Bitburner Fury - Hashrate Protection

[mgio]

Somehow I finally got all 8 boards hashing at once!!

I wasn't able to get my homemade CAN-BUS cable to work with all 8 miners at the same time but I was able to get it work with 4 pretty consistently. I plugged the cable into the 3 miners with broken USB connectors and one good one and left the other plugs unconnected. For those other 4 miners I used USB cables. So 1 CANBUS cable and 5 USB cables.

I started up CGMiner with the 4 CAN-BUS-ed miners plugged in first and then plugged in the other 4 USB connectors one by one as they were recognized and they all seem to be running now!

I really have no idea what the correct CGMiner options should be though. The "number of chips" option doesn't seem to do anything at all. Right now I have 1 cluster of 4 miners and 4 single miners all connected in cgminer. It looks like this:

 BTB 0: 50C 270 1023mV | 190.1G/190.9Gh/s | A:51789 R: 42 HW:5731 WU:2667.9/m
 BTB 1: 49C 270 1026mV | 47.67G/47.58Gh/s | A:12772 R: 15 HW:1992 WU: 668.0/m
 BTB 2: 50C 270 1028mV | 42.02G/44.93Gh/s | A:12643 R: 15 HW:2550 WU: 628.0/m
 BTB 3: 39C 270 1022mV | 41.79G/45.84Gh/s | A:11613 R:127 HW:2268 WU: 641.8/m
 BTB 4: 43C 270 1025mV | 47.17G/45.31Gh/s | A:10220 R:316 HW:2041 WU: 634.8/m

My CGMiner command line was:

sudo cgminer -o stratum+tcp://stratum.mining.eligius.st:3334 -u <bitcoin address> -p x --avalon-options 115200:128:10:47:270 --bitburner-voltage 1300

What does that "number of chips" setting actually do?

The difficulty seems to be stabilized at 128. That is likely a little high for the individual miners. Maybe I should run them in a separate CGMiner process.

These seem to be mining happily at 360 GH/s or so right now. I'm going to try and play with the settings to get a little bit more out of them.

I also epoxied the USB connectors for the miners that still have the connectors attached. I don't know if I will ever be able to remove the cables from them again but they are securely attached now.

[Starcraftman]

Can I plug in or remove the power connector while the psu is running? Don't wanna blow the board up.

Yes, this is safe. I have done it many times and there are no problems with my PSU or board.

[someone42]

When trying to connect boards with the CAN-BUS cable, I frequently get errors like:

[2013-10-10 18:38:49] BTB0: Comms error(buffer)
 [2013-10-10 18:38:49] usb_write error on avalon_write
 [2013-10-10 18:38:50] usb_write error on avalon_write
 [2013-10-10 18:38:50] usb_write error on avalon_write

It seems ot be kind of random. I've trying every combination of 2 - 8 boards that I can. The only thing consistent is the first and last boards. Sometimes it works, sometimes it doesn't but it is repeatable. I haven't been able to get more than 4 boards working at a time. But not always the same 4 boards.

What could be going wrong?

Are you using the latest firmware (https://www.dropbox.com/s/zn94oofozdn0eqb/BitBurnerFury_08.10.13.hex)? This firmware deals specifically with a "CAN-BUS causes usb_write error" issue. It only needs to be written onto the USB-connected boards, so it's okay if the CAN-connected ones aren't updated.

I really have no idea what the correct CGMiner options should be though. The "number of chips" option doesn't seem to do anything at all. Right now I have 1 cluster of 4 miners and 4 single miners all connected in cgminer.

...

What does that "number of chips" setting actually do?

The baud rate, number of chips and timeout are all ignored (they're vestigial from the Avalon-based BitBurners). The number of miners should be 16 * number of boards in the largest stack. Since you've divided your stack up, it should be 64. But if you had all 8 boards in a single stack it should be 128. It's okay if the number of miners is larger than necessary, though you might get a slightly higher stale rate due to cgminer queueing up more work than necessary.

