[ANN] Bitfury ASIC sales in EU and Europe - page 74.

Meizirkki

hero member

Activity: 616

Merit: 500

Quote from: dani on September 04, 2013, 04:39:54 PM

I'm not giving up on this so easily. I seperated that faulty board and put it on a m-board as a single. Strange thing:

it just works "fine" (by all means) except for 3 chips, I don't care right now.

Even more strange? The two hboards left on the first mboard perform badly, one between 10 and 15 and the other around 10-12. I saw both hashing at >20 happily before.

I also checked the mboards for soldering issues, they look pretty clean, no problems. I'll try to put the other two hboards one by one to the second master board, let's see how this works out. Maybe its a faulty mboard after all?

or a bad raspi, maybe? Those things corrupt memorycards a lot, you might want to check that too

dani

hero member

Activity: 525

Merit: 500

..yeah

I'm not giving up on this so easily. I seperated that faulty board and put it on a m-board as a single. Strange thing:

it just works "fine" (by all means) except for 3 chips, I don't care right now.

Even more strange? The two hboards left on the first mboard perform badly, one between 10 and 15 and the other around 10-12. I saw both hashing at >20 happily before.

I also checked the mboards for soldering issues, they look pretty clean, no problems. I'll try to put the other two hboards one by one to the second master board, let's see how this works out. Maybe its a faulty mboard after all?

Edit:
Did that. Getting around 18gh + 18gh on the 2nd mboard and 19gh on the single left on the original mboard. Anyone else finds this interesting? Cheesy

They still do a pretty bad job compared to others I've seen. But a little tuning should help, right Mxrider? Wink

tccd

newbie

Activity: 51

Merit: 0

I have one suspicious chip too:

http://img28.imageshack.us/img28/8028/ojwe.png

Look at number 16 on board 1. Just tried spiccioli's 50 value but the noncerate is still 0 for this chip.

Micky25

legendary

Activity: 974

Merit: 1000

My opinion: we should be happy having received anything at all Wink

WynX

member

Activity: 77

Merit: 10

Quote from: dani on September 04, 2013, 04:09:02 PM

I guess I'm just very unlucky, ordering 50gh and receiving 45gh.. Huh

I changed i to I, d to D, in every possible way, no way to turn on 38 or any later chip..
Thank you guys for your help, I will try to fire up the 2nd masterboard with the faulty hcard and another psu, but my hopes are already crushed.

Right now its hashing with 20gh, what the..

Have you checked for bad soldering on that particular board? Also the underside of the M-card?
See for ref pics: http://imgur.com/a/t3AKx

EDIT: especially the I/O pins shorting on that chip 38 would match that behaviour (if you can locate it?)

dani

hero member

Activity: 525

Merit: 500

..yeah

Quote from: spiccioli on September 04, 2013, 04:02:39 PM

Quote from: Isokivi on September 04, 2013, 03:53:58 PM

Im not entirely sure, but I'd say that chip 38 is dead and since they are chained any chips beyond it will not function.

I have a h-card with chips 7-16 inclusive which are dead, chip 16 barely hashes setting its speed at 50, but throws a ton of miso-errors.

spiccioli

I guess I'm just very unlucky, ordering 50gh and receiving 45gh.. Huh

I changed i to I, d to D, in every possible way, no way to turn on 38 or any later chip..
Thank you guys for your help, I will try to fire up the 2nd masterboard with the faulty hcard and another psu, but my hopes are already crushed.

Right now its hashing with 20gh, what the..

spiccioli

legendary

Activity: 1379

Merit: 1003

nec sine labore

Quote from: spiccioli on September 04, 2013, 04:02:39 PM

Quote from: Isokivi on September 04, 2013, 03:53:58 PM

Im not entirely sure, but I'd say that chip 38 is dead and since they are chained any chips beyond it will not function.

I have a h-card with chips 7-16 inclusive which are dead, chip 16 barely hashes setting its speed at 50, but throws a ton of miso-errors.

spiccioli

I know that speed range should be 52-57 but I've found that I can go as low as 50 when I have a problematic chip, see this for example

Code:

24 aIfDSo 50 0.786 1.250 52 39 0 0 112 [4:7] 707 2 3 4 4 2 2 2 2 4 4 4 3 3 4 5 4 4 3 2 2 3 3 3 3 1 2 2 3 3 2 1 2

spiccioli

legendary

Activity: 1379

Merit: 1003

nec sine labore

Quote from: Isokivi on September 04, 2013, 03:53:58 PM

Im not entirely sure, but I'd say that chip 38 is dead and since they are chained any chips beyond it will not function.

I have a h-card with chips 7-16 inclusive which are dead, chip 16 barely hashes setting its speed at 50, but throws a ton of miso-errors.

spiccioli

Micky25

legendary

Activity: 974

Merit: 1000

I would try different slots and make sure the boards sit really good. But probably Isokivi is right. I'm no pro.

Isokivi

hero member

Activity: 910

Merit: 1000

Items flashing here available at btctrinkets.com

Quote from: dani on September 04, 2013, 03:50:28 PM

Quote from: MXRider on September 04, 2013, 01:15:29 PM

No, not that much.

