[ANN] Bitfury ASIC sales in EU and Europe - page 22.

Trupik

member

Activity: 136

Merit: 13

Quote from: GH on November 23, 2013, 05:20:02 PM

Hello all,

recently had my very first "oops" experience with my bitfury miner.
Was mining absolutely(!) rock stable with cron miner restarts 2 hourly at 225-230 Gh/s with lately 6 pro-modded(*) h-board for two month now.
Suddenly only the first 3 h-boards (0/4/5) had full useable nonces, 8/9/C were hashing at normal speed but did produce absolutely 0% useable nonces. No exceptional SPI nor MISO errors were detected.
Miner restarts (which work even in cases of a "hot-plug") and even complete reboot did not help. Only a halt with power-cycle restored full function.
Why? Any similar experience or explainations, anyone?

Thanks,
Gerald

*) 20x precision spindle trimmers for overvolting, additional power cabling to compensate for voltage drop over PCBs and even voltage for all chips, not heatsinks, 7 fans.

I do have similar experience. About once a week, the noncerate reported by the pool drops to 15-40% of former value. The bitfury itself reports full noncerate as if nothing happened. I have to log in and reboot the RaspberryPi (sudo shutdown -r now), then it works for another week or so. I was thinking about a watchdog script. Any ideas on how to do it?

GH

member

Activity: 117

Merit: 10

Quote from: GH on November 23, 2013, 05:20:02 PM

Hello all,

recently had my very first "oops" experience with my bitfury miner.
Was mining absolutely(!) rock stable with cron miner restarts 2 hourly at 225-230 Gh/s with lately 6 pro-modded(*) h-board for two month now.
Suddenly only the first 3 h-boards (0/4/5) had full useable nonces, 8/9/C were hashing at normal speed but did produce absolutely 0% useable nonces. No exceptional SPI nor MISO errors were detected.
Miner restarts (which work even in cases of a "hot-plug") and even complete reboot did not help. Only a halt with power-cycle restored full function.
Why? Any similar experience or explainations, anyone?

Thanks,
Gerald

*) 20x precision spindle trimmers for overvolting, additional power cabling to compensate for voltage drop over PCBs and even voltage for all chips, no heatsinks, 7 fans.

Hmmm. Now one of my 96 friends is missing in action...

Code:

89	aIfDSo	54	2.792	2.600	195	1	0	0	246	[C:8]	461	13 13 12 12 12 12 12 12 12 12 12 12 11 12 13 13 	0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 
90	aIfDSo	54	0.000	2.600	0	3375	0	0	246	[C:9]	756	0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 	338 338 337 337 0 0 337 337 337 0 0 338 338 338 0 0 
91	aIfDSo	54	2.319	2.484	162	0	0	0	235	[C:A]	496	10 11 11 11 10 10 10 10 10 10 10 10 10 10 10 9 	0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

Edit:
Hmmm 2: found his way home himself this morning, a little wounded...

Code:

89	aIfDSo	54	2.649	2.600	185	0	0	0	246	[C:8]	275	12 12 12 11 11 11 11 11 11 11 12 12 12 12 12 12 	0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 
90	aIfDSo	54	2.534	2.600	177	8	0	0	246	[C:9]	301	10 11 11 11 12 12 10 11 12 12 11 11 10 11 11 11 	1 0 0 0 0 0 2 1 0 0 1 1 2 0 0 0 
91	aIfDSo	54	2.534	2.473	177	0	0	0	234	[C:A]	332	12 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 	0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

klondike_bar

legendary

Activity: 2128

Merit: 1005

ASIC Wannabe

Quote from: spiccioli on November 23, 2013, 05:26:29 PM

Quote from: klondike_bar on November 23, 2013, 01:57:08 PM

Quote from: spiccioli on November 22, 2013, 08:37:26 AM

BTW,
having some hardware way to isolate "broken" chips without having to cut traces on the board would be a good thing.
spiccioli

pretty sure it was demonstrated that shorting the jumpers by the faulty chip causes it to simply by skipped over the same was as cutting the traces would achieve *i think thats what i read*

all of them or a particular couple?
spiccioli

if you look at the chip's traces, there are two sets: 1 runs to the capacitors (or where capacitors were in rev1) and the other set runs to vias to another layer of the PCB. These have very small silver pads that pair them up, and i belive are responsible for the MISO commands and other chaining functions. if the small pairs of pads are shorted, the chip is no longer included in the chain

https://bitcointalk.org/index.php?topic=250249.1040

step 1: short the small jumper pads and see how it works
step 2: if problem persists, either remove the chip or cut the traces (preferably a little better than how some have done it on here lol)

i know there is more info elsewhere but cant find it offhand

spiccioli

legendary

Activity: 1379

Merit: 1003

nec sine labore

Quote from: klondike_bar on November 23, 2013, 01:57:08 PM

Quote from: spiccioli on November 22, 2013, 08:37:26 AM

BTW,

having some hardware way to isolate "broken" chips without having to cut traces on the board would be a good thing.

