| It's here | |
|---|---|
| Tweet Topic Started: Oct 28 2013, 09:13 AM (295 Views) | |
| Corather | Oct 28 2013, 09:13 AM Post #1 |
|
Administrator
|
The unit arrived about 20 minutes ago. The outside box was in good shape. Upon opening, the unit was intact with no dents, dings or scratches. The fan on the heatsink has fallen off, so I'll have to reinstall that. I have to go to work now, but I'll be back in a couple of hours to work on it. Hopefully after reinstalling the fan and a second visual check I'll have her up and running by mid afternoon. ![]() Pictures will follow shortly. Edited by Corather, Oct 28 2013, 09:13 AM.
|
![]() |
|
| tripppn | Oct 28 2013, 12:40 PM Post #2 |
|
It's a good day!!! |
![]() |
|
| tonychau7 | Oct 28 2013, 06:39 PM Post #3 |
|
Best news today! |
![]() |
|
| Corather | Oct 28 2013, 09:25 PM Post #4 |
|
Administrator
|
I have bad news. Looks like the ASIC board is toast. It would run for a few minutes, get some odd errors and then quit. I left it on for 6 hours and now it won't submit any shares and the temperature doesn't rise at all. It never went higher than 50c when it did work which wasn't for very long. I've submitted a ticket. I don't know or if we'll ever ROI on this unit. I'm very disappointed.
|
![]() |
|
| Gump | Oct 28 2013, 10:08 PM Post #5 |
|
That may be, but 50c is about right. Lets try some stuff first, I want to see what's happening. What are the odd errors? We have $ in this and we don't want to lose that easily. Also, KNC is completely unresponsive to emails and they aren't answering the phone. They did say they have a firmware update coming out tomorrow (hopefully) that is supposed to address the problem with all three(!) of my jupiters, so we'll see. So, try to SSH to the box (using Putty) and run it manually. Lets see what the errors even are. Here is the command that the gui starts. 3447 3446 root S 83148 16% 61% /usr/bin/cgminer --default-config /config/cgminer.conf First stop it: '/etc/init.d/cgminer.sh stop' Then restart with this, which will output to the screen: /usr/bin/cgminer --default-config /config/cgminer.conf Also, try re-flashing the firmware. Start with the version it's running now. You can download from KNC. There is a process that monitors the dies (/sbin/monitordcdc). When bad voltage readings are seen, it restarts cgminer: /etc/init.d/cgminer restart so your process may get killed but you can restart. Edited by Gump, Oct 28 2013, 10:10 PM.
|
![]() |
|
| Gump | Oct 28 2013, 10:12 PM Post #6 |
|
Also, what did it hash at when running? Which pool? I think we would like some visibility there :-) |
![]() |
|
| Corather | Oct 28 2013, 10:16 PM Post #7 |
|
Administrator
|
Do you have ventrilo gump? 66.177.134.124 port 3784 Lets try and get this thing going. |
![]() |
|
| Corather | Oct 28 2013, 10:17 PM Post #8 |
|
Administrator
|
It was doing around 120-130 gh when it was hashing. I upgraded to .97 and was having some troubles, so I downgraded and that's when it went to hell. I think I need to try enablecores.bin? I tried it on bitminter as nothing else worked well. same user name. Edited by Corather, Oct 28 2013, 10:17 PM.
|
![]() |
|
| Corather | Oct 28 2013, 10:26 PM Post #9 |
|
Administrator
|
DC/DC 4:0 Too Low current is what I get. |
![]() |
|
| Corather | Oct 28 2013, 10:28 PM Post #10 |
|
Administrator
|
And how do I upload bertmod? Normal upgrade feature? |
![]() |
|
| Gump | Oct 28 2013, 10:43 PM Post #11 |
|
I should have warned against the upgrade :-) This should hash at 200+ i think? The .98 firmware is supposed to address this KNC fiasco. Yes, bertmod is a normal upgrade, but DON'T restart. Restart undoes the 'upgrade'. Bertmod behaves kind of funky on the browser so clear your browser cache and cookies and log on with: <ip>/index.html Then you can see your cores enabled/disabled and you'll know if you need enablecores. *EDIT* I have a commitment in a few mins and I can't troubleshoot now. I downloaded ventrilo...are you free any time tomorrow? Edited by Gump, Oct 28 2013, 10:52 PM.