Could you post another picture (or a link to that picture) of the .stat.log -file. I'm pretty sure you have autotune on and several chips are turned off.

sure, here we go:

http://i.imgur.com/pqaq5Qo.png

Im not entirely sure, but I'd say that chip 38 is dead and since they are chained any chips beyond it will not function.

Micky25

legendary

Activity: 974

Merit: 1000

Quote from: dani on September 04, 2013, 03:17:07 PM

Can i put the third card inti c2 oder D1? Maybe the card has an issue with the pins?

the h-boards should work in any slot, but optimal in a1 b1 c1 d1. Probably its only some kind of voodoo.

dani

hero member

Activity: 525

Merit: 500

..yeah

Quote from: MXRider on September 04, 2013, 01:15:29 PM

No, not that much.

Could you post another picture (or a link to that picture) of the .stat.log -file. I'm pretty sure you have autotune on and several chips are turned off.

sure, here we go:

http://i.imgur.com/pqaq5Qo.png

mr_rulezzz

member

Activity: 61

Merit: 10

sooo ... now i have chip.cnf with not tuned lines in and best.cnf with auto tune on selected chips only ... i start from webface new ... which cnf ist now in use?

Micky25

legendary

Activity: 974

Merit: 1000

maybe you want to try 16. This seems to work nice for me.

dani

hero member

Activity: 525

Merit: 500

..yeah

Quote from: Micky25 on September 04, 2013, 01:48:39 PM

dani, you mine on a pool? What's your difficulty?

Yes, on eligius, set the difficulty to 32 over there before i left the house, seems like it didnt change anything. I'll be home in 30mins, then i can check it.

Can i put the third card inti c2 oder D1? Maybe the card has an issue with the pins?

arorts

sr. member

Activity: 408

Merit: 250

Quote from: punin on September 04, 2013, 12:59:21 PM

Sorry for the delay guys.. Lot's of things happening here..

Here are photos of some of the soldering issues we've encountered. Please check your boards!

Thanks for the heads-up, Punin!

Just to get a feel of how likely I'll run into this issue, approx what percentage of boards are having this issues?

mr_rulezzz

member

Activity: 61

Merit: 10

Quote from: MXRider on September 04, 2013, 02:30:46 PM

I'm raising or lowering the speed of the chips, depending on the number of errors. I have also turned off the autotune. The point of manual tuning is that it works a little bit better than autotune.

Let's take the first chip as an example. Autotune is on (AIfDSo). It can be turned off by changing the capital A to a normal a (aIfDSo). Now the chip will always run at the speed you have set. Current speed is 55. Minimum is 52 and maximum is 57, higher the better of course. In this case I would raise the speed to 56 since there are practically no errors.

Chip two: change the "aifdso" to "aIfDSo" and let's see how it does (it won't work at all with aifdso).

Chip three: Something wrong with it. Turn off the autotune and keep the speed at 52. At least this is what I would do because 0,6 GH/s is still better than 0.

More info here http://punin.com/README_BFSB_v0.1.txt

thank you very much! =) ... (every time i fall in this trap, read teh fckn manual)

Isokivi

hero member

Activity: 910

Merit: 1000

Items flashing here available at btctrinkets.com

Quote from: MXRider on September 04, 2013, 02:30:46 PM

Let's take the first chip as an example. Autotune is on (AIfDSo). It can be turned off by changing the capital A to a normal a (aIfDSo). Now the chip will always run at the speed you have set. Current speed is 55. Minimum is 52 and maximum is 57, higher the better of course. In this case I would raise the speed to 56 since there are practically no errors.

In my (limited) experience there is no sweet spot between 1 and 10+ errors. I've only been raising the speed on chips that constantly have 0 errors. This ofcourse is something that should be discussed more. Whats the efective hashing speed gain when raising one chip by 1 ? ..and subsequently whats the error persentage that justifies the gain ?

I would collect a sample on it myself, but am for now in the middle of trying to find a correlation between temperature, noncerate and errors.

MXRider

sr. member

Activity: 466

Merit: 250

I'm raising or lowering the speed of the chips, depending on the number of errors. I have also turned off the autotune. The point of manual tuning is that it works a little bit better than autotune.

Let's take the first chip as an example. Autotune is on (AIfDSo). It can be turned off by changing the capital A to a normal a (aIfDSo). Now the chip will always run at the speed you have set. Current speed is 55. Minimum is 52 and maximum is 57, higher the better of course. In this case I would raise the speed to 56 since there are practically no errors.

Chip two: change the "aifdso" to "aIfDSo" and let's see how it does (it won't work at all with aifdso).

Chip three: Something wrong with it. Turn off the autotune and keep the speed at 52. At least this is what I would do because 0,6 GH/s is still better than 0.

More info here http://punin.com/README_BFSB_v0.1.txt

Micky25

legendary

Activity: 974

Merit: 1000

my stuff runs steady at 65 GH/s. 1 M and 3 H-boards. Pool at diff. 16. No direct air flow. Check proper seating of the hashing boards, this seems error-prone.

Topic: [ANN] Bitfury ASIC sales in EU and Europe - page 74. (Read 250482 times)