[mgio]

When trying to connect boards with the CAN-BUS cable, I frequently get errors like:

[2013-10-10 18:38:49] BTB0: Comms error(buffer)
 [2013-10-10 18:38:49] usb_write error on avalon_write
 [2013-10-10 18:38:50] usb_write error on avalon_write
 [2013-10-10 18:38:50] usb_write error on avalon_write

It seems ot be kind of random. I've trying every combination of 2 - 8 boards that I can. The only thing consistent is the first and last boards. Sometimes it works, sometimes it doesn't but it is repeatable. I haven't been able to get more than 4 boards working at a time. But not always the same 4 boards.

What could be going wrong?

Are you using the latest firmware (https://www.dropbox.com/s/zn94oofozdn0eqb/BitBurnerFury_08.10.13.hex)? This firmware deals specifically with a "CAN-BUS causes usb_write error" issue. It only needs to be written onto the USB-connected boards, so it's okay if the CAN-connected ones aren't updated.

I really have no idea what the correct CGMiner options should be though. The "number of chips" option doesn't seem to do anything at all. Right now I have 1 cluster of 4 miners and 4 single miners all connected in cgminer.

...

What does that "number of chips" setting actually do?

The baud rate, number of chips and timeout are all ignored (they're vestigial from the Avalon-based BitBurners). The number of miners should be 16 * number of boards in the largest stack. Since you've divided your stack up, it should be 64. But if you had all 8 boards in a single stack it should be 128. It's okay if the number of miners is larger than necessary, though you might get a slightly higher stale rate due to cgminer queueing up more work than necessary.

That's good to know. I'll try updating the firmware on the miners with the USB connections tomorrow. Only problem is I don't have a Windows PC. Well, I have a PC running Linux, and I do have a Windows 7 disc so I guess I could install it but I don't want to if I don't have to.

The installation instructions say: "The next value is the number of miners, this must be the total number of chips in cluster multiplied by 2 (32 for 16 chips or one board)." so wouldn't it be 256 for a stack of 8 boards (8 * 2 * 16 = 256)?

[shapemaker]

WTF - where is my order? I should get it last week & again nothing happend - no answer to mail mail & PM.  

+1

order #115x, ordered before september 17th, nothing yet.

+1
order #116x

same here, #114x, ordered and paid before september 17th.
2nd week of october is almost over  

edit: order picked up by fedex today

No sign of order 111x ...

[mccampser]

my order 116x just arrived. thanks to cryptx for fast and complete delivery. now going to start it :-)

[Min€r]

my order 116x just arrived. thanks to cryptx for fast and complete delivery. now going to start it :-)

How can this be? When I have #108X & paid immediately... and again no answer to my mail.

[someone42]

That's good to know. I'll try updating the firmware on the miners with the USB connections tomorrow. Only problem is I don't have a Windows PC. Well, I have a PC running Linux, and I do have a Windows 7 disc so I guess I could install it but I don't want to if I don't have to.

The installation instructions say: "The next value is the number of miners, this must be the total number of chips in cluster multiplied by 2 (32 for 16 chips or one board)." so wouldn't it be 256 for a stack of 8 boards (8 * 2 * 16 = 256)?

To update firmware from Linux, you can try using pic32prog. Some modifications are required: see https://bitcointalk.org/index.php?topic=179769.msg2897971#msg2897971. I have made those changes in https://www.dropbox.com/s/7g7nygo7jbazd7z/pic32prog-an1388.tar.gz. After compiling it, power-on the boards and run "./pic32prog <hex file>" within the 12 second bootloader window.

About the "multiplied by 2" thing, that's a relic from when an input queue was introduced (you needed to multiply the actual number of miners by 2 to compensate for the input queue). cgminer versions >= 3.4.0 do not need the multiplication by 2. I think that instruction is there to account for people using older cgminer versions.

[gyverlb]

That's good to know. I'll try updating the firmware on the miners with the USB connections tomorrow. Only problem is I don't have a Windows PC. Well, I have a PC running Linux, and I do have a Windows 7 disc so I guess I could install it but I don't want to if I don't have to.

Same problem here, I compiled pic32prog for Linux and tried to upadate it but it wouldn't work : during a board boot (red led blinking) it times out and after the board is ready to hash (green led blinking) it can't find a device to work on.

