[DIY] - Reward $100 | Antminer S1/S3 Blade on Raspberry Pi

[zziggi]

I think it will work if you do not exceed 1V) I choose the cheapest aluminum as heatsink - about 8 $ 2m.  ROI:(
But low voltage can not raise your frequency - above 275 speed drops again probably because of HW errors.

Same here! Compiled under MinGW but will not detect the devices!

I tried to compile with a MinGW different versions of cgminer ( bitmain,kano,original),
all give an error in usbutils.c  when compiling with --enable-bmsc --enable-bitmain  --enable-icarus (--enable-icarus   it is necessary to detect the usb device).

One anormally though is that, using the precompiled version 3.8.5, I can not have two CP2102's in the same computer. I can hav one pluged into the PC / laptop, and another plugged into a hub connected to another usb port then it is ok. Also, I can not start with both plugged in, I have to either start without the modules pluged in, or just one module, else cgminer crashes!

Try first to run 4.7.0 when it detects all the devices as ANU,start 3.8.5.If detect as LIX press U and then reset device.4.7.0 can change the frequency with --anu-freq

[forpblat]

I got my heatsinks and UARTs in so I put together one blade to start hashing

I hooked up TX->RX and RX->TX. I tried with GND connected and disconnected.

However, with the bitmaintech cgminer 4.7 and compiled with "./configure --enable-icarus --enable-bmsc --enable-bitmain", I get a whole slew of

 [2014-11-04 19:30:15] ICA 0 usb transfer err:(-9) LIBUSB_ERROR_PIPE
 [2014-11-04 19:30:15] FAIL: USB get_lock not found (6:18)
 [2014-11-04 19:30:22] ICA 0 usb transfer err:(-9) LIBUSB_ERROR_PIPE
 [2014-11-04 19:30:22] FAIL: USB get_lock not found (6:19)
 [2014-11-04 19:30:34] ICA 0 usb transfer err:(-9) LIBUSB_ERROR_PIPE
 [2014-11-04 19:30:34] FAIL: USB get_lock not found (6:20)

I'm using a PL2303 device, and when I start cgminer, I lose the /dev/ttyUSB0 (I think cgminer is grabbing it)

Any thoughts?

[zziggi]

Use cp2102 ,PL2303 has a different PID/VID

[forpblat]

Use cp2102 ,PL2303 has a different PID/VID

Could I recompile bfgminer for PL2303 using the correct PID/VID (I'm on Linux)? Sorry for the silly questions, this is my first exposure to ASIC mining that wasn't plug n' play.

[forpblat]

I ended up changing all the PIDs/VIDs in the "rules" file of both cgminer and bfgminer to match a PL2303, but still no dice. I assume that there's more to it than that, and it's further hardcoded based on each asic type?

[pekatete]

I tried to compile with a MinGW different versions of cgminer ( bitmain,kano,original),
all give an error in usbutils.c  when compiling with --enable-bmsc --enable-bitmain  --enable-icarus (--enable-icarus   it is necessary to detect the usb device).

OK this is quite easy to resolve, though after compiling and running, only one board is detected and it still throws errors. Here's what you need to do to get rid of the errors in usbutils.

1. The first set of errors refers to previously defined structures (13 in total), i.e there are ifdef conditions that overlap for ICARUS and BMSC. I opened the file usbutils.c and commented out the initial conflicting declarations inside #ifdef USE_ICARUS (it does not matter whether you do the other!). Not all declarations inside here need to be commented out, declarations 3, 4 and 5 are not duplicated.

2. The next error thrown was a conflict of declaration of the function rev. After removing the static marker on the function inside driver-icarus.c, the compiler threw another error relating to the same function as having been included from another header file. So I un-commented the function out completely from the driver-icarus.c file and re-compiled.

After making the above changes, recompilation was successfull and the executable runs, albeit with the error mentioned above.

