Topic: BFL board project COINTAMINATION - EU facility - ORDERS OPENED for Chili - page 17. (Read 73149 times)
December 15, 2013, 02:48:38 PM
Appreciate your hard work. Hope you figure it out soon.
December 15, 2013, 02:35:36 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...
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
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...
December 15, 2013, 02:26:03 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...
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
December 15, 2013, 02:12:46 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...
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...
December 15, 2013, 01:35:07 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...
December 15, 2013, 01:29:33 PM
Yep again I'm jumping to conclusions without reading everything. Thanks for clearing it up, I'm kinda embarrassed after that one.
December 15, 2013, 01:22:46 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 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 December 15, 2013, 12:32:58 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).
December 15, 2013, 12:10:34 PM
Figure out the interface problem...
December 14, 2013, 07:40:54 PM
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...
December 14, 2013, 10:23:45 AM
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. Can you disable engine 0 within the software? and then test without engine 0 running on each chip?
December 14, 2013, 09:48:19 AM
I recall my lot of chips being divided into 2 groups of "A" and "B", with the laser etchings bearing different numbers. I think A was week 38 and B was week 37, or something like that. I wonder if they behave differently.
December 14, 2013, 09:22:50 AM
Any new firmware for the chili board
.?
.?
December 14, 2013, 07:47:35 AM
No. I don't have source code for firmware...
December 14, 2013, 07:37:44 AM
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?
Can you disable engine 0 within the software? and then test without engine 0 running on each chip?
December 14, 2013, 07:29:17 AM
1 board 23GH no problems... Only 70 engines.
The rest of the boards are over 118 engines...
5 boards looks to be 30+ but can't get them to hash for longer then minute to get to speed...
2 boards 36 and 38(with additional cooling 41 and 39) but they don't run every time for cold start... But if I stop mining software and restart is fast enough(so that the voltage is still over 1V) board will start every time...
In most cases if problems starts at 24 to 29GH. Only one board crashes at 38GH.
Now I can't figure it out why boards run for some time then restarts. I would say that it is put together right but no idea why it then restart and they restart even USB so they get disconnected from computer.
The rest of the boards are over 118 engines...
5 boards looks to be 30+ but can't get them to hash for longer then minute to get to speed...
2 boards 36 and 38(with additional cooling 41 and 39) but they don't run every time for cold start... But if I stop mining software and restart is fast enough(so that the voltage is still over 1V) board will start every time...
In most cases if problems starts at 24 to 29GH. Only one board crashes at 38GH.
Now I can't figure it out why boards run for some time then restarts. I would say that it is put together right but no idea why it then restart and they restart even USB so they get disconnected from computer.
December 14, 2013, 06:39:23 AM
Hm, how much pcbs you now have allready tested?
Maybe they are only spikes in the whole order and so are really lemons
?
How much hashing engines are activated after some hashing?
Do you tried to use bfgminer instead of cgminer?
I've here a new Thermaltake 730W PSU which looks like give also "not so ideal" hashing results with my both boards which have hashing problems. They have also problems with my new Enermax Revo PSU, but have little bit less hw errors so they hash ~ 1 GH/s better than with the Thermaltake. Connected to my good boards I cannot see a difference.
But if I read some posts in Mr. Teals forum it looks like that some have problem where most other didn't have. If hashing with 30 GH/s is possible without special cooling of the vrm the board should be ok.
So my last question is, how much boards never hashes or hashes really pool, arrive <25 GH/s, > 25 GH/s, >30 GH/s, >35 GH/s? Also how boards are affected and when will the boards which are not affected send to us?
Maybe they are only spikes in the whole order and so are really lemons
?How much hashing engines are activated after some hashing?
Do you tried to use bfgminer instead of cgminer?
I've here a new Thermaltake 730W PSU which looks like give also "not so ideal" hashing results with my both boards which have hashing problems. They have also problems with my new Enermax Revo PSU, but have little bit less hw errors so they hash ~ 1 GH/s better than with the Thermaltake. Connected to my good boards I cannot see a difference.
But if I read some posts in Mr. Teals forum it looks like that some have problem where most other didn't have. If hashing with 30 GH/s is possible without special cooling of the vrm the board should be ok.
So my last question is, how much boards never hashes or hashes really pool, arrive <25 GH/s, > 25 GH/s, >30 GH/s, >35 GH/s? Also how boards are affected and when will the boards which are not affected send to us?
December 13, 2013, 03:34:16 PM
The problem is that is not stable... Every Chili has voltage/hashing speed(they do go up in parallel) that is critical. And if it goes over it it will work but if not will crash... Using more then one is impossible since it crashes mining program...
I really can't find anything wrong with them... Today test showed noting strange. Nothing is to hot no spikes... I do need to make some tests with Digital Power Designer software(need to figure out how to use it) that control chip power supply... That will be done tomorrow...
And it is different problem with different firmware on Chili. If I use old one it start producing HW errors if I use new one it restarts...
One card that is working is probably below problematic hashing speed...
I really can't find anything wrong with them... Today test showed noting strange. Nothing is to hot no spikes... I do need to make some tests with Digital Power Designer software(need to figure out how to use it) that control chip power supply... That will be done tomorrow...
And it is different problem with different firmware on Chili. If I use old one it start producing HW errors if I use new one it restarts...
