IN STOCK - OneStringMiner boards, up to 39GH/s last ITEMS: from US$15...FUN!!! - page 12.

Taugeran

hero member

Activity: 658

Merit: 500

CCNA: There i fixed the internet.

Quote from: cscape on March 19, 2014, 02:46:55 AM

Sending unnecessary flush commands is bad for performance, since it wipes the entire work queue inside the device. For typical bitcoin mining, the device works fine without flush, as it will automatically flush when a new block is detected. Also, it is unnecessary to send target or maxroll repeatedly. They only need to be send again when they're changed.

Also, a low maxroll value is also bad for performance. For instance, with maxroll = 0, a single OSM board needs about 11-12 work items per second to keep busy. And if the boards are chained through the serial link, they could require nearly 200 work items per second. That's a lot. Setting maxroll to 60, only requires 1/60th of the workload, so about 1 work item every 5 seconds for a board.

In addition, it is good to send the pool difficulty to the board using 'target' command, to avoid unnecessary traffic going the other way.

Thanks for the insight cscape.

A question that comes to mind does the clock command accept 1 value per chip so clock command for an OSM would be CLOCK followed by 15 space separated values?

BenTuras

hero member

Activity: 826

Merit: 1001

Quote from: wttbs on March 18, 2014, 05:58:22 PM

Got my One String Miners today ! and... I loved them ! They are great.

I have 4 DIY1 and 4 DIY2 kits. This evening I build up the 4 DIY2 kits and build a small tower. Quality of the boards and other materials is high. Building was very easy. Screw the pcb to the heatsink, connect the wires to the powerconnectors, that’s about it. I used a BeQuiet PurePower 600W PSU I had laying around. Attached two fans, all in a temporary setup. 10 Port USB Hub is a Rosewill RHB-500. I used an ArchLinux PC, downloaded and compiled CGminer 3.12.3, turned on the miners, started CGMiner and it detected all miners. 3 miners were “ZOMBIE” at first but after just 10 seconds all boards were recognized and hashing. All works great with any issues. Temperatuur is stable with only 2 fans (in the most efficient position), 30 to 40 degrees Celcius. I haven’t tuned them yet so they hash around 25GH/s per board.

Photos: http://miningonmyown.com/2014/03/18/one-string-miner/

Thank you for the compliments.
With heat sinks, the board should do 30+ GH/s each, so no tuning should be necessary.
Only if you want to go way over 30GH/s, you can overvolt them, but that voids your warranty.
Thanks for the nice webpage too, I hope you don't mind me sending the link to some people Wink

Oh, and the boards are designed in Holland, but produced in Asia.

cscape

sr. member

Activity: 251

Merit: 250

Sending unnecessary flush commands is bad for performance, since it wipes the entire work queue inside the device. For typical bitcoin mining, the device works fine without flush, as it will automatically flush when a new block is detected. Also, it is unnecessary to send target or maxroll repeatedly. They only need to be send again when they're changed.

Also, a low maxroll value is also bad for performance. For instance, with maxroll = 0, a single OSM board needs about 11-12 work items per second to keep busy. And if the boards are chained through the serial link, they could require nearly 200 work items per second. That's a lot. Setting maxroll to 60, only requires 1/60th of the workload, so about 1 work item every 5 seconds for a board.

In addition, it is good to send the pool difficulty to the board using 'target' command, to avoid unnecessary traffic going the other way.

Taugeran

hero member

Activity: 658

Merit: 500

CCNA: There i fixed the internet.

Quote from: cscape on March 18, 2014, 04:17:42 AM

Quote from: Taugeran on March 18, 2014, 04:00:35 AM

Code:

 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: SEND version
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV hwerror 0
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV job 67 5327f8d0 0
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV needwork 3
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV hwerror 1
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV hwerror 2
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV job 69 5327f8d1 2
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV needwork 3
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV hwerror 3
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV hwerror 4
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV job 6c 5327f8d3 4
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV needwork 3
  [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV version 1.2 rev 1 chips 2

Ben the preceding is a snippet of a log from bfgminer run with --device-protocol-dump

it is plainly clear that the firmware shipped on the OSM boards is hard coded with a response the version command. if this is so and you have access to the source it would need modified to 15. if not could you request cscape recompile a fw specifically for the OSM boards with this modification.