|
![]() |
|
| Gump | Oct 28 2013, 10:49 PM Post #12 |
|
Yes, that's in one of the scripts. It checks current then disables dies. I get the same thing on all 3 of mine. I wish I'd cancelled and bought btc in August, I'd be up 10K right now. But...there are other coins to mine too ;-) |
![]() |
|
| Corather | Oct 28 2013, 10:52 PM Post #13 |
|
Administrator
|
I think I do need enablecores because it was working before I flashed and now it's not. Not that it was working well, but it did seem to. Where do I find that file? |
![]() |
|
| Corather | Oct 28 2013, 10:52 PM Post #14 |
|
Administrator
|
I am free tomorrow. |
![]() |
|
| tripppn | Oct 29 2013, 12:01 AM Post #15 |
|
Oh No!!!
|
![]() |
|
| tonychau7 | Oct 29 2013, 12:12 AM Post #16 |
|
So was that a software or hardware issue? |
![]() |
|
| Corather | Oct 29 2013, 12:48 AM Post #17 |
|
Administrator
|
Well I got it working somewhat with kev's help. But it seems I'm experiencing the same issues many others are having. It appears to be a software issue because the asic starts off working and then gradually drops off to a cascade of non submitted shares. Kev suggests I leave it run like that, I'm inclined to agree, but I'm still trying to coax this thing into life. Bah. |
![]() |
|
| Corather | Oct 29 2013, 01:01 AM Post #18 |
|
Administrator
|
HOLY GUACAMOLE! IT WORKS! IT WORKS! IT WORKS! On a whim I tried firmware .95, and IT WORKS! IT WORKS!
|
![]() |
|
| Corather | Oct 29 2013, 01:04 AM Post #19 |
|
Administrator
|
Am I dreaming? Someone pinch me. |
![]() |
|
| Gump | Oct 29 2013, 10:30 AM Post #20 |
|
Consider yourself pinched. :-) So, what are we getting at the pool? Can you post a screenshot? Also, KNC released .98 firmware this morning. It didn't help any of my Jupiters so far. I'm letting it run for several hours to make sure. Honestly, if you get the advertised rate, I'd leave the firmware at .95. The cgminer 3.4 output lies, btw, by counting hardware errors in the total hash rate. Firmware .98/cgminer 3.6.6 corrects this so it's more accurate. |
![]() |
|
| Corather | Oct 29 2013, 01:31 PM Post #21 |
|
Administrator
|
It's mixed. I am mining on two pools with load balancing between Eligius and BitMinter. The results look good. I'll post everything later on tonight, I'm in between work and some paperwork I have to get done here. I will say it has an annoying problem. The miner will end cgminer for whatever reason and then startup cgminer again. It kills hash rate a little but the miner has gone up to 140 gh/s for a few moments sometimes. It's usually around 100-130 gh/s. Very few hardware errors also, and yes cgminer may be lying about that but it looks pretty stable aside from the restart issue. |
![]() |
|
| Corather | Oct 29 2013, 01:35 PM Post #22 |
|
Administrator
|
And you saved us again Gump. Thank you. I think you're the new tech expert on the team.So we have tripppn in charge of group buys, gump in charge of tech support, and tony is our oversees link. ![]() And me? I'm just a low paid public servant.
|
![]() |
|
| Corather | Oct 29 2013, 01:37 PM Post #23 |
|
Administrator
|
Have you tried firmware .95 on your Jupiters? It's a bit sparse on features but if it gets the unit to run man it's a winner. |
![]() |
|
| Corather | Oct 29 2013, 01:49 PM Post #24 |
|
Administrator
|
Is there a way to just run the new cgminer without the new firmware? That would probably work out okay. |
![]() |
|
| Gump | Oct 29 2013, 01:52 PM Post #25 |
|
The restart is part of /sbin/monitordcdc service. It grabs the temps from the dies and saves in /var/run/stats.knc and also restarts cgminer if there are any cores that are turned off. In .98, they just removed the restart and added a function that disables cores after 10 HW errors and re-enables them after 30 secs or so. There may be other enhancements, but .98 just has the effect of hiding hardware problems, which is BS. This 577gh number is bs too. look at the hardware errors. I'm barely touching 400gh at the pool. It beats 275, but I still think I have faulty hardware. KnC 0: | 577.2G/423.8Gh/s | A:680940 R:6206 HW:18499 WU: 5910.9/ Miner1 [2013-10-29 18:47:12] KnC: core 0-130 was disabled due to 10 HW errors in a row [2013-10-29 18:47:27] KnC: core 0-124 was disabled due to 10 HW errors in a row miner2 [2013-10-29 18:48:31] KnC: core 3-144 was disabled due to 10 HW errors in a row [2013-10-29 18:48:44] KnC: core 3-180 was disabled due to 10 HW errors in a row |
![]() |
|