One card that is working is probably below problematic hashing speed...
December 13, 2013, 02:34:49 PM
Lucky boy,
my chilies are hashing with to less performance:
BAL 1 and BAJ0 are Jallies, BAL1 with cooling problem.
BAL6 and BAL2 are Chillies, also with cooling problem.
Rest looks ok for non good vrm cooled boards at ambient temperature of 30 °C.
Some of them has not a backplate which I will mount on weekend. Maybe than they will hashing faster.
What I'm missing is a way to
understand the which temps are shown in bfgminer
the detailed view of each board as like it does for BFL miners in bfgminer (processor like stats)
check the internal temperatures of each chip
shows the frequencies of each chip
...
Whishes over whishes...
Cheers...
my chilies are hashing with to less performance:
Code:
cgminer version 3.8.5 - Started: [2013-12-13 08:11:43]
--------------------------------------------------------------------------------
ALL (10s):162.7G (avg):162.6Gh/s | A:1579858 R:6000 HW:100679 WU:2137.2/m
ST: 10 SS: 39 NB: 83 LW: 1705533 GF: 1 RF: 1
...
Block: 2dce8686... Diff:908M Started: [20:21:53] Best share: 995K
--------------------------------------------------------------------------------
[P]ool management [S]ettings [D]isplay options [Q]uit
BAL 0: max 69C 1.16V | 27.92G/28.05Gh/s | A:277760 R:1168 HW: 4781 WU:385.4/m
BAL 1: max 43C 1.00V | 9.745G/9.694Gh/s | A: 46864 R: 192 HW:54790 WU: 64.1/m
BAL 2: max 69C 1.15V | 21.23G/21.42Gh/s | A:215920 R: 576 HW: 3540 WU:294.1/m
BAJ 0: max 49C 1.00V | 7.614G/7.752Gh/s | A: 82432 R: 464 HW: 399 WU:107.6/m
BAL 3: max 69C 1.16V | 25.02G/25.45Gh/s | A:261538 R: 896 HW: 6695 WU:345.3/m
BAL 4: max 67C 1.16V | 27.25G/27.23Gh/s | A:265600 R:1056 HW:18941 WU:354.5/m
BAL 5: max 69C 1.16V | 27.31G/27.39Gh/s | A:272928 R: 944 HW: 5337 WU:376.1/m
BAL 6: max 70C 1.02V | 15.60G/15.66Gh/s | A:156880 R: 704 HW: 6196 WU:210.0/m
--------------------------------------------------------------------------------
--------------------------------------------------------------------------------
ALL (10s):162.7G (avg):162.6Gh/s | A:1579858 R:6000 HW:100679 WU:2137.2/m
ST: 10 SS: 39 NB: 83 LW: 1705533 GF: 1 RF: 1
...
Block: 2dce8686... Diff:908M Started: [20:21:53] Best share: 995K
--------------------------------------------------------------------------------
[P]ool management [S]ettings [D]isplay options [Q]uit
BAL 0: max 69C 1.16V | 27.92G/28.05Gh/s | A:277760 R:1168 HW: 4781 WU:385.4/m
BAL 1: max 43C 1.00V | 9.745G/9.694Gh/s | A: 46864 R: 192 HW:54790 WU: 64.1/m
BAL 2: max 69C 1.15V | 21.23G/21.42Gh/s | A:215920 R: 576 HW: 3540 WU:294.1/m
BAJ 0: max 49C 1.00V | 7.614G/7.752Gh/s | A: 82432 R: 464 HW: 399 WU:107.6/m
BAL 3: max 69C 1.16V | 25.02G/25.45Gh/s | A:261538 R: 896 HW: 6695 WU:345.3/m
BAL 4: max 67C 1.16V | 27.25G/27.23Gh/s | A:265600 R:1056 HW:18941 WU:354.5/m
BAL 5: max 69C 1.16V | 27.31G/27.39Gh/s | A:272928 R: 944 HW: 5337 WU:376.1/m
BAL 6: max 70C 1.02V | 15.60G/15.66Gh/s | A:156880 R: 704 HW: 6196 WU:210.0/m
--------------------------------------------------------------------------------
BAL 1 and BAJ0 are Jallies, BAL1 with cooling problem.
BAL6 and BAL2 are Chillies, also with cooling problem.
Rest looks ok for non good vrm cooled boards at ambient temperature of 30 °C.
Some of them has not a backplate which I will mount on weekend. Maybe than they will hashing faster.
What I'm missing is a way to
understand the which temps are shown in bfgminer
the detailed view of each board as like it does for BFL miners in bfgminer (processor like stats)
check the internal temperatures of each chip
shows the frequencies of each chip
...
Whishes over whishes...
Cheers...
December 12, 2013, 05:07:00 PM
If anyone is interested how it looks...
https://www.dropbox.com/s/rn85e93my1a9gcc/2013-12-12%2013.39.24.mp4
With only a single LED flashing, it looks like maybe the MCU has rebooted. https://www.dropbox.com/s/rn85e93my1a9gcc/2013-12-12%2013.39.24.mp4
The LEDs are numbered 1 (closest to USB connector) to 8 (closest to the PCI-E power connector). Is it LED 7 that is blinking?
Jump to: