ChipGeek
|
|
December 14, 2013, 03:23:45 PM |
|
BFL did have problems with engine 0 making the miner unstable,
Can you disable engine 0 within the software? and then test without engine 0 running on each chip?
This was only true for the first revision of chips. All of those were shipped in BFL products except a few early sample chips. All of the other chips sent out to us were the second revision where engine 0 is working.
|
Tip jar: 1ChipGeeK7PDxaAWG4VgsTi31SfJ6peKHw
|
|
|
|
|
|
|
|
"If you don't want people to know you're a scumbag then don't be a scumbag." -- margaritahuyan
|
|
|
Advertised sites are not endorsed by the Bitcoin Forum. They may be unsafe, untrustworthy, or illegal in your jurisdiction.
|
Lucko (OP)
|
|
December 15, 2013, 12:40:54 AM |
|
OK today/yesterday(since it is 1:35am) progress report. We getting sure it is a problem with 1V power supply. Problem is that interface for checking it doesn't work on any. MrTeal did look at used components and didn't find anything wrong at the glance but will go over line by line tomorrow. I also got some additional documentation on the board(thanks MrTeal) so I can do batter job debugging it... It might be bad batch of components...
|
|
|
|
Lucko (OP)
|
|
December 15, 2013, 05:10:34 PM |
|
Figure out the interface problem...
|
|
|
|
GrapeApe
|
|
December 15, 2013, 05:32:58 PM |
|
Figure out the interface problem...
Oh man you're going to have to share more than that. What was the problem and are they hashing now? If so what are the numbers? Oh and glad to hear it!!! (didn't mean to sound ungrateful).
|
|
|
|
Lucko (OP)
|
|
December 15, 2013, 06:22:46 PM |
|
Figure out the interface problem...
Oh man you're going to have to share more than that. What was the problem and are they hashing now? If so what are the numbers? Oh and glad to hear it!!! (didn't mean to sound ungrateful). OK one post up... OK today/yesterday(since it is 1:35am) progress report. We getting sure it is a problem with 1V power supply. Problem is that interface for checking it doesn't work on any. MrTeal did look at used components and didn't find anything wrong at the glance but will go over line by line tomorrow. I also got some additional documentation on the board(thanks MrTeal) so I can do batter job debugging it... It might be bad batch of components...
I can see errors in 1V power supply... So MrTeal will probably be able to tell me what is happening... We are not hashing any way different then they were... They run to some speed and then reboots
|
|
|
|
GrapeApe
|
|
December 15, 2013, 06:29:33 PM |
|
Yep again I'm jumping to conclusions without reading everything. Thanks for clearing it up, I'm kinda embarrassed after that one.
|
|
|
|
Lucko (OP)
|
|
December 15, 2013, 06:35:07 PM |
|
Yep again I'm jumping to conclusions without reading everything. Thanks for clearing it up, I'm kinda embarrassed after that one.
No problem it happens... They do hash to some point. I can tell you that if I don't count one with only 60% of engines working they all go over 35GH. One is over 41. Just not long... It has all engines working... So yes board is realy good one. Just need to figure out what is the reason it resets...
|
|
|
|
asjfdlksfd
|
|
December 15, 2013, 07:12:46 PM |
|
Hmm,
do you mean you can send the boards which hash > 30 GH/s allready? It looks like these are ok for now and every day is a loss in 10 € a day with actual diff. (not to forget the lost allready we've got for the time). I don't need an immediate fix for more hashing rate for an board which can be throw in a trash bin in a half year but is running for now . :-(
Cheers...
|
|
|
|
one4many
|
|
December 15, 2013, 07:26:03 PM |
|
Hmm,
do you mean you can send the boards which hash > 30 GH/s allready? It looks like these are ok for now and every day is a loss in 10 € a day with actual diff. (not to forget the lost allready we've got for the time). I don't need an immediate fix for more hashing rate for an board which can be throw in a trash bin in a half year but is running for now . :-(
Cheers...
+1 IF, it is fixable via software ... otherwise please fix the hardware. Get Mr. Teal a little more involved, he designed it after all. Cheers one4many
|
|
|
|
Lucko (OP)
|
|
December 15, 2013, 07:35:36 PM |
|
Hmm,
do you mean you can send the boards which hash > 30 GH/s allready? It looks like these are ok for now and every day is a loss in 10 € a day with actual diff. (not to forget the lost allready we've got for the time). I don't need an immediate fix for more hashing rate for an board which can be throw in a trash bin in a half year but is running for now . :-(
Cheers...
