Pages:
Author

Topic: X6500 Custom FPGA Miner - page 35. (Read 220050 times)

newbie
Activity: 12
Merit: 0
January 23, 2012, 11:50:04 AM
Hey guys,

Been following bitcoin for a year or so, speculating in it for a few months, considering mining for a few weeks and following the thread a few days. I'm pretty caught up on where everythings at and I'm looking to join the project.

Consider this my introduction and sincere thanks to Fizzisist and the other project leaders.

Looking to make a purchase on a board in the next week or so once my wall wort arrives.

Plan on running the board from my mac mini initially.

This is my first dive into mining but I think FPGA boards and the x6500 are an opportune entry point.

Thanks again and I look forward to seeing what we can build.

JR
full member
Activity: 148
Merit: 100
January 21, 2012, 12:30:29 PM
anyone tried to make the miner-software tor-useable?
why? my pool has an .onion adress which should be avaible in case of ddos-attacks
i know tsocks should run on linux, but i am on windows, a open-source or freeware alternative would be nice but i didn't find something yet
as far as i get it i have to rewrite rpcclient.py
sr. member
Activity: 378
Merit: 250
Why is it so damn hot in here?
January 21, 2012, 12:21:15 PM
In addition to cooler mounting holes, I would like to see the power supply come from the 12V pin as opposed to the 5V pin as the standard configuration.  The reason being, most power supplies manufactured now are able to supply far more power down the 12v rail than the 5v rail.  While this may not be a big deal for most end users running just a few of these, I'm debating on buying (If I can get the funding) 20 or more and having to switch them all myself from 5V to 12V doesn't sound like happy fun time.

Yep, this is planned for the next revision as well. But, switching them over isn't as hard as it sounds. I use one adapter modified like so and then a Molex splitter like the one Cablesaurus sells to connect to the stack of 5 boards. Works great for me!

I'm wondering, do you folks have any recommendations for heatsinks we should allow for (in terms of clearance and mounting hole locations)? If you have a heatsink at hand and are willing to measure it, we can try to make the next board compatible with it. Unfortunately, most of these Northbridge heatsinks don't offer CAD drawings...

This...
"In doing this, you are bringing the 12V wire over to the pin where 5V is usually applied. If you forget and plug this adapter into some other device, you will probably blow it up."   Grin

hero member
Activity: 720
Merit: 525
January 21, 2012, 12:18:26 PM
In addition to cooler mounting holes, I would like to see the power supply come from the 12V pin as opposed to the 5V pin as the standard configuration.  The reason being, most power supplies manufactured now are able to supply far more power down the 12v rail than the 5v rail.  While this may not be a big deal for most end users running just a few of these, I'm debating on buying (If I can get the funding) 20 or more and having to switch them all myself from 5V to 12V doesn't sound like happy fun time.

Yep, this is planned for the next revision as well. But, switching them over isn't as hard as it sounds. I use one adapter modified like so and then a Molex splitter like the one Cablesaurus sells to connect to the stack of 5 boards. Works great for me!

I'm wondering, do you folks have any recommendations for heatsinks we should allow for (in terms of clearance and mounting hole locations)? If you have a heatsink at hand and are willing to measure it, we can try to make the next board compatible with it. Unfortunately, most of these Northbridge heatsinks don't offer CAD drawings...
hero member
Activity: 720
Merit: 525
January 21, 2012, 12:09:09 PM
Hi guys! Ive been away (as you might have noticed) for the past couple of months (learning to fly gyrocopters FWIW) and so since I´ve returned I have finalized my order for the four Dominator FGPA boards. Happy to say they have just arrived and don´t those little munchins look beauuuuuuwtiful? 

I´m on a Mandriva linux box with a 32bit Intel BTW. So far I´ve installed the FTDI drivers. I then downloaded PyUSB-1.6-fizzisist-linux.tar.gz extracted it and attempted to install as per the instructions. Here´s where I have gotten a little stumped:

Code:
[root@localhost PyUSB-1.6-fizzisist-linux]# python setup.py build
running build
running build_py
warning: build_py: byte-compiling is disabled, skipping.

