The bitcointalk.org forum is great and contains lots of info about the bitcoin, market, software, hardware, and so on. It’s a worldwide community and most people are sharing lots of info and help people who have some problems or questions.
Recently I bought 10 Hashbuster Nanos and today I took 1 board away from my tower and started to experiment with it. Try to get it hashing at a higher rate. Standard hashrate is bewteen 2.5 and 2.7 Gh/s. On the board you can play with the voltage which feeds the ASIC chip. Higher voltage means higher hashrate but also more heat! So (extra) cooling is essential. Bitcointalkmember XBOCT posted some very useful tips and tricks.
I did several tests with the board:
- 0.981 V -> hashrate: 2.5 – 2.6 Gh/s = stock / unmodified
- 1.025 V -> hashrate: 2.6 – 2.7 Gh/s = test run #1
- 1.050 V -> hashrate: 2.7 – 2.9 Gh/s = test run #2
- 1.124 V -> hashrate: 2.9 – 3.07 Gh/s = test run #3 going nuts
What the maximum voltage is for the ASIC-chip I don’t know, I have to find out later. Have the guts to nuts and just do some crazy experiments 🙂
As XBOCT mentioned, the bottom of the chip gets realy hot too, so an extra heatsink on the back of the board is no luxery. I scraped off the coating on the board on the back, just locate the small square with the little holes. Then I put a small heatsink with some coolingpaste on the board. XBOCT used 3M Thermal Heatsink Tape but I just have some lowbudget paste, so I used a tiwrap to keep the heatsink in place. A fan provides some fresh air and I pointed it at the back of the board as I felt with my fingers the back is hotter than the front.
For now I am just curious if the board survives the night and what the avarage hashspeed will be over the next 8 to 12 hours.
The Experiment-test-computer is an old Pentium 4 with 512MB RAM, a 20GB sata disk. Debian Wheezy Linux is the operating system. On the picture you also see two “oldskool” Cairnsmore CM1 boards, they are hashing @ 800Mh/s each…
Update Sunday Oct 27: Last evening I noticed that the hashrate dropped to 2.7 Gh/s after only some minutes.. Back to the forum. I’ve got another tip from XBOCT regarding cgminer settings.
My version of cgminer uses clock auto selection in range default+1 — default-1 Default clock value can be found in driver-bitfury.c: “#define BITFURY_DEFAULT_CLOCKBITS …” Internal clock generator is voltage dependent, so optimal clock value is in the range 53-54 for 0.9V or 51-53 for 1.1V. Try to play with default clock value in source code or by command line (note: double ‘=’ in command line)
I added the options –bitfury-clockbits==52 and it starts hashing @ 3 Gh/s I looked at it for 5 minutes and it stayed at 3. This sundaymorning I took a look again: back 2.7 🙁
cgminer sometimes gives the message: Frequency (55) not changed for chip 0
Looks like cgminer loses it’s 52 clockbit setting and automaticly switches back to some default, 55 in this case. Next mission: investigate and modify the driver-bitfury.c
Update 10:00h: modified driver-bitfury.c ,compiled and make cgminer again. Open driver-bitfury.c with a texteditor, in the beginning you find the right code:
#define USB_VID 0xfa04
#define USB_PID 0x0011
//#define SPIMAXSZ 256*1024
#define BITFURY_DEFAULT_CLOCKBITS 52
then save the file and exit the texteditor.
./configure --disable-opencl --disable-adl --disable-bitforce --disable-icarus --disable-modminer --disable-ztex --disable-scrypt --enable-bitfury
sudo make install
Now start cgminer. It looks goods now (after a few minutes)
cgminer version 3.5.1 – Started: [2013-10-27 10:28:13]
(5s):2.826G (avg):3.174Gh/s | A:124 R:2 HW:0 WU:44.3/m
ST: 2 SS: 0 NB: 1 LW: 183 GF: 0 RF: 0
Connected to eu-stratum-lb489kj.btcguild.
Block: 000af5f5159e7580… Diff:391M Started: [10:28:13] Best share: 47
[P]ool management [S]ettings [D]isplay options [Q]uit
BFR 0: R:0 Cl:52 D1: 107 | 3.274G/3.174Gh/s | A:124 R:2 HW:0 WU:44.3/m
[2013-10-27 10:30:05] Accepted 6c6b6e85 Diff 2/2 BFR 0
[2013-10-27 10:30:06] Accepted 638d4efa Diff 2/2 BFR 0
[2013-10-27 10:30:12] Accepted 3d59119f Diff 4/2 BFR 0
[2013-10-27 10:30:13] Accepted 384a1b6b Diff 4/2 BFR 0
[2013-10-27 10:30:14] Accepted 438b55d4 Diff 3/2 BFR 0
Board is stable and hashrate is 3 or above ! 🙂