| Gump | Oct 29 2013, 01:56 PM Post #26 |
|
It came with .95, which gave me ~275GH. It was so low, in part, because of the constant restarts, I think. I believe you'll get better results with .98. |
![]() |
|
| Gump | Oct 29 2013, 02:01 PM Post #27 |
|
The new cgminer (3.6.6) didn't help me. The change in the monitor script that removed the restart function is helping I think. Maybe it's a combination. I'm sure there are other considerations, I'm just learning how it works by reading the scripts running in mem and following other scripts they call when run. |
![]() |
|
| Corather | Oct 29 2013, 02:19 PM Post #28 |
|
Administrator
|
Asics tend to have high HW error rates. I could try upgrading to .98 just afraid it's going to "break" again. |
![]() |
|
| Corather | Oct 29 2013, 02:29 PM Post #29 |
|
Administrator
|
Alright, I held my breath and jumped in. 0.98 does work on this Mercury. And it does seem to be more stable also. The temp never rises above 52.5c which tells me the heatsink is doing it's job. That's actually pretty cool for what it's doing. My 7950 GPU's run at 71c undervolted and that's cool for them, haha. |
![]() |
|
| Corather | Oct 29 2013, 02:33 PM Post #30 |
|
Administrator
|
And hey look. ORSoC made the ASICS. KnC's hardware actually works very well. And the software fix is the best thing KnC can do to serve their clients. If it can be even hobbled together with software to coax under performing hardware along, that's huge. I know I wouldn't be so chipper if this miner wasn't working at all, and I know you're not happy either Gump, but it is what it is, and it is an improvement. I actually have to take my hat off to those fellas. They're fixing a worldwide issue with the power of their software genius. Yeah working hardware is nicer but shit, when life hands you lemons...
|
![]() |
|
| Corather | Oct 29 2013, 02:38 PM Post #31 |
|
Administrator
|
Oops, Asic is now up to 54.5c and holding. Probably it's maximum. That still seems about right to me. |
![]() |
|
| Gump | Oct 29 2013, 03:13 PM Post #32 |
|
An error rate around 5%… certainly less than 10%...is not unexpected. 20% in the first hour? I don't think so. There are many folks out there getting 2% or so, which is what their fully functional hardware is capable of. Heck, one guy received his miner with a bunch of water in it. After drying it out he is getting 550 to 600. That's true, KNC did not make the boards, but I didn't purchase from ORSoc, I purchased from KNC, who didn't test at all (apparently) before sending out a bunch of lemons. Bottom line is I didn't purchase $20,000 worth of lemons. Basically, I have 50% of a Jupiter given that two of the dies on all 3 aren't doing anything at all. I mean, who would be happy with these results with any other product? No one, I think, so my displeasure should not be unexpected. ;-) In the end, they developed a great miner, they just let a bunch of folks down with faulty hardware. I'll be happy when I consistently hit (exceed) the 500+ mark, like a bunch of other folks out there (with good hardware), which is what they were advertising. ... I'm still working with them to fix it. Then I will either try and sell the miners on eBay or mine BTC until it's no longer profitable, which may not be too long, and then mine different coins. I haven't decided yet Edited by Gump, Oct 29 2013, 03:18 PM.
|
![]() |
|
| Gump | Oct 29 2013, 03:16 PM Post #33 |
|
Max operating temperature on these is right about 120c. I see several people reporting that lower temperatures give higher rates while others saying higher temperatures do. I guess it all depends on your hardware's "personality." |
![]() |
|
| Corather | Oct 29 2013, 03:37 PM Post #34 |
|
Administrator
|
Oh I absolutely agree. I emailed KnC last night about my apparent hardware issues and they sent me an email today. I know I have hardware issues with this unit. It's doing around 106 gh/s average which is advertised but still low according to their maximum. On burst it reports 140 gh/s. But I know I have a few cores that aren't happy. I do hope KnC comes through for all of us, and especially everyone like yourself who's had to deal with critical hardware problems. Keeping the optimism, I know they'll make it right. And now I almost want to buy a Saturn upgrade for this Mercury. Hah!
|
![]() |
|
| 1 user reading this topic (1 Guest and 0 Anonymous) | |
| « Previous Topic · Announcements & News · Next Topic » |







3:43 AM Jul 11