+1 IF, it is fixable via software ... otherwise please fix the hardware. Get Mr. Teal a little more involved, he designed it after all. Cheers one4many It is one board not boards... I don't know if it is hardware or software... I have MrTeal as involved as he has time allows... He is looking what I did what parts did I put on and I updating him on everything... He is as puzzled as I'm on this problem. I also have 15 boards waiting so I fill your pain...
|
|
|
|
dwdoc
Legendary
Offline
Activity: 966
Merit: 1000
- - -Caveat Aleo- - -
|
|
December 15, 2013, 07:48:38 PM |
|
Appreciate your hard work. Hope you figure it out soon.
|
|
|
|
bx8389
Member
Offline
Activity: 80
Merit: 10
|
|
December 15, 2013, 11:22:20 PM |
|
Hmm,
do you mean you can send the boards which hash > 30 GH/s allready? It looks like these are ok for now and every day is a loss in 10 € a day with actual diff. (not to forget the lost allready we've got for the time). I don't need an immediate fix for more hashing rate for an board which can be throw in a trash bin in a half year but is running for now . :-(
Cheers...
+1 IF, it is fixable via software ... otherwise please fix the hardware. Get Mr. Teal a little more involved, he designed it after all. Che one4many It is one board not boards... I don't know if it is hardware or software... I have MrTeal as involved as he has time allows... He is looking what I did what parts did I put on and I updating him on everything... He is as puzzled as I'm on this problem. I also have 15 boards waiting so I fill your pain... If you have pinpointed problem to 1V power supply i think real problem should be identified in next 1-2 days: probably is better to wait before take decisions to send boards "as-is" or "fixed". Question may be stupid but have you checked if 1 V power supply is really 1v ? ( don't rely only on value read by controller ) power is also clean and constant? Regards
|
|
|
|
MrTeal
Legendary
Offline
Activity: 1274
Merit: 1004
|
|
December 16, 2013, 04:25:13 AM |
|
Lucko and I have been trying to work through the errors that he is getting and looking for ways to solve them. I also believe Lucko is going to send me a board to debug so that I can have the same hardware in hand that he is looking at to speed things along.
|
|
|
|
dwdoc
Legendary
Offline
Activity: 966
Merit: 1000
- - -Caveat Aleo- - -
|
|
December 16, 2013, 04:36:42 AM |
|
MrTeal thanks for all your help!
|
|
|
|
Bogart
Legendary
Offline
Activity: 966
Merit: 1000
|
|
December 16, 2013, 04:37:56 AM |
|
Yep again I'm jumping to conclusions without reading everything. Thanks for clearing it up, I'm kinda embarrassed after that one.
No problem it happens... They do hash to some point. I can tell you that if I don't count one with only 60% of engines working they all go over 35GH. One is over 41. Just not long... It has all engines working... So yes board is realy good one. Just need to figure out what is the reason it resets... So the strongest board is the one that resets? I imagine there's a correlation. Like the board is running so hard it overwhelms some supporting piece. I know the Avalon boards had an issue with voltage drop due to the PCB copper being too thin. I wonder if something similar is happening here.
|
"All safe deposit boxes in banks or financial institutions have been sealed... and may only be opened in the presence of an agent of the I.R.S." - President F.D. Roosevelt, 1933
|
|
|
Lucko (OP)
|
|
December 16, 2013, 01:46:29 PM |
|
So the strongest board is the one that resets?
I imagine there's a correlation. Like the board is running so hard it overwhelms some supporting piece.
I know the Avalon boards had an issue with voltage drop due to the PCB copper being too thin. I wonder if something similar is happening here.
I don't think it is as easy as that. It is a speed and voltage combination. If this is off the problem will not appear... So if you stop CGminer and voltage will not drop to 0,85V(or below crash voltage for board not sure) and restart CGminer it the board will not crash. Board made by facility did have 25% less copper on outer layers. But we did boards with found PCB to and same problem... But yes I agree to a point. One normally working board is probably below problematic hashing/voltage speed. So there is probably a strongest link... EDIT: If MrTeal agrees I can also give you tracking data for board send to him...
|
|
|
|
Lucko (OP)
|
|
December 16, 2013, 05:46:07 PM |
|
OK found something but not sure what to think... But it might be that I have a batch of defective components... Will need to look at all made boards... This is one of a better running boards so... https://www.dropbox.com/s/wvxuvls6502pdov/temps.jpgProblem is that this was detected on this one but not on other we looked at... Tomorrow will go over all. It was late and I was asked to live friend lab...
|
|
|
|
Lucko (OP)
|
|
December 17, 2013, 10:59:22 PM |
|
Unfortunately only one board do this... And when running at full board get much hotter...
I did notice that marking on one of the components is a bit different then on a sample board but component datasheet say it is only a date code...
|
|
|
|
Swimmer63
Legendary
Offline
Activity: 1593
Merit: 1004
|
|
December 18, 2013, 12:07:39 AM |
|
Unfortunately only one board do this... And when running at full board get much hotter...
I did notice that marking on one of the components is a bit different then on a sample board but component datasheet say it is only a date code...
So are you saying we still don't know what is wrong with the boards? Has Mr. Teal made any progress?
|
|
|
|
Lucko (OP)
|
|
December 18, 2013, 06:51:43 PM |
|
Unfortunately only one board do this... And when running at full board get much hotter...
I did notice that marking on one of the components is a bit different then on a sample board but component datasheet say it is only a date code...
So are you saying we still don't know what is wrong with the boards? Has Mr. Teal made any progress? No. We will see what will happen when MrTeal get his board...
|
|
|
|
|