Following this post, i will be trying to hack together a modification to bfgminer's Bifury driver to see if this fixes the poor performance.

the issue lies in the way bfgminer uses the FW response to VERSION to allocate and setup the queue. since it replies only 2 chip in the response the other 12 are left hungry as f*** and any values they find arent being logged since as far as BFG knows they dont exist

We'll change the firmware, but note that the 'chip' parameter was never intended to be used to calculate work load. Instead, if the board requires more work, it sends the 'needwork ' command to the miner. This is a much more accurate representation of exactly how many jobs are required by the firmware to ensure smooth mining.

Yea I'm still trying to work out the fundamental differences between cgminer BXF driver and bfgminer BIF driver.

I'll be quite honest bfgminers BiFury driver needs a lot of work. Just from looking it seems that on every poll it tells the device to flush, set target, and set maxroll. No idea how much workload that is on the PIC but I'm sure several times a second can't be light.

TO the drawing board... tonite after work and a few angry orchards

wttbs

legendary

Activity: 2210

Merit: 1109

Got my One String Miners today ! and... I loved them ! They are great.

I have 4 DIY1 and 4 DIY2 kits. This evening I build up the 4 DIY2 kits and build a small tower. Quality of the boards and other materials is high. Building was very easy. Screw the pcb to the heatsink, connect the wires to the powerconnectors, that’s about it. I used a BeQuiet PurePower 600W PSU I had laying around. Attached two fans, all in a temporary setup. 10 Port USB Hub is a Rosewill RHB-500. I used an ArchLinux PC, downloaded and compiled CGminer 3.12.3, turned on the miners, started CGMiner and it detected all miners. 3 miners were “ZOMBIE” at first but after just 10 seconds all boards were recognized and hashing. All works great with any issues. Temperatuur is stable with only 2 fans (in the most efficient position), 30 to 40 degrees Celcius. I haven’t tuned them yet so they hash around 25GH/s per board.

Photos: http://miningonmyown.com/2014/03/18/one-string-miner/

BenTuras

hero member

Activity: 826

Merit: 1001

Quote from: yslyung on March 18, 2014, 02:38:33 AM

watching & really interested in this. awesome project. do you have anything more than 420gh? atm i'm running 9x290 cards and my power bill is insane here in korea. if i can get a good deal i will also help to promote your miner.

Each DIY0 is tested and hash at least 25GH/s. Some hash over 30GH/s. If you want to go higher, you'll need a heat sink(we are still waiting for our 2nd batch of heat sinks; only have a few left). Overvolting voids warranty and needs a heat sink for sure.

Number of DIY0 sets	outside EU price	price including VAT inside EU	price excluding VAT for business owners
1	$244	€213	€176
2	$214	€187	€155
3	$204	€178	€147
4-9	$199	€173	€143
10-25	$194	€169	€140

Prices are a piece when you buy the number of sets in the left column.
The price differences are caused by lower shipment cost per board, less additional items(for example 1 fan per 3 boards) and an applied discount when you buy more than 3.

Don't forget that the sets include everything you need to start mining, excluding a power supply and a Raspberry Pi.

And remember, shipping is FREE(FedEx), as is real and friendly customer support.

BenTuras

hero member

Activity: 826

Merit: 1001

Quote from: klondike_bar on March 17, 2014, 11:08:32 PM

...just needs to be scaled up and it would make for some mean, cheap 1TH machines.

Working on it Wink

Taugeran

hero member

Activity: 658

Merit: 500

CCNA: There i fixed the internet.

and now im utterly confused. time for a drink and some GDB magic. if all else to luke

Taugeran

hero member

Activity: 658

Merit: 500

CCNA: There i fixed the internet.