spiccioli

pretty sure it was demonstrated that shorting the jumpers by the faulty chip causes it to simply by skipped over the same was as cutting the traces would achieve *i think thats what i read*

Hi,

all of them or a particular couple?

spiccioli

GH

member

Activity: 117

Merit: 10

Hello all,

recently had my very first "oops" experience with my bitfury miner.
Was mining absolutely(!) rock stable with cron miner restarts 2 hourly at 225-230 Gh/s with lately 6 pro-modded(*) h-board for two month now.
Suddenly only the first 3 h-boards (0/4/5) had full useable nonces, 8/9/C were hashing at normal speed but did produce absolutely 0% useable nonces. No exceptional SPI nor MISO errors were detected.
Miner restarts (which work even in cases of a "hot-plug") and even complete reboot did not help. Only a halt with power-cycle restored full function.
Why? Any similar experience or explainations, anyone?

Thanks,
Gerald

*) 20x precision spindle trimmers for overvolting, additional power cabling to compensate for voltage drop over PCBs and even voltage for all chips, no heatsinks, 7 fans.

intron

sr. member

Activity: 427

Merit: 251

- electronics design|embedded software|verilog -

Quote from: klondike_bar on November 23, 2013, 01:57:08 PM

Quote from: spiccioli on November 22, 2013, 08:37:26 AM

BTW,

having some hardware way to isolate "broken" chips without having to cut traces on the board would be a good thing.

spiccioli

pretty sure it was demonstrated that shorting the jumpers by the faulty chip causes it to simply by skipped over the same was as cutting the traces would achieve *i think thats what i read*

Also it might be needed to remove the defective chip.
You can use an hot air gun for that. Be careful not to
heat up the rest of the board to much though.

intron

klondike_bar

legendary

Activity: 2128

Merit: 1005

ASIC Wannabe

Quote from: spiccioli on November 22, 2013, 08:37:26 AM

BTW,

having some hardware way to isolate "broken" chips without having to cut traces on the board would be a good thing.

spiccioli

pretty sure it was demonstrated that shorting the jumpers by the faulty chip causes it to simply by skipped over the same was as cutting the traces would achieve *i think thats what i read*

rammy2k2

legendary

Activity: 1974

Merit: 1003

Quote from: Micky25 on November 23, 2013, 09:07:21 AM

Niko, one question:

As you put the full kits together, you tested which card works best in what slot and marked them accordingly (A1, A2...).
Is there a special way to do this, or is it just like swapping the cards around and let them run some time?

I ask because I have a couple of single cards and would like to find the optimal way (slot?) to run them.

move them all around ... thats what i did

Micky25

legendary

Activity: 974

Merit: 1000

Niko, one question:

As you put the full kits together, you tested which card works best in what slot and marked them accordingly (A1, A2...).
Is there a special way to do this, or is it just like swapping the cards around and let them run some time?

I ask because I have a couple of single cards and would like to find the optimal way (slot?) to run them.

nowywbitcoinowymswiecie

member

Activity: 71

Merit: 10

Quote from: punin on November 21, 2013, 01:09:16 PM

Actually, I'd like some comments from you...

- Would you like a USB miner or are you happy with standalone RasPi controlled unit?
- Do you feel modular design is important, or would you prefer a nice tidy box?
- What about form factor: free or rack?
- What dB level are you comfortable with?
- Internal custom PSU or ATX?

- stand-alone with Ethernet. Its the most convenient way. No PC/Laptop but self sustainable mashine.
- any
- any
- dont care
- any

Foofighter

hero member

Activity: 882

Merit: 547

BTC Mining Hardware, Trading and more

Quote from: intron on November 21, 2013, 03:25:55 PM

Quote from: punin on November 21, 2013, 01:09:16 PM

Actually, I'd like some comments from you...

- Would you like a USB miner or are you happy with standalone RasPi controlled unit?
- Do you feel modular design is important, or would you prefer a nice tidy box?
- What about form factor: free or rack?
- What dB level are you comfortable with?
- Internal custom PSU or ATX?

- stand-alone with Ethernet
- tidy box (foofighter)
- free
- less then BFL
- external power supply

And 200+ GH per board.

intron

+1

spiccioli

legendary

Activity: 1379

Merit: 1003

nec sine labore

BTW,

having some hardware way to isolate "broken" chips without having to cut traces on the board would be a good thing.

spiccioli

KNK

hero member

Activity: 692

Merit: 502

Quote from: punin on November 21, 2013, 01:09:16 PM

Actually, I'd like some comments from you...

- Would you like a USB miner or are you happy with standalone RasPi controlled unit?
- Do you feel modular design is important, or would you prefer a nice tidy box?
- What about form factor: free or rack?
- What dB level are you comfortable with?
- Internal custom PSU or ATX?