running build_ext
building 'd2xx._d2xx' extension
gcc -pthread -fno-strict-aliasing -O2 -g -frecord-gcc-switches -Wstrict-aliasing=2 -pipe -Wformat -Werror=format-security -Wp,-D_FORTIFY_SOURCE=2 -fstack-protector --param=ssp-buffer-size=4 -fomit-frame-pointer -mtune=generic -march=i586 -fasynchronous-unwind-tables -DNDEBUG -O2 -g -frecord-gcc-switches -Wstrict-aliasing=2 -pipe -Wformat -Werror=format-security -Wp,-D_FORTIFY_SOURCE=2 -fstack-protector --param=ssp-buffer-size=4 -fomit-frame-pointer -mtune=generic -march=i586 -fasynchronous-unwind-tables -g -fPIC -Iftdi-win32 -I/usr/include/python2.7 -c d2xx/_d2xx.c -o build/temp.linux-i686-2.7/d2xx/_d2xx.o
In file included from d2xx/_d2xx.c:23:0:
d2xx/_d2xx.h:23:20: fatal error: Python.h: No such file or directory
compilation terminated.
error: command 'gcc' failed with exit status 1
[root@localhost PyUSB-1.6-fizzisist-linux]#

I never have any joy when expected to compile code in order to install it... NEVER.  Sad I figured there was something missing between Python and gcc, so I installed a package (gcc-plugin-python), but this had no effect. Any ideas?  Huh


Welcome back!!

I think that's an easy problem to fix, and I saw it when I installed on Linuxcoin a while back. You need the developer version of python:
apt-get install python-dev (or whatever Mandriva uses for packages)

Hopefully that does it.
sr. member
Activity: 447
Merit: 250
January 21, 2012, 11:19:06 AM
Did anyone ever figure out what's causing this error?

Code:
2012-01-21 08:13:56 | Long-poll: ValueError! No JSON object could be decoded

Still getting this on ABCpool...
sr. member
Activity: 378
Merit: 250
Why is it so damn hot in here?
January 21, 2012, 11:04:30 AM
In addition to cooler mounting holes, I would like to see the power supply come from the 12V pin as opposed to the 5V pin as the standard configuration.  The reason being, most power supplies manufactured now are able to supply far more power down the 12v rail than the 5v rail.  While this may not be a big deal for most end users running just a few of these, I'm debating on buying (If I can get the funding) 20 or more and having to switch them all myself from 5V to 12V doesn't sound like happy fun time.
sr. member
Activity: 447
Merit: 250
January 21, 2012, 10:53:25 AM
I never have any joy when expected to compile code in order to install it... NEVER.  Sad I figured there was something missing between Python and gcc, so I installed a package (gcc-plugin-python), but this had no effect. Any ideas?  Huh

What version of python are you running? Try installing 2.6.7 and using that directly i.e.: python2.6 setup.py build

full member
Activity: 157
Merit: 100
January 21, 2012, 08:34:13 AM
Code:
340.93 MH/s | 0: 22569/123/121 0.5%/0.5% | 1: 22708/161/125 0.7%/0.5% | 1w29m56s | AH00WOWI

Anyone (besides fizzisist) successfully running the 180Mhz bitstream? Care to comment on your speed and what cooling you're using?

Given how stable the 166Mhz bitstream has been for me, not sure if the upgrade is worth it.

I'm running a 190Mhash bitstream per FPGA. Which gives me 380 total. My heatsink isn't anything special, however it's attached on with thermal epoxy instead of tape and has a high performance fan blowing on them.

I see alot of posts requesting for mounting holes, rest assured they will come in future board revisions. I'm thinking of working out some cheap hardware that'll enable users to clamp various heatsinks onto the current board, until then thermal epoxy might be the next best solution.
full member
Activity: 185
Merit: 121
January 21, 2012, 06:25:27 AM
With regards to the cooling issues, if you are willing to forgo the board stacking configuration, itś not too hard to conceive of a set of risers coming up to just above the heat-sinks to a pair of flat aluminum strips (mounting rails), counter-sunk and drilled to accept an 80mm fan facing downwards over the whole board. This config, fits with the board quite well and puts the airflow right down into the heatsinks. I can try to knock up a prototype for this and then some jigs if it works out OK.

Stacking for vertical space could still be implemented if the mounting rails were extended out past the board and separate holes were provided for longer ´stacking risers´.

For extra cooling, how about a sponge (cheap kitchen sponge should do) cut so that it slips down over the heat-sink fins. That can then be fed water by capillary action, with a wick going to a water bottle mounted nearby? Then you effectively have an evaporative cooling system. 

Given the lack of header pins for a fan, you can also use the disused Mollinex plug for the fan, if you are using the barrel connector for power (or vice versa I assume).

Itś great to see everybody doing so well with the Dominator so far. The code seems to be maturing nicely and the hash rates... OH WOW!!  Shocked
full member
Activity: 185
Merit: 121
January 21, 2012, 03:48:37 AM
Hi guys! Ive been away (as you might have noticed) for the past couple of months (learning to fly gyrocopters FWIW) and so since I´ve returned I have finalized my order for the four Dominator FGPA boards. Happy to say they have just arrived and don´t those little munchins look beauuuuuuwtiful? 