I had to split the CAN bus in two halves to get it to mostly work with 2 stacks of 4 boards. The hashrate reported by cgminer is ~320GH/s after booting the 2 stacks but the pool reports 290GH/s. I can't deviate much from 1200mV and 256MHz or several boards freeze. Even with standard settings one of the 8 boards freeze after a while and need to be power cycled bringing my total hashrate at ~250GH/s at the pool.

Hope there's room for improvement: currently it doesn't even reach the minimum advertised speed of 320GH/s.

[gyverlb]

To update firmware from Linux, you can try using pic32prog. Some modifications are required: see https://bitcointalk.org/index.php?topic=179769.msg2897971#msg2897971. I have made those changes in https://www.dropbox.com/s/7g7nygo7jbazd7z/pic32prog-an1388.tar.gz. After compiling it, power-on the boards and run "./pic32prog <hex file>" within the 12 second bootloader window.

Could someone post this information on the instructions page? I had to cut my CAN bus cable to use my 8 boards and would have avoided it if I had known this : I only have 1 spare USB cable, cutting was the only option at the time.

[mccampser]

i just did the firmwareupdate via windows7 to version from 8.10.2013. after that cgminer could connect via usb :-) now i am trying to get some overclocking.... with defaults i get 43-48ghas.

[zulover]

got mine working last night, followed rages instructional post. nice one.

settings clock:270  volt:1330 getting around 48- 52ghs.   

at 300mhz and 1400v hit 53-54ghs that was tops but that setting might need some extra help with cooling, i dont know.

temp at 48c

[mccampser]

i think the voltage setting "--bitburner-voltage" is not correctly implemented.
if i set bitburner-voltage 1300 cgminer shows about 1020mV. if i set it to 1400 (maximum value) cgminer shows 1109mV.

I think when the voltage is correctly set to ~1300mV-1400mV it should be possible to reach 64ghash.
With avalon-setting "45:282" and bitburner-voltage "1400" i reach ~54ghash.

( i used both cgminer 3.5.0 from someone42 and default )

[Roy Badami]

i think the voltage setting "--bitburner-voltage" is not correctly implemented.
if i set bitburner-voltage 1300 cgminer shows about 1020mV. if i set it to 1400 (maximum value) cgminer shows 1109mV.

I think when the voltage is correctly set to ~1300mV-1400mV it should be possible to reach 64ghash.
With avalon-setting "45:282" and bitburner-voltage "1400" i reach ~54ghash.

( i used both cgminer 3.5.0 from someone42 and default )

That is correct.  The default voltagefor the Bitfury chips is 900mV, but the Bitburner Fury boards support under- and overvolting, so they support a voltage range of 700mv-1100mV.  The current firmware emulates an Avalon so it subtracts 300mV from the specified bitburner voltage - if it didn't the default voltage of 1200mV would fry your board!

So "--bitburner-voltage 1300" selects a voltage of 1000mV, which is approximately what we see.  If you could really set the core voltage to 1300mV you'd almost certainly destroy the chips.

roy

[zulover]

Makes sense, so whats need to up the hashrate to at least 60 if not 80 safely?

[mccampser]

ok, thanks for explaining it.
ok so i get maximum of ~52ghash @ 45:282 with 1300mV(1020mV).

far away from promised 64ghash.

[fydel]

Did any of you with the broken USB connectors got an answer from crpytx?

I have a similar problem and I mailed him four days ago. But did not receive an answer.
I don't wont to start a disput with paypal.

[mgio]

ok, I just got the most hilarious phone call from US Customs.

They said there was a international FedEx package for me from "cryptx" (Thank you Cryptx!! I wasn't expecting anything!) and the contents said "CANNABIS" or something and they wanted to know what it was.

I cracked up laughed and explained it was just a CANBUS cable and not drugs.

[zulover]

Loooool


Oh dear uncle sam.

[h@shKraker]

All,

To make a proper CAN-BUS ribbon cable with connectors you'll need the following parts ->

https://www.sparkfun.com/products/10647 (I ordered 1 because it's 15 feet)
https://www.sparkfun.com/products/10650 (I ordered 16 because I plan to have two stacks of 8 boards)

If you live in the US and ship USPS Priority this turns out to be cheaper and faster than ordering from asic-hardware.  Cheers.

H@shKraker