I've made similar pool few months ago and it seems the results have not changed much.

Keefe

hero member

Activity: 681

Merit: 500

Quote from: punin on November 21, 2013, 01:09:16 PM

Actually, I'd like some comments from you...

- Would you like a USB miner or are you happy with standalone RasPi controlled unit?
- Do you feel modular design is important, or would you prefer a nice tidy box?
- What about form factor: free or rack?
- What dB level are you comfortable with?
- Internal custom PSU or ATX?

- I hate how often I have to replace rPi SD cards, and I already have a host pc for other miners, so I lean toward usb connectivity instead of ethernet.
- I don't mind lots of separate units.
- Free is fine as long as it's stackable.
- Noise doesn't matter to me. I can't tolerate any noise in the house, so I run my miners elsewhere.
- ATX

Keefe

hero member

Activity: 681

Merit: 500

Quote from: punin on November 21, 2013, 01:59:10 PM

What if I said I have a PSU that can deliver 3kW at a price of ~400€ and ~90% efficiency?

How about 4 of these ($483 total including shipping in the US)?:
http://www.tigerdirect.com/applications/SearchTools/item-details.asp?EdpNo=7580853&CatId=2533
http://www.hardwaresecrets.com/article/XFX-PRO-750-W-Black-Edition-Full-Modular-Power-Supply-Review/1695/13

Up to 92% efficient, fully modular, and top quality (Seasonic KM3 core). Each can power one full rig and be about 90% efficient. Watch out for similarly named older products that aren't as good.

Micky25

legendary

Activity: 974

Merit: 1000

Quote from: punin on November 21, 2013, 01:09:16 PM

Actually, I'd like some comments from you...

I like them as they are, no changes needed for me, I just want more of them

Quote from: punin on November 21, 2013, 01:59:10 PM

What if I said I have a PSU that can deliver 3kW at a price of ~400€ and ~90% efficiency?

If one can power 2 full rigs I would buy 2.

BenTuras

hero member

Activity: 826

Merit: 1001

Quote from: punin on November 21, 2013, 01:09:16 PM

Actually, I'd like some comments from you...

- Would you like a USB miner or are you happy with standalone RasPi controlled unit?
- Do you feel modular design is important, or would you prefer a nice tidy box?
- What about form factor: free or rack?
- What dB level are you comfortable with?
- Internal custom PSU or ATX?

1. Raspi is best, available everywhere in case of malfunction. Al this proprietary stuff is a pain to replace when needed.
2. modular, easiest to service
3. free
4. not an issue, I am using very silent shark fans
5. atx, again available everywhere. currently using Corsair AX1200i, shows me the Amps on every PCI power connector!

greaterninja

hero member

Activity: 924

Merit: 1000

IMO, you guys need to bring chip prices down...not january... now. Europe, America, Asia should pool funds so we can order 500k to 1,500k chips for cents on the USD each.

I am also willing to throw 10k toward chip packaging services and help setup the deal.

gingernuts

member

Activity: 89

Merit: 10

Quote from: punin on November 21, 2013, 01:09:16 PM

Actually, I'd like some comments from you...

- Would you like a USB miner or are you happy with standalone RasPi controlled unit?
- Do you feel modular design is important, or would you prefer a nice tidy box?
- What about form factor: free or rack?
- What dB level are you comfortable with?
- Internal custom PSU or ATX?

Standalone with ethernet - doesn't need to be a rpi, I don't want to have to manage a load more linux boxes, any <10Euro micro that can talk to a Stratum proxy will do.
Probably nice-tidy box or board on heatsink.
Free
Noise level - about the same as a PC if possible
External PSU

And I think these things need to start at 50-100GH/s - the network's moved on from 10GH/s being a load!

spiccioli

legendary

Activity: 1379

Merit: 1003

nec sine labore

Quote from: punin on November 21, 2013, 01:09:16 PM

Actually, I'd like some comments from you...

- Would you like a USB miner or are you happy with standalone RasPi controlled unit?
- Do you feel modular design is important, or would you prefer a nice tidy box?
- What about form factor: free or rack?
- What dB level are you comfortable with?
- Internal custom PSU or ATX?

Ok, you asked for it:

- No USB, it makes no sense right now to have 2 GHs units, it is just a waste of time using them
- break even in BTC and so it HAS to have a reasonable cost
- power efficiency, in the end, if you don't live in very peculiar places on earth energy costs a lot!

Noise and form factor have very marginal value.

What I'd really like to have is a more robust systems.

A system which does not depend on board order or ambient temperature.

A more robust m-board, current ones have too much flaky connectors.

Price, price and price, or as I said earlier, break-even in BTC is a MUST.

spiccioli

Topic: [ANN] Bitfury ASIC sales in EU and Europe - page 22. (Read 250482 times)