Bitcoin Forum
March 28, 2024, 11:32:05 PM *
News: Latest Bitcoin Core release: 26.0 [Torrent]
 
   Home   Help Search Login Register More  
Pages: « 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 [16] 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 »
  Print  
Author Topic: X6500 Custom FPGA Miner  (Read 219788 times)
Qoheleth
Legendary
*
Offline Offline

Activity: 960
Merit: 1028


Spurn wild goose chases. Seek that which endures.


View Profile WWW
January 04, 2012, 07:32:46 PM
 #301

Code:
2012-01-04 11:11:49 | Long-poll: ValueError! No JSON object could be decoded

Maybe this has something to do with ABC Pool though? I'll try out a different pool.
For what it's worth, I'm on Eligius and have not been seeing such errors.

Also still getting lots of these "Golden nonce found" errors, not sure if they actually matter though?
That's not an error, really. It's just an announcement by the FPGA that it's found something that it thinks is a qualifying hash. For me, at least, every such announcement is followed by an "accepted" or "rejected" indicating whether the server agrees on the matter.

If there is something that will make Bitcoin succeed, it is growth of utility - greater quantity and variety of goods and services offered for BTC. If there is something that will make Bitcoin fail, it is the prevalence of users convinced that BTC is a magic box that will turn them into millionaires, and of the con-artists who have followed them here to devour them.
The Bitcoin software, network, and concept is called "Bitcoin" with a capitalized "B". Bitcoin currency units are called "bitcoins" with a lowercase "b" -- this is often abbreviated BTC.
Advertised sites are not endorsed by the Bitcoin Forum. They may be unsafe, untrustworthy, or illegal in your jurisdiction.
1711668725
Hero Member
*
Offline Offline

Posts: 1711668725

View Profile Personal Message (Offline)

Ignore
1711668725
Reply with quote  #2

1711668725
Report to moderator
1711668725
Hero Member
*
Offline Offline

Posts: 1711668725

View Profile Personal Message (Offline)

Ignore
1711668725
Reply with quote  #2

1711668725
Report to moderator
freshzive
Sr. Member
****
Offline Offline

Activity: 447
Merit: 250


View Profile
January 04, 2012, 08:12:23 PM
 #302

Code:
2012-01-04 11:11:49 | Long-poll: ValueError! No JSON object could be decoded

Maybe this has something to do with ABC Pool though? I'll try out a different pool.
For what it's worth, I'm on Eligius and have not been seeing such errors.

Yeah switching to Eligius completely abolished this error. Something specific to ABCPool.

sirky
Sr. Member
****
Offline Offline

Activity: 404
Merit: 250



View Profile
January 04, 2012, 09:16:12 PM
 #303

Version 0.2 is out!

The main changes in this are:
  • Long-polling
  • Progress line during bitstream pre-preprocessing
  • And probably most importantly, fixed a bug in formatting data for share submission. Thanks to m0mchil for finding this! In my little bit of testing so far, this has reduced the number of rejects down to maybe 2%. I didn't want to wait to get this out, so please report the reject rates that you find in your own testing.

Download it from Github here.

Thanks again, m0mchil!!

m0mchil is the man.
freshzive
Sr. Member
****
Offline Offline

Activity: 447
Merit: 250


View Profile
January 05, 2012, 12:52:22 AM
 #304

Ran v0.2 this afternoon for about 5 hours, worked great. Thanks for the update Smiley

Code:
Run Summary:
-------------
Device: 0
Serial: AH00WOVL
JTAG chain: 2
Number of FPGAs: 2
Running time: 5h4m
Getwork interval: 30 secs
Chain 0:
  Accepted: 576
  Rejected: 4 (0.69%)
  Invalid: 1 (0.17%)
  Accepted hashrate: 135.50 MH/s
  Hashrate w/ rejects: 136.44 MH/s
Chain 1:
  Accepted: 523
  Rejected: 4 (0.76%)
  Invalid: 1 (0.19%)
  Accepted hashrate: 123.03 MH/s
  Hashrate w/ rejects: 123.97 MH/s
Total hashrate for device: 258.54 MH/s / 260.42 MH/s

tf101
Newbie
*
Offline Offline

Activity: 18
Merit: 0


View Profile
January 05, 2012, 02:13:27 AM
 #305

Certainly less errors:

Run Summary:                                                     
-------------
Device: 0
Serial: AH00WOWE
JTAG chain: 2
Number of FPGAs: 2
Running time: 1h13m
Getwork interval: 20 secs
Chain 0:
  Accepted: 134
  Rejected: 4 (2.90%)
  Invalid: 0 (0.00%)
  Accepted hashrate: 130.63 MH/s
  Hashrate w/ rejects: 134.53 MH/s
  Hashrate w/ invalids: 134.53 MH/s
Chain 1:
  Accepted: 111
  Rejected: 4 (3.48%)
  Invalid: 0 (0.00%)
  Accepted hashrate: 108.21 MH/s
  Hashrate w/ rejects: 112.11 MH/s
  Hashrate w/ invalids: 112.11 MH/s
thirdlight
Sr. Member
****
Offline Offline

Activity: 445
Merit: 250



View Profile
January 05, 2012, 06:39:29 AM
 #306

I'm seeing rejects between 1% and 4%, with hash rates of around 250 MH/s.

tgmarks
Donator
Hero Member
*
Offline Offline

Activity: 490
Merit: 500


View Profile
January 05, 2012, 06:45:43 AM
 #307

I'm pulling a consistent 260 mh/s with and average of 0.9% reject

li_gangyi
Full Member
***
Offline Offline

Activity: 157
Merit: 100



View Profile
January 05, 2012, 11:08:45 AM
 #308

Pre-fix was getting 20% avg rejects. Now I'm down to under 3%.
freshzive
Sr. Member
****
Offline Offline

Activity: 447
Merit: 250


View Profile
January 05, 2012, 04:24:09 PM
 #309

Another user had this problem on Linux, and it turned out it was because he used the new D2XX drivers that came out on 12/23. I haven't had a chance to hack PyUSB to work with that, so please try it with 1.0.4 (download here). Please let us know if that works for you!

Tried this, now I'm getting a different error:

Code:
Traceback (most recent call last):
  File "program.py", line 23, in <module>
    from ft232r import FT232R, FT232R_PortList
  File "/Users/christian/x6500/ft232r.py", line 22, in <module>
    import d2xx
  File "/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/d2xx/__init__.py", line 24, in <module>
    from _d2xx import *
ImportError: dlopen(/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/d2xx/_d2xx.so, 2): Symbol not found: _FT_Close
  Referenced from: /Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/d2xx/_d2xx.so
  Expected in: dynamic lookup

imac:x6500 chris$ python2.6 --version
Python 2.6.7

Running Python 2.6.7, using the D2XX 1.04 drivers and the PyUSB-0.6 from fpamining.com . I've tried messing around with different versions but nothing I do will get it to work in Lion. Works beautifully in Win7 though.

Any other suggestions? Thanks for the help.

fizzisist (OP)
Hero Member
*****
Offline Offline

Activity: 720
Merit: 525



View Profile WWW
January 05, 2012, 06:00:10 PM
 #310

Running Python 2.6.7, using the D2XX 1.04 drivers and the PyUSB-0.6 from fpamining.com . I've tried messing around with different versions but nothing I do will get it to work in Lion. Works beautifully in Win7 though.

Any other suggestions? Thanks for the help.

Sorry if this is a stupid question, but you used the Mac version of PyUSB, right? http://fpgamining.com/software/PyUSB-1.6-fizzisist-mac.tar.gz

Other than that, I'm really not sure what could be wrong. It might be helpful to go through and clear out all of the old D2XX stuff in case something is hanging around. First, do the reverse of the installation process you followed in the FTDI D2XX readme file. Then, remove PyUSB by removing the d2xx folder in /Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/. Maybe something is lingering around from one of the other versions and mucking things up.

fizzisist (OP)
Hero Member
*****
Offline Offline

Activity: 720
Merit: 525



View Profile WWW
January 05, 2012, 09:49:29 PM
 #311

FPGAMiner pushed a new project to the Open Source FPGA Bitcoin Miner Github repository called X5000_ztexmerge. This is the new firmware capable of achieving much higher hashrates because he ported over ztex's very fast hashers to our platform. Thanks for open sourcing that, ztex!

I'll be posting some pre-built bitstreams very soon, but everyone should be even more careful to monitor their cooling using these higher clocks because the potential to burn out your FPGA is much higher. Also, in my limited tests I've seen that the error rates vary from FPGA to FPGA so you'll have to experiment with your setup to see what works well for you. I've seen some FPGAs tolerate 180 MHz no problem, but others report over 10% errors at that clock rate.

Qoheleth
Legendary
*
Offline Offline

Activity: 960
Merit: 1028


Spurn wild goose chases. Seek that which endures.


View Profile WWW
January 05, 2012, 10:23:44 PM
 #312

I'll be posting some pre-built bitstreams very soon, but everyone should be even more careful to monitor their cooling using these higher clocks because the potential to burn out your FPGA is much higher. Also, in my limited tests I've seen that the error rates vary from FPGA to FPGA so you'll have to experiment with your setup to see what works well for you. I've seen some FPGAs tolerate 180 MHz no problem, but others report over 10% errors at that clock rate.
In preparation for this, I'd like to reiterate my request for instructions on removing the stock heatsinks. I have a feeling my current cooling situation might be insufficient for running the new logic on my boards (especially the slightly anemic one we've discussed before), so I'd like to be ready to switch to something a little higher-end if needs be.

If there is something that will make Bitcoin succeed, it is growth of utility - greater quantity and variety of goods and services offered for BTC. If there is something that will make Bitcoin fail, it is the prevalence of users convinced that BTC is a magic box that will turn them into millionaires, and of the con-artists who have followed them here to devour them.
fizzisist (OP)
Hero Member
*****
Offline Offline

Activity: 720
Merit: 525



View Profile WWW
January 05, 2012, 10:41:05 PM
 #313

I'll be posting some pre-built bitstreams very soon, but everyone should be even more careful to monitor their cooling using these higher clocks because the potential to burn out your FPGA is much higher. Also, in my limited tests I've seen that the error rates vary from FPGA to FPGA so you'll have to experiment with your setup to see what works well for you. I've seen some FPGAs tolerate 180 MHz no problem, but others report over 10% errors at that clock rate.
In preparation for this, I'd like to reiterate my request for instructions on removing the stock heatsinks. I have a feeling my current cooling situation might be insufficient for running the new logic on my boards (especially the slightly anemic one we've discussed before), so I'd like to be ready to switch to something a little higher-end if needs be.

Thanks for the reminder and sorry for not testing this out already! I'll do some experiments tonight!

freshzive
Sr. Member
****
Offline Offline

Activity: 447
Merit: 250


View Profile
January 05, 2012, 11:22:25 PM
 #314

Running Python 2.6.7, using the D2XX 1.04 drivers and the PyUSB-0.6 from fpamining.com . I've tried messing around with different versions but nothing I do will get it to work in Lion. Works beautifully in Win7 though.

Any other suggestions? Thanks for the help.

Sorry if this is a stupid question, but you used the Mac version of PyUSB, right? http://fpgamining.com/software/PyUSB-1.6-fizzisist-mac.tar.gz

Other than that, I'm really not sure what could be wrong. It might be helpful to go through and clear out all of the old D2XX stuff in case something is hanging around. First, do the reverse of the installation process you followed in the FTDI D2XX readme file. Then, remove PyUSB by removing the d2xx folder in /Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/. Maybe something is lingering around from one of the other versions and mucking things up.

Yes, I've been using the Mac version you posted. I was able to replicate the exact error I posted above on a 2nd Mac, so it isn't something weird with my Macbook. I'll try going through and manually removing all the old D2XX stuff tonight, didn't remove anything from the site-packages folder when reinstalled PyUSB last time (I'm assuming it needs to be reinstalled with each D2XX change?).


On another note, are there suggestions for something to mount these FPGAs on? I just ordered another and plan on picking up a few more once the speed improvements hit. I was thinking some sort of small case with some powerful fans in it would be sweet.

Thanks again for all of the help/advice!

coblee
Donator
Legendary
*
Offline Offline

Activity: 1651
Merit: 1284


Creator of Litecoin. Cryptocurrency enthusiast.


View Profile
January 07, 2012, 09:05:49 AM
 #315

Looks like there's a new 166 mhash/s bitstream: http://fpgamining.com/bitstreams/
I assume it's legit?

tf101
Newbie
*
Offline Offline

Activity: 18
Merit: 0


View Profile
January 07, 2012, 09:17:53 AM
 #316

-------------
Device: 0
Serial: AH00WOWE
JTAG chain: 2
Number of FPGAs: 2
Running time: 59m39s
Getwork interval: 20 secs
Chain 0:
  Accepted: 133
  Rejected: 3 (2.21%)
  Invalid: 0 (0.00%)
  Accepted hashrate: 159.57 MH/s
  Hashrate w/ rejects: 163.16 MH/s
  Hashrate w/ invalids: 163.16 MH/s
Chain 1:
  Accepted: 103
  Rejected: 2 (1.90%)
  Invalid: 2 (1.87%)
  Accepted hashrate: 123.57 MH/s
  Hashrate w/ rejects: 125.97 MH/s
  Hashrate w/ invalids: 128.37 MH/s
Total hashrate for device: 283.14 MH/s / 289.14 MH/s / 291.54 MH/s

Thats using the 166MHz bitstream.
freshzive
Sr. Member
****
Offline Offline

Activity: 447
Merit: 250


View Profile
January 07, 2012, 04:40:59 PM
 #317

Is it getting a lot hotter with the 166mhz bitstream?

freshzive
Sr. Member
****
Offline Offline

Activity: 447
Merit: 250


View Profile
January 07, 2012, 05:06:15 PM
 #318

By the way, I did end up getting this run in Mac OS X Lion. The PyUSB-1.6 package you posted refused to build/install with most of the Python 2.6.x versions. The only one I could get to work with was the ActiveState Python 2.6.7, but then I'd get that error when running program.py ^^. However, this morning I installed ActiveState Python 2.7.2, rebuilt/installed the PyUSB-1.6 package and everything is working perfectly.

So yeah, if you're using a Mac, try the Python 2.7.2 package from the activestate site.

fizzisist (OP)
Hero Member
*****
Offline Offline

Activity: 720
Merit: 525



View Profile WWW
January 07, 2012, 05:48:25 PM
 #319

In preparation for this, I'd like to reiterate my request for instructions on removing the stock heatsinks. I have a feeling my current cooling situation might be insufficient for running the new logic on my boards (especially the slightly anemic one we've discussed before), so I'd like to be ready to switch to something a little higher-end if needs be.

It looks to me like removing the heatsinks when hot works really well. At least it did for me. Fortunately, there's a very nice little heater right under the heatsink. Smiley I simply ran the board for a minute without a fan, then turned it off and unplugged it. Then, the slightest twisting motion broke the heatsink free from the chip. When I say twist, I mean parallel to the board. And when I say slight, I mean slight. If you're using a lot of force, you risk pulling the FPGA off the board.

On another note, are there suggestions for something to mount these FPGAs on? I just ordered another and plan on picking up a few more once the speed improvements hit. I was thinking some sort of small case with some powerful fans in it would be sweet.

I've been thinking a lot about enclosures lately. I've drawn up a few sketches of different ideas for custom enclosures, but haven't decided on what I think would be the best.

I plan to do some experiments with one idea today, which would be simpler than a custom case if it works. I ordered a standard ATX mid-tower case and I'm going to install a mini-itx motherboard and PSU in there. Then, I'll try to remove the hard drive and optical drive mounts from the front of the case. I'll do my best to fill up that empty space with as many X6500s and fans as I can. The front of the case is mesh, so fresh air will be pulled in from there and exhausted out the back or top of the case.

Looks like there's a new 166 mhash/s bitstream: http://fpgamining.com/bitstreams/
I assume it's legit?

Yes, it's legit. Smiley In my testing, it seems to use about a watt more than the 133 MHz (17 instead of 16), which means cooling is even more important. Keep an eye on it. I'll be posting some higher hashrates soon, but those come along with even more of a warning.

By the way, I did end up getting this run in Mac OS X Lion. The PyUSB-1.6 package you posted refused to build/install with most of the Python 2.6.x versions. The only one I could get to work with was the ActiveState Python 2.6.7, but then I'd get that error when running program.py ^^. However, this morning I installed ActiveState Python 2.7.2, rebuilt/installed the PyUSB-1.6 package and everything is working perfectly.

So yeah, if you're using a Mac, try the Python 2.7.2 package from the activestate site.

That's great! Glad to hear you got it working, and I'll update the guide to suggest that version of Python. I simply used the pre-installed version of Python, but I'm not even sure what version that was.

sadpandatech
Hero Member
*****
Offline Offline

Activity: 504
Merit: 500



View Profile
January 07, 2012, 06:57:26 PM
 #320

I like the progress I'm seeing here. keep it up!  Smiley


cheers

If you're not excited by the idea of being an early adopter 'now', then you should come back in three or four years and either tell us "Told you it'd never work!" or join what should, by then, be a much more stable and easier-to-use system.
- GA

It is being worked on by smart people.  -DamienBlack
Pages: « 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 [16] 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 »
  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!