X6500 Custom FPGA Miner - page 45. | Bitcointalksearch.org

fizzisist

hero member

Activity: 720

Merit: 528

Quote from: sirky on December 27, 2011, 07:11:28 PM

Quote from: sirky on December 26, 2011, 02:16:02 PM

Quote from: fizzisist on December 24, 2011, 11:58:41 AM

That error is probably because of that "sparkline" stuff in ConsoleLogger.py that I once thought was a great idea. I think that character is the largest one on the list of available characters, and I just removed it from the latest version in Github. Could you try it now to see if the problem is fixed for you? If it isn't we can completely comment out that section of the code and be done with it (unless you really want that sparkline in there).

I just had a crash with \u2587 this time. It took a bit longer to get to than normal.

I manually changed the python file, and replaced the fancy ascii characters with numerals between 0 and 6. It isn't as beautiful, but it doesn't crash the system, and the same info is conveyed.

Ah, nice work! Sorry, I guess those characters don't work on Windows XP although they seemed to work fine on Windows 7 when I tested it. I think I'll just remove that whole feature from the code for now, until I figure out what needs to be done to make it totally cross platform.

My focus this week will be to get to higher hashrates and implement long-polling, but please everyone let me know if you come across any more bugs or have any ideas for new features. Thanks!

sirky

sr. member

Activity: 404

Merit: 250

Quote from: sirky on December 26, 2011, 02:16:02 PM

Quote from: fizzisist on December 24, 2011, 11:58:41 AM

Quote from: sirky on December 24, 2011, 11:27:41 AM

I am getting an error on my old XP box that says it cannot encode character \u2588

Anyone else get this.

And sorry if I missed something.

After it errors it seems like the FPGA gets a bit hotter too. It runs cool when it is running though.

That error is probably because of that "sparkline" stuff in ConsoleLogger.py that I once thought was a great idea. I think that character is the largest one on the list of available characters, and I just removed it from the latest version in Github. Could you try it now to see if the problem is fixed for you? If it isn't we can completely comment out that section of the code and be done with it (unless you really want that sparkline in there).

I just had a crash with \u2587 this time. It took a bit longer to get to than normal.

I manually changed the python file, and replaced the fancy ascii characters with numerals between 0 and 6. It isn't as beautiful, but it doesn't crash the system, and the same info is conveyed.

sirky

sr. member

Activity: 404

Merit: 250

Quote from: fizzisist on December 24, 2011, 11:58:41 AM

Quote from: sirky on December 24, 2011, 11:27:41 AM

I am getting an error on my old XP box that says it cannot encode character \u2588

Anyone else get this.

And sorry if I missed something.

After it errors it seems like the FPGA gets a bit hotter too. It runs cool when it is running though.

That error is probably because of that "sparkline" stuff in ConsoleLogger.py that I once thought was a great idea. I think that character is the largest one on the list of available characters, and I just removed it from the latest version in Github. Could you try it now to see if the problem is fixed for you? If it isn't we can completely comment out that section of the code and be done with it (unless you really want that sparkline in there).

I just had a crash with \u2587 this time. It took a bit longer to get to than normal.

sirky

sr. member

Activity: 404

Merit: 250

Quote from: fizzisist on December 24, 2011, 11:58:41 AM

Quote from: sirky on December 24, 2011, 11:27:41 AM

I am getting an error on my old XP box that says it cannot encode character \u2588

Anyone else get this.

And sorry if I missed something.

After it errors it seems like the FPGA gets a bit hotter too. It runs cool when it is running though.

That error is probably because of that "sparkline" stuff in ConsoleLogger.py that I once thought was a great idea. I think that character is the largest one on the list of available characters, and I just removed it from the latest version in Github. Could you try it now to see if the problem is fixed for you? If it isn't we can completely comment out that section of the code and be done with it (unless you really want that sparkline in there).

I am heading out for a few days for XMas, and don't want to test when I am not home. So I will check it out when I get back on Monday.