[idonothave]

Here's your heat sink.

https://www.onlinemetals.com/merchant.cfm?pid=1162&step=4&showunits=inches&id=997&top_cat=60

Just drill a few holes in it, and run some water through it. I designed a waterblock a couple of months ago but it would have cost around $80 to make. I gave up on it for a while and decided to make my own. I borrowed a buddy's drill press and tried to go the long way. I ended up going through one of the sides. I only ordered I piece of metal, so I put it on the back burner for a while. I thought about routing a piece and using another thin piece to create a channel, but I don't have access to a cnc machine. For mounting the boards I was going to drill holes all the way through and use a machine screw and nut with nylon washers to pull the boards tight against the waterblock. If anyone has access to  machining equipment and can make some simple waterblocks for a reasonable price I would be interested in a few.

Great job J4bberwock for getting a control board together. I wanted to do it, but I'm not that smart.

look this https://www.gme.cz/chl255e-p620-064

[pekatete]

Here's your heat sink.

https://www.onlinemetals.com/merchant.cfm?pid=1162&step=4&showunits=inches&id=997&top_cat=60

Just drill a few holes in it, and run some water through it. I designed a waterblock a couple of months ago but it would have cost around $80 to make. I gave up on it for a while and decided to make my own. I borrowed a buddy's drill press and tried to go the long way. I ended up going through one of the sides. I only ordered I piece of metal, so I put it on the back burner for a while. I thought about routing a piece and using another thin piece to create a channel, but I don't have access to a cnc machine. For mounting the boards I was going to drill holes all the way through and use a machine screw and nut with nylon washers to pull the boards tight against the waterblock. If anyone has access to  machining equipment and can make some simple waterblocks for a reasonable price I would be interested in a few.

Great job J4bberwock for getting a control board together. I wanted to do it, but I'm not that smart.

look this https://www.gme.cz/chl255e-p620-064

and this? http://www.ebay.co.uk/itm/300-140-20mm-Aluminum-HeatSink-for-LED-Power-IC-Transistor-Module-PBC-/321199970078?pt=UK_BOI_Electrical_Components_Supplies_ET&hash=item4ac902971e

[idonothave]

I tried to compile with a MinGW  win32 but still no luck.compiled without errors but  cgminer finds no usb devices,maybe this code is only for Linux?

PS Proto №2) Heatsink is made of aluminum corner 50*50mm but necessary if the voltage is above 0.9v
 ......

Same here! Compiled under MinGW but will not detect the devices!

I am running 2 boards without any heatsink, fully powered but with a 150 CFM fan at full blast blowing over them. I have seen average speeds (poolside) of up to 106 GH/s (when I set a high minimum diff), and ~85 GH/s when the pool sets the diff.

One anormally though is that, using the precompiled version 3.8.5, I can not have two CP2102's in the same computer. I can hav one pluged into the PC / laptop, and another plugged into a hub connected to another usb port then it is ok. Also, I can not start with both plugged in, I have to either start without the modules pluged in, or just one module, else cgminer crashes!

J4bberwock -  do you have similar issues with the board? Also, I put myself down for the protos (also to work on windows software, sorry not a C++ geek yet!), are we ready to go and do I need to cover your postage / outlay?

PS. zziggi - how are those heatsinks coping? I have seen some heatsinks 300mm X 140 for approx $15 each, also for normal operation, syscooling do the C1 heatink for $49 (I think + shipping) ....

[CHAOSiTEC]

finally got home :-)

ive connected 1 blade:

--bmsc-options 115200 --bmsc-freq 0781


this blade is underclocked to 200 MHz running aprox 45 GHs...

estimated speed at this setting should be 50 GHs (+/-10%)

and it seems like it within those specs




Im waiting for my hot air solder that ive ordered, and some 12k ohm resistors, which should bring me to 3.25 kohm (per bitmaintechs official undervolt) for testing

edit:

after setting the correct timeout with --bmsc-options 115200:0.6 im hitting the specs:

the 0.6 = timeout of 60



and this is after 5 minutes:



ill do a last test, with default MH setting

edit too many errors at default speed and more or less boiling hot, but i did get it to run ok at 325 MHz timeout at 43 (0.43 in bmsc-options)



i dont have anything to do a descent cooling atm, so im not going to push it more until i get cooling fins

edit:

J4bberwock, ill look into building a driver for your board when i recieve it

[pekatete]

^^^
I know you are on *nix but I have to ask, is that a stock 4.7.0 or did you recompile with modifications? I ask because I have been able to compile a useable 4.7.0 on MinGW but it is not working ... detects one board as a LIX and the other as a ANU then on other occassions as AMU ... (not really *nix related, but just to give some context).

Also, I can not set the timeout to 0.6 on 3.8.5 (windows) as it expects, according to the error message before it bails out, a number greater than 0 (which I thought 0.6 was!).

[CHAOSiTEC]

^^^
I know you are on *nix but I have to ask, is that a stock 4.7.0 or did you recompile with modifications? I ask because I have been able to compile a useable 4.7.0 on MinGW but it is not working ... detects one board as a LIX and the other as a ANU then on other occassions as AMU ... (not really *nix related, but just to give some context).

Also, I can not set the timeout to 0.6 on 3.8.5 (windows) as it expects, according to the error message before it bails out, a number greater than 0 (which I thought 0.6 was!).

im using bitmaintechs branch (4.7) with --enable-bmsc plus these changes:

https://bitcointalk.org/index.php?topic=671128.msg9399337#msg9399337

and im working on some other changes as well but those above should help

ill post a git when i have more time, since there are a few issues i would like to get fixed, and i will make it into a driver for the official cgminer when i get to play with j4bberwocks prototype

[pekatete]

I thought I'd missed something you put out earlier! Thanks for the pointer.

I managed to compile having made the changes to driver-bmsc.c, aka
replace memcpy((char *)&nonce, nonce_bin, sizeof(nonce_bin)); with memcpy((char *)&nonce, nonce_bin, sizeof(nonce));

Then issued:
./autogen.sh --enable-bmsc CFLAGS="-02 -Wall -W -march=i686" --host=i686-pc-mingw32
make

I did NOT include driver-icarus as has been suggested somewhere else, and the device was detected albeit with some issues. Below are the two screen shots i grabbed:

EDIT: I run the executable with the options: --bmsc-options 115200:0.6  --bmsc-freq 0F81

[CHAOSiTEC]

I thought I'd missed something you put out earlier! Thanks for the pointer.

I managed to compile having made the changes to driver-bmsc.c, aka
replace memcpy((char *)&nonce, nonce_bin, sizeof(nonce_bin)); with memcpy((char *)&nonce, nonce_bin, sizeof(nonce));

Then issued:
./autogen.sh --enable-bmsc CFLAGS="-02 -Wall -W -march=i686" --host=i686-pc-mingw32
make

I did NOT include driver-icarus as has been suggested somewhere else, and the device was detected albeit with some issues. Below are the two screen shots i grabbed:

<!snip>

what usb driver have you installed? to me that seems like a driver issue

[pekatete]

I thought I'd missed something you put out earlier! Thanks for the pointer.

I managed to compile having made the changes to driver-bmsc.c, aka
replace memcpy((char *)&nonce, nonce_bin, sizeof(nonce_bin)); with memcpy((char *)&nonce, nonce_bin, sizeof(nonce));

Then issued:
./autogen.sh --enable-bmsc CFLAGS="-02 -Wall -W -march=i686" --host=i686-pc-mingw32
make

I did NOT include driver-icarus as has been suggested somewhere else, and the device was detected albeit with some issues. Below are the two screen shots i grabbed:

<!snip>

what usb driver have you installed? to me that seems like a driver issue

I copied the libusb-1.0.dll into the directory where the exe resides, aside from that, normal windows (though I probably installed a zadig sometime ago ... but I may remember removing it ..... !? could this be the issue?)

[CHAOSiTEC]

I thought I'd missed something you put out earlier! Thanks for the pointer.

I managed to compile having made the changes to driver-bmsc.c, aka
replace memcpy((char *)&nonce, nonce_bin, sizeof(nonce_bin)); with memcpy((char *)&nonce, nonce_bin, sizeof(nonce));

Then issued:
./autogen.sh --enable-bmsc CFLAGS="-02 -Wall -W -march=i686" --host=i686-pc-mingw32
make

I did NOT include driver-icarus as has been suggested somewhere else, and the device was detected albeit with some issues. Below are the two screen shots i grabbed:

<!snip>

what usb driver have you installed? to me that seems like a driver issue

I copied the libusb-1.0.dll into the directory where the exe resides, aside from that, normal windows (though I probably installed a zadig sometime ago ... but I may remember removing it ..... !? could this be the issue?)

try reinstalling zadig

http://sourceforge.net/projects/libwdi/files/zadig/
http://zadig.akeo.ie/

[pekatete]

I thought I'd missed something you put out earlier! Thanks for the pointer.

I managed to compile having made the changes to driver-bmsc.c, aka
replace memcpy((char *)&nonce, nonce_bin, sizeof(nonce_bin)); with memcpy((char *)&nonce, nonce_bin, sizeof(nonce));

Then issued:
./autogen.sh --enable-bmsc CFLAGS="-02 -Wall -W -march=i686" --host=i686-pc-mingw32
make

I did NOT include driver-icarus as has been suggested somewhere else, and the device was detected albeit with some issues. Below are the two screen shots i grabbed:

<!snip>

what usb driver have you installed? to me that seems like a driver issue

I copied the libusb-1.0.dll into the directory where the exe resides, aside from that, normal windows (though I probably installed a zadig sometime ago ... but I may remember removing it ..... !? could this be the issue?)

try reinstalling zadig

http://sourceforge.net/projects/libwdi/files/zadig/
http://zadig.akeo.ie/

I re-installed zadig from the earlier one I downloaded from ckolivas's site (v 2.0) and rebooted but STILL get the same error messages. Actually, the reason I rmoved the drivers before was that with version 3.8.5 and the drivers installed, it will only detect (and use) one board only, with the drivers removed, it detects and uses both boards (though I have to plug them in after starting cgminer).

Any pointers?

[CHAOSiTEC]

I thought I'd missed something you put out earlier! Thanks for the pointer.

I managed to compile having made the changes to driver-bmsc.c, aka
replace memcpy((char *)&nonce, nonce_bin, sizeof(nonce_bin)); with memcpy((char *)&nonce, nonce_bin, sizeof(nonce));

Then issued:
./autogen.sh --enable-bmsc CFLAGS="-02 -Wall -W -march=i686" --host=i686-pc-mingw32
make

I did NOT include driver-icarus as has been suggested somewhere else, and the device was detected albeit with some issues. Below are the two screen shots i grabbed:

<!snip>

what usb driver have you installed? to me that seems like a driver issue

I copied the libusb-1.0.dll into the directory where the exe resides, aside from that, normal windows (though I probably installed a zadig sometime ago ... but I may remember removing it ..... !? could this be the issue?)

try reinstalling zadig

http://sourceforge.net/projects/libwdi/files/zadig/
http://zadig.akeo.ie/

I re-installed zadig from the earlier one I downloaded from ckolivas's site (v 2.0) and rebooted but STILL get the same error messages. Actually, the reason I rmoved the drivers before was that with version 3.8.5 and the drivers installed, it will only detect (and use) one board only, with the drivers removed, it detects and uses both boards (though I have to plug them in after starting cgminer).

Any pointers?

since i dont use windows, and my gf does not like me playing with her laptop :-p im unable to give any pointers atm

edit:

do you have a rasbpi nearby? try it with that if you do

[CHAOSiTEC]

mining for 20 minutes, with timeout of 60 and 200 MHz with voltmodded board: (diff forced to 256 at pool)

[pekatete]

since i dont use windows, and my gf does not like me playing with her laptop :-p im unable to give any pointers atm

edit:

do you have a rasbpi nearby? try it with that if you do

rPi is currently pre-occupied with her indoors ....., but yeah, thanks anyhow.
I'll keep toying with the zadig's, as having looked closer, it seems it only installed the driver for one CP2102 and not the other, so I will try associating both with the extracted drivers manually and report back if that yields anything useful, else assume a status quo as it is now.