. Most heat generated by the bitfury chip comes from the bottom of the chip -> back of the PCB. Try to fit a heatsink on the back of PCB but I know there is little space for that.

I have heatsinks on the backs of my boards:



(It's an old picture from when I only had two H-boards, but all my H-boards are set up like this.)

I have done this with all my boards, plus some pencil modding (v1.2 boards) to achieve about 34-38GH per board. however, it stops mining every so often (about twice a month) and needs a reboot to go back to normal operation

24" PCI-E cables with 16AWG wires and stripped ends - great for server PSU mods, best prices https://bitcointalk.org/index.php?topic=563461
No longer a wannabe - now an ASIC owner!
Pages: [1]
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!