Thanks for the quick response!

fizzisist

hero member

Activity: 720

Merit: 528

Quote from: Qoheleth on December 24, 2011, 11:36:40 AM

228.66 MH/s | 0: 2238/210/0 8.6% | 1: 2309/227/0 9.0% | 1d21m22s

Running the 133MHz bitstreams, not sure exactly why the rate is so low. Maybe just a run of bad luck.

Glad to see it running for an extended period now! The hashrate you see is based on the number of shares submitted in the last hour, so it can fluctuate. If I look at the total number of shares submitted the whole time it's been running, you can see that you're still a little bit lower than typical:

Shares submitted: 2238+210+2309+227 = 4984
Running time: 1d21m22s = (24*60 + 21) * 60 + 22 = 87682 s
Effective hashrate = Shares * 2^32 / Time = 244.1 MH/s

That's lower than the lowest performance I saw during testing. Could you PM me your serial number so I can look up the performance I saw with this board? That will tell us if it's an issue with the board or some difference in how it runs on your system.

Thanks!

fizzisist

hero member

Activity: 720

Merit: 528

Quote from: sirky on December 24, 2011, 11:27:41 AM

I am getting an error on my old XP box that says it cannot encode character \u2588

Anyone else get this.

And sorry if I missed something.

After it errors it seems like the FPGA gets a bit hotter too. It runs cool when it is running though.

That error is probably because of that "sparkline" stuff in ConsoleLogger.py that I once thought was a great idea. I think that character is the largest one on the list of available characters, and I just removed it from the latest version in Github. Could you try it now to see if the problem is fixed for you? If it isn't we can completely comment out that section of the code and be done with it (unless you really want that sparkline in there).

Qoheleth

legendary

Activity: 960

Merit: 1028

Spurn wild goose chases. Seek that which endures.

228.66 MH/s | 0: 2238/210/0 8.6% | 1: 2309/227/0 9.0% | 1d21m22s

Running the 133MHz bitstreams, not sure exactly why the rate is so low. Maybe just a run of bad luck.

sirky

sr. member

Activity: 404

Merit: 250

I am getting an error on my old XP box that says it cannot encode character \u2588

Anyone else get this.

And sorry if I missed something.

After it errors it seems like the FPGA gets a bit hotter too. It runs cool when it is running though.

shad

full member

Activity: 148

Merit: 100

269.42 MH/s | 0: 3089/339/0 9.9% | 1: 2993/311/0 9.4% | 1d9h6m and sill running Wink

stales rate is because of no LP, but i think LP is on the way

CA Coins

donator

Activity: 305

Merit: 250

Following to see how the software issues work out and what hash rates people are getting

fizzisist

hero member

Activity: 720

Merit: 528

Quote from: thirdlight on December 23, 2011, 04:46:46 PM

Quote from: fizzisist on December 23, 2011, 12:11:59 PM

behind a proxy, etc.?

Yes, I use "Flexible Mining Proxy". Going direct does get rid of those errors.

But.. it won't connect to BTC Guild (which does work thru the proxy). Are there "good" pools & "bad" pools for this software? My experience is:

Eclipse: works
Guild: fails to connect
p2pool: shares are "invalid"

Maybe the LP support will fix it?

Hmm, I use the mining proxy regularly and never saw that error... I did have a problem early on where I would get empty responses from the proxy occasionally. I had to increase the KeepAliveTimeout in Apache's config to avoid this. Still, it should just retry when it encounters that instead of getting an error. Try increasing that setting to longer than the getwork interval (default 20 seconds) and see if it gets better.

As for the various pools, I haven't tested it with many others. The few I tried worked perfectly, but I need to do a careful study of a lot of them, along with recording the raw traffic to know what's really going on. Thanks for reporting what you've found!

fizzisist

hero member

Activity: 720

Merit: 528

I decided to put up a list of useful accessories for mining with your X6500, in hopes that it will save others a lot of time in searching around the web for these things.

Cablesaurus.com
As long as you're ordering your X6500s from Cablesaurus, you might as well pick up some useful accessories there and not have to pay shipping to another supplier. We're trying to get as many accessories stocked by Cablesaurus.com as possible, but only have a few basics right now.

10 port USB hub:
http://cablesaurus.com/index.php?main_page=product_info&cPath=10&products_id=55

USB cable:
http://cablesaurus.com/index.php?main_page=product_info&cPath=10&products_id=54

Molex splitter:
http://cablesaurus.com/index.php?main_page=product_info&cPath=10&products_id=51

24-Pin ATX Power-On Dummy Plug:
http://cablesaurus.com/index.php?main_page=product_info&cPath=13&products_id=45

20-Pin ATX Power-On Dummy Plug:
http://cablesaurus.com/index.php?main_page=product_info&cPath=13&products_id=44

Elsewhere on the web:
Unfortunately, not everything that you might want is available from Cablesaurus.com yet. Here are some accessories that might be useful.

USB powered fan:
http://www.amazon.com/Thermaltake-Mobile-Fan-External-Cooling/dp/B00080G0BK

Fan with AC adapter (80mm):
http://www.buyextras.com/bxtulsi80x80.html

Fan with AC adapter (120mm):
http://www.buyextras.com/yalo120x25ul2.html

AC to Molex power supply (the current on the 5V rail is not enough to power an X6500, so you must use an adapter to bring the 12V line to the correct pin):
http://www.amazon.com/110v-4pin-molex-Power-Adapter/dp/B000MGG6SC

SATA to Molex adapter:
http://www.amazon.com/KingWin-Power-Adapter-Cable-SAC-05/dp/B001JE6WYY

Low-profile SATA to Molex:
http://www.amazon.com/Sata-LP4-Pin-Power-Adapter/dp/B000NPKGGU

I'll add to this list as I find more useful stuff.

thirdlight

sr. member

Activity: 445

Merit: 250

Quote from: fizzisist on December 23, 2011, 12:11:59 PM

behind a proxy, etc.?

Yes, I use "Flexible Mining Proxy". Going direct does get rid of those errors.

But.. it won't connect to BTC Guild (which does work thru the proxy). Are there "good" pools & "bad" pools for this software? My experience is:

Eclipse: works
Guild: fails to connect
p2pool: shares are "invalid"

Maybe the LP support will fix it?

fizzisist

hero member

Activity: 720

Merit: 528

Quote from: thirdlight on December 22, 2011, 04:31:32 PM

Seeing frequent Errno 10053, and very occasional Errno 10054.

Not sure what's going on there... I'm guessing those are socket errors, which would show up as "IOError" in the log. Is that what you see? Do you have some strange network config (behind a proxy, etc.)? I've never seen those errors myself.

Quote from: fizzisist on December 12, 2011, 02:16:24 PM

Quote from: coblee on December 12, 2011, 01:59:54 PM

What's the status on the Mac miner?

So far not tested yet. I was putting a little more focus on the ARM miner, but haven't had success yet. I'll try Mac soon.

I got it to run on my Mac today! I posted some instructions at fpgamining.com/software, but your testing and feedback on that will be much appreciated!

thirdlight

sr. member

Activity: 445

Merit: 250

Now using v 0.11, thanks!

Seeing frequent Errno 10053, and very occasional Errno 10054.

Qoheleth

legendary

Activity: 960

Merit: 1028

Spurn wild goose chases. Seek that which endures.

Quote from: fizzisist on December 22, 2011, 12:09:11 PM

Quote from: Qoheleth on December 22, 2011, 06:14:17 AM

New problem.

I've got both my X6500 boards up and "running", but while one of them (-d 1) is returning golden nonces every few jobs, the other one has been running for ~10 minutes without a single golden nonce.

Is this expected? What could the problem be?

Edit: Had a problem with my jury-rigged power supplies and had to reload -d 1, and when I came back it was spinning its wheels too.

We need to find a solution to this kind of problem. I believe what happened was the power was removed momentarily from the FPGA. When that happens, it loses it's configuration and goes back to being an unprogrammed FPGA. The mining software doesn't know that it's unprogrammed so just keeps asking it for nonces. Of course, none ever come. We'll work on a fix to this (easy to add a bit that says, "I'm programmed" or something).

Until then, if anyone gets strange problems like this, the first thing to try is restarting mine.py. The next thing is to rerun program.py. Rerunning program.py is essentially the same as rebooting the FPGA, so it should clear up any strangeness.

Hmm. I suppose that'd explain it - like I said, my solutions to powering the boards are pretty seriously jury-rigged right now, so I'm not surprised that they'd cut out momentarily. I'll see if I can't get that part of the system to be more stable when I get home tonight.

sadpandatech

hero member

Activity: 504

Merit: 500

Are there any plans to add a small eprom to future boards in order to save the configuration to the boards until we purposely wish to change it?

fizzisist

hero member

Activity: 720

Merit: 528

Quote from: Qoheleth on December 22, 2011, 06:14:17 AM

New problem.

I've got both my X6500 boards up and "running", but while one of them (-d 1) is returning golden nonces every few jobs, the other one has been running for ~10 minutes without a single golden nonce.

Is this expected? What could the problem be?

Edit: Had a problem with my jury-rigged power supplies and had to reload -d 1, and when I came back it was spinning its wheels too.

We need to find a solution to this kind of problem. I believe what happened was the power was removed momentarily from the FPGA. When that happens, it loses it's configuration and goes back to being an unprogrammed FPGA. The mining software doesn't know that it's unprogrammed so just keeps asking it for nonces. Of course, none ever come. We'll work on a fix to this (easy to add a bit that says, "I'm programmed" or something).

Until then, if anyone gets strange problems like this, the first thing to try is restarting mine.py. The next thing is to rerun program.py. Rerunning program.py is essentially the same as rebooting the FPGA, so it should clear up any strangeness.

fizzisist

hero member

Activity: 720

Merit: 528

Quote from: Qoheleth on December 22, 2011, 04:18:10 AM

I'm having some trouble as well.

Running program.py gives this error:

Code:

Traceback (most recent call last):
  File "program.py", line 23, in 
    from ft232r import FT232R, FT232R_PortList
  File "/home/[redacted]/x6500-miner/ft232r.py", line 22, in 
    import d2xx
  File "/usr/lib/python2.7/site-packages/d2xx/__init__.py", line 24, in 
    from _d2xx import *
ImportError: /usr/lib/python2.7/site-packages/d2xx/_d2xx.so: undefined symbol: FT_CyclePort

I ran a quick nm command on the libftd2xx.so I downloaded from the site you indicated, and sure enough, there's no such symbol in the library.

I'm on 32-bit Linux, if it matters.

Edit: Bar that; solution found. Turns out there's four functions whose metadata you commented out in the modified python libraries, but you forgot to comment out the functions themselves. So it tried to link to them, and.... bang!

Nice detective work! That's exactly the kind of modification I had to make to get it to work, and I'm not sure why you still had to make those changes. I'll look at my modified library again and update it so that this doesn't happen to others. Thanks, and sorry for the problems!

fizzisist

hero member

Activity: 720

Merit: 528

Quote from: thirdlight on December 22, 2011, 02:14:10 AM

Now getting:

Code:

(FPGA0) Golden nonce found
IO Error!
(FPGA0) rejected 72ba42c7L
(FGPA0) Connecting...
(FGPA0) accepted 72ba42c7L

Any thoughts?

I suspect that that IO Error was a connection error from the pool. Since it retried the submission after reconnecting, it shouldn't have counted that as rejected. I believe that behavior was fixed a few commits back in Github. Are you using v0.1? I decided to release a v0.11 today to try to bring everybody to the same page. This version and updated documentation can be found here. Sorry about all the trouble with that!

I was hoping to release after long-polling was added, but I think it's a good idea to get this out now, because of all these little fixes that have been added over the weeks. This version has been incredibly stable while I've been out of town for a week. I haven't once had to ssh in and restart a miner or anything like that. You can see the stability here: http://eligius.st/~artefact2/5/1FPGAxkfHskMq9fg2gbAWhPrnqwXydbX93

Also, this status line is pretty awesome:

Code:

273.35 MH/s | 0: 18082/1664/19 8.4% | 1: 18107/1665/19 8.4% | 1w23h39m | AHxxxxxx

Topic: X6500 Custom FPGA Miner - page 45. (Read 220107 times)