I´m on a Mandriva linux box with a 32bit Intel BTW. So far I´ve installed the FTDI drivers. I then downloaded PyUSB-1.6-fizzisist-linux.tar.gz extracted it and attempted to install as per the instructions. Here´s where I have gotten a little stumped:

Code:
[root@localhost PyUSB-1.6-fizzisist-linux]# python setup.py build
running build
running build_py
warning: build_py: byte-compiling is disabled, skipping.

running build_ext
building 'd2xx._d2xx' extension
gcc -pthread -fno-strict-aliasing -O2 -g -frecord-gcc-switches -Wstrict-aliasing=2 -pipe -Wformat -Werror=format-security -Wp,-D_FORTIFY_SOURCE=2 -fstack-protector --param=ssp-buffer-size=4 -fomit-frame-pointer -mtune=generic -march=i586 -fasynchronous-unwind-tables -DNDEBUG -O2 -g -frecord-gcc-switches -Wstrict-aliasing=2 -pipe -Wformat -Werror=format-security -Wp,-D_FORTIFY_SOURCE=2 -fstack-protector --param=ssp-buffer-size=4 -fomit-frame-pointer -mtune=generic -march=i586 -fasynchronous-unwind-tables -g -fPIC -Iftdi-win32 -I/usr/include/python2.7 -c d2xx/_d2xx.c -o build/temp.linux-i686-2.7/d2xx/_d2xx.o
In file included from d2xx/_d2xx.c:23:0:
d2xx/_d2xx.h:23:20: fatal error: Python.h: No such file or directory
compilation terminated.
error: command 'gcc' failed with exit status 1
[root@localhost PyUSB-1.6-fizzisist-linux]#

I never have any joy when expected to compile code in order to install it... NEVER.  Sad I figured there was something missing between Python and gcc, so I installed a package (gcc-plugin-python), but this had no effect. Any ideas?  Huh
full member
Activity: 148
Merit: 100
January 20, 2012, 01:31:57 AM
the real WC problem is that if you really have a cluster you need a lot of material

i think the real goal hear is to get some mounting-holes on future boards

looks like northbridge-cooling-stuff could be the right solution, but is there a standard of those holes?
sr. member
Activity: 274
Merit: 250
January 19, 2012, 11:39:57 PM
AC vs WC:
1. Ac has beter cooling of whole PCB and WC cools only Spartans.
2. When you wont to cool cluster, you need lots of those blocks AND some fans on the radiator, where in case of AC, you need just somme fans.

In my opinion, in case of FPGA AC wins. Why??
1. There`s not a lot of heat to cool.
2. Power Section is as hot as spartan is, but it`s harder to watercool.
3. making of fullcover is pointless - too small market.

Best sollution:
Put your cluster into nice box, and blow few fans into it Wink

@fizzisist
I think you can use only 2 fans, when you put more walls. Now you got lots of "fresh/cool" air around, but in stad you got 6 fans.
sr. member
Activity: 447
Merit: 250
January 19, 2012, 11:11:54 PM
Code:
340.93 MH/s | 0: 22569/123/121 0.5%/0.5% | 1: 22708/161/125 0.7%/0.5% | 1w29m56s | AH00WOWI

Anyone (besides fizzisist) successfully running the 180Mhz bitstream? Care to comment on your speed and what cooling you're using?

Given how stable the 166Mhz bitstream has been for me, not sure if the upgrade is worth it.
full member
Activity: 411
Merit: 101
🦜| Save Smart & Win 🦜
January 19, 2012, 10:32:18 PM
I second the idea of having mounting holes for cooling devices on the next version of this board. There are some water blocks for chipsets such as this one that would work wonderfully I think.
hero member
Activity: 720
Merit: 525
January 19, 2012, 06:53:56 PM
My board is back!  Tongue
I changed to 12V connector and it works. On the 5 V molex which I used before no current gets pulled at all. Seems you are right, the fuse is gone. Don't know how this could have happend. I am using a standard PC power supply....
Maybee under 5V voltage drop? The current gets higher if the voltage drops and fuses are likely to burn?

Can you give me the position and data of the fuse? I can relace it by myself - less hassle comparing to sending the board back and forward through the customs...
Is it the part with the "w" on it?