Quote from: BenTuras on March 18, 2014, 04:08:44 AM

Quote from: Taugeran on March 18, 2014, 04:00:35 AM

...the issue lies in the way bfgminer uses the FW response to VERSION to allocate and setup the queue. since it replies only 2 chip in the response the other 12 are left hungry as f*** and any values they find arent being logged since as far as BFG knows they dont exist
^_^ hoora stubbornness for digging

Thank you for investigating. I will forward your message. And love your phrase 'hungry as f***' Wink

well its true. they arent being fed work items. since bfg only sees 2 chips.

**update

manually changing the procs count in the cgpu initialization (hardcode from .procs = chips, to .procs = 15,) leads to a moderate increase in hashrate. moderate as in from utility hashrate of 7Ghash to about 10 but that was a short 45 second run. but still not optimal. more investigating to do. and there goes my sleep. just downed a glass and a half of turkish coffee

on a side note debug outputs a s**tload of useful data. 45 seconds produced 4MB logfile

cscape

sr. member

Activity: 251

Merit: 250

Quote from: Taugeran on March 18, 2014, 04:00:35 AM

Code:

 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: SEND version
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV hwerror 0
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV job 67 5327f8d0 0
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV needwork 3
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV hwerror 1
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV hwerror 2
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV job 69 5327f8d1 2
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV needwork 3
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV hwerror 3
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV hwerror 4
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV job 6c 5327f8d3 4
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV needwork 3
  [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV version 1.2 rev 1 chips 2

Ben the preceding is a snippet of a log from bfgminer run with --device-protocol-dump

it is plainly clear that the firmware shipped on the OSM boards is hard coded with a response the version command. if this is so and you have access to the source it would need modified to 15. if not could you request cscape recompile a fw specifically for the OSM boards with this modification.

Following this post, i will be trying to hack together a modification to bfgminer's Bifury driver to see if this fixes the poor performance.

the issue lies in the way bfgminer uses the FW response to VERSION to allocate and setup the queue. since it replies only 2 chip in the response the other 12 are left hungry as f*** and any values they find arent being logged since as far as BFG knows they dont exist

We'll change the firmware, but note that the 'chip' parameter was never intended to be used to calculate work load. Instead, if the board requires more work, it sends the 'needwork ' command to the miner. This is a much more accurate representation of exactly how many jobs are required by the firmware to ensure smooth mining.

BenTuras

hero member

Activity: 826

Merit: 1001

Quote from: Taugeran on March 18, 2014, 04:00:35 AM

...the issue lies in the way bfgminer uses the FW response to VERSION to allocate and setup the queue. since it replies only 2 chip in the response the other 12 are left hungry as f*** and any values they find arent being logged since as far as BFG knows they dont exist
^_^ hoora stubbornness for digging

Thank you for investigating. I will forward your message. And love your phrase 'hungry as f***' Wink

Taugeran

hero member

Activity: 658

Merit: 500

CCNA: There i fixed the internet.

Code:

 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: SEND version
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV hwerror 0
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV job 67 5327f8d0 0
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV needwork 3
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV hwerror 1
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV hwerror 2
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV job 69 5327f8d1 2
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV needwork 3
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV hwerror 3
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV hwerror 4
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV job 6c 5327f8d3 4
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV needwork 3
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV hwerror 5
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV hwerror 6
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV hwerror 8
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV hwerror 9
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV hwerror 10
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV hwerror 12
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV hwerror 13
 [2014-03-18 07:42:47] bifury fd=8: DEVPROTO: RECV version 1.2 rev 1 chips 2

Ben the preceding is a snippet of a log from bfgminer run with --device-protocol-dump

it is plainly clear that the firmware shipped on the OSM boards is hard coded with a response the version command. if this is so and you have access to the source it would need modified to 15. if not could you request cscape recompile a fw specifically for the OSM boards with this modification.

Following this post, i will be trying to hack together a modification to bfgminer's Bifury driver to see if this fixes the poor performance.

the issue lies in the way bfgminer uses the FW response to VERSION to allocate and setup the queue. since it replies only 2 chip in the response the other 12 are left hungry as f*** and any values they find arent being logged since as far as BFG knows they dont exist

^_^ hoora stubbornness for digging

yslyung

legendary

Activity: 1500

Merit: 1002

Mine Mine Mine

watching & really interested in this. awesome project. do you have anything more than 420gh? atm i'm running 9x290 cards and my power bill is insane here in korea. if i can get a good deal i will also help to promote your miner.

BenTuras

hero member

Activity: 826

Merit: 1001

And in the [email protected] inbox another happy customer:
"I received my first DIY1, and it's kind of amazing -- thanks!! Now I'm ready for 2 more, please. I would like to pay with BTC."

BenTuras

hero member

Activity: 826

Merit: 1001

Quote from: hephaist0s on March 17, 2014, 10:03:19 PM

Received mine, and I gotta say that these are awesome. Can you explain to me again the black magic used to get gen3 performance out of a gen2 chip which has been out for many months? Why hasn't everyone been running Bitfury chips on crazy-mega-power saving mode all along? Heck, even the board is simpler than most designs -- and thus, presumably, with fewer ways to fail.

I am somewhat floored by the gloriousness.

Thanks!

klondike_bar

legendary

Activity: 2114

Merit: 1005

ASIC Wannabe

Quote from: hephaist0s on March 17, 2014, 10:03:19 PM

Received mine, and I gotta say that these are awesome. Can you explain to me again the black magic used to get gen3 performance out of a gen2 chip which has been out for many months? Why hasn't everyone been running Bitfury chips on crazy-mega-power saving mode all along? Heck, even the board is simpler than most designs -- and thus, presumably, with fewer ways to fail.

I am somewhat floored by the gloriousness.

just needs to be scaled up and it would make for some mean, cheap 1TH machines.

Punin is starting to sell off Bitfury 6U 2.5TH/3kW miners now (overpriced) so they must be getting into gen2 heavily

hephaist0s

hero member

Activity: 711

Merit: 532

Received mine, and I gotta say that these are awesome. Can you explain to me again the black magic used to get gen3 performance out of a gen2 chip which has been out for many months? Why hasn't everyone been running Bitfury chips on crazy-mega-power saving mode all along? Heck, even the board is simpler than most designs -- and thus, presumably, with fewer ways to fail.

I am somewhat floored by the gloriousness.

BenTuras

hero member

Activity: 826

Merit: 1001

Quote from: ZBC3 on March 15, 2014, 03:29:39 AM

You are right on about having someone that answers questions . I really do appreciate great customer service when I receive it. I have a couple of ants, and they make me appreciate my one string boards. with the one string boards; my room isn't noticeably warmer, they make much less noise, and they don't drive my power bill up like the ants do.

I'll be adding to my one string collection soon.

A happy customer !
Cheers, and looking forward to your next order.

ZBC3

sr. member

Activity: 361

Merit: 250

You are right on about having someone that answers questions . I really do appreciate great customer service when I receive it. I have a couple of ants, and they make me appreciate my one string boards. with the one string boards; my room isn't noticeably warmer, they make much less noise, and they don't drive my power bill up like the ants do.

I'll be adding to my one string collection soon.

BenTuras

hero member

Activity: 826

Merit: 1001

Quote from: Stinky_Pete on March 14, 2014, 02:03:43 PM

Do you have the 6 chip USB devices? Do you intend to get them in future?

We don't have the 6 chip USB devices.
And what's in the future ... is yet to be written Wink

Topic: IN STOCK - OneStringMiner boards, up to 39GH/s last ITEMS: from US$15...FUN!!! - page 12. (Read 38740 times)