Awesome! Yep, the fuse is the part with the W on it. You'll notice that there's one on each input. I think you're right that the voltage probably dropped and the current rose to keep up. The fuse is Bourns SF-1206S400-2 available from Digikey here for $0.68. If your power supply is regularly putting out a slightly low voltage, you might want to consider installing the 5A part or switch from 5V to 12V on the Molex connector.

Again, I'm sorry this happened but I'm glad to hear it's a pretty easy fix!
newbie
Activity: 16
Merit: 0
January 19, 2012, 03:27:37 PM
Hi,

My board is back!  Tongue
I changed to 12V connector and it works. On the 5 V molex which I used before no current gets pulled at all. Seems you are right, the fuse is gone. Don't know how this could have happend. I am using a standard PC power supply....
Maybee under 5V voltage drop? The current gets higher if the voltage drops and fuses are likely to burn?

Can you give me the position and data of the fuse? I can relace it by myself - less hassle comparing to sending the board back and forward through the customs...
Is it the part with the "w" on it?

It sounds to me like you had a good system going, and I doubt your FPGAs are dead. Do you have a volt-meter that you could measure the output from your power supply with? If that looks good, it could be that the protection fuse on X6500 has blown. This fuse should only blow as a last resort, particularly if you supplied reverse voltage, so I don't know why it would blow in your case. Were you using the Molex connector or barrel connector? Each power input has it's own fuse so you could try using the other one (if you have a suitable supply) to figure out if the fuse is to blame. If it is, please send it back and I'll replace the fuse for free.

If you don't feel comfortable testing it, I'll gladly diagnose it for you if you send it back. If you're pretty handy, you could replace the fuse yourself or bypass it entirely (although you'll lose the protection it provided, of course). I'll email you the address to ship it to.

Sorry for the trouble!

hero member
Activity: 720
Merit: 525
January 19, 2012, 01:11:33 PM
I had two 40mm fans above each heatsink, the did not get hot, therefore I don't believe they are burnt.
Maybee something about the power supply went wrong?
Can you gibe me some instruction what to do? Can I check the Voltage regulators?

It sounds to me like you had a good system going, and I doubt your FPGAs are dead. Do you have a volt-meter that you could measure the output from your power supply with? If that looks good, it could be that the protection fuse on X6500 has blown. This fuse should only blow as a last resort, particularly if you supplied reverse voltage, so I don't know why it would blow in your case. Were you using the Molex connector or barrel connector? Each power input has it's own fuse so you could try using the other one (if you have a suitable supply) to figure out if the fuse is to blame. If it is, please send it back and I'll replace the fuse for free.

If you don't feel comfortable testing it, I'll gladly diagnose it for you if you send it back. If you're pretty handy, you could replace the fuse yourself or bypass it entirely (although you'll lose the protection it provided, of course). I'll email you the address to ship it to.

Sorry for the trouble!
hero member
Activity: 720
Merit: 525
January 19, 2012, 01:04:30 PM
The reason that we leave it up to the customer to decide how to cool the board and what is sufficient is because, to be honest, we don't know! Also, what works for us might not work for you and I'm sure many of you will find better ways to do it. That said, at the very least I should share the cooling set up that I use, because it works for me and I run 180 MHz boards with it by default. Sorry for not doing that earlier. Let me be clear though, that this is simply what works for me. Because there are so many variables, I'm not endorsing this as the perfect setup that everyone should use. It works well enough for me, is all, and it will hopefully guide you in building your own system.

I have 80 mm fans blowing sideways as in the User Guide. Currently, I'm running 10 x6500s in 2 stacks of 5. To get that height, I have two 80 mm fans stacked one on top of the other and zip-tied together, as in:



Then, using 3 of those towers, I built this:



All of the fans are oriented so that they're blowing air to the right, and they're stuck down with double sided tape. I'm not totally sure the last tower of fans is necessary, but it seemed nice to have it.

Finally, and this may be overkill, I taped on some pieces of card stock to help channel the air flow a little bit more:



I fully plan to build a real enclosure that basically duplicates this and replaces the wonky cardboard walls with something a little more permanent.

All of those boards are running the 180 MHz bitstream and all are getting ~360 MH/s with very few invalids (<1%). When I touch a PCB next to the FPGA with my finger, it's almost imperceptibly warm.
member
Activity: 73
Merit: 10
January 19, 2012, 09:42:12 AM
I had two 40mm fans above each heatsink, the did not get hot, therefore I don't believe they are burnt.
Maybee something about the power supply went wrong?

That's an interesting thought. What's your power supply? What's its max amperage? Do you have a spare you can swap it with?
Pages:
Jump to: