Upgrade 0.94 has been released.
I'm getting way more hardware errors with that, the miner I update was steadily hashing at 525GH/s (+3 hours average) and its now struggling to keep up to 500GH/s (just a few minutes hashing thou). I'd not recommend upgrading if your miners are running fine. You know how we say, "if ain't broke..." What they added to the webui is the temps, pretty scary stuff, ASIC slots preaking at 70 Celsius thanks for sharing Rampion. edit: are you planning to downgrade the fw was installed before? |
|
|
|
Upgrade 0.94 has been released.
again no changelog, thou  well you brave enough to pop it in yet and if so...whats the info on power use etc etc....(jeez I don't even have my Jupiter yet and still i'm curious) as soon as I'll arrive to my colo, I'll test it and for sure I'll share my findings but don't hold your breath thou, I have to work for a living so I'll not be able to test till late afternoon |
|
|
|
KnC just released 0.94  |
|
|
|
Upgrade 0.94 has been released.
again no changelog, thou |
|
|
|
I've posted this on KNC forum but I put it here to. It's about the FPGA issue
I received my jupiter yesterday. I've upgraded the firmware abd had the same problem. BUT I think it's not a faulty card. I have 6 connenction for the hashing cards at the mainboard. I spent all night (from Europe) to figure out what is the problem. If I connceted only 1 hashing card for example slot 1 it made 5Ghs but if I connected to slot 2, it went up for 140Ghs. All night I was playing with the variations of the cards and the slots and in my opinion this problem is not came from the hashing card itself. I think it comes from wrong fpga routing or something like that. I could make a variation with the cards what made 440Ghs stable and without this issue. Now I left on a variation that makes about 530Ghs but restarts every 5-10 minutes. I had to give up cause it was 3 o'clock in the night.
I've tried avery single hashing card and they ar works fine but it depending on wich slot.
If I plug in only 1 or 2 hashing cards the HW error about 5-7%, but if I plug in all four it runs up to 20%.
Today I will continue the playing and waiting for the new firmware.....
keep up the good work bro! and don't forget to keep us posted.
Sorry for my poor English but I hope it helps.
don't worry mine is worst than yours but I post as long as I think I've something worth sharing |
|
|
|
My Jupiter won't run the .92 or .93 firmware it just causes CGminer to crash over and over. I've found .91 to be the most stable and fast avg. 460ghs@535 watts and guys get a meter so you can see when you are hashing, It's been a life saver. Using 50btc.com pool
this is precisely what's happening to me. stick with 0.91 cause .92/.93 cause cgminer to restart very frequently |
|
|
|
thanks really appreciated |
|
|
|
you know that? it just happened to me on one go my jups. with fw 0.91. the strange thing is that this miner was running for almost 4 days without problem.
restarting cgminer didn't help. i'll try to do hard reset as soon as someone will do it for me at the colo.
And the FPGA Error once again. I can tell that this thing is struggling with the lack of adequate airflow now. However it is not as bad as the first firmware v0.9.0 i already took the cover off and it doesnt help i have a full screen accepted by FPGA 21 works, but onlt 0/2/5/13 submitted it's so much fun i dont even think it has half hot as my 7970 rigs doing a cold reboot (shutdown -h now, switch psu off, wait 30s, swtich psu on again) fix the problem for me, for now. anyway I think you could be right about lack of adequate airflow. probably just putting a heatsink on top of the FPGA shiuld be enough... who knows. the problem for me is magnified because I'm monitoring the miners remotely and I can't send someone at will to tweak phisically the machine. regarding the fans, I can't remember where I saw it but someone posted a sort o minu-guide on how to check if your internal fans are correctly mounted as soon as I find it I'll post it here edit: found out (thanks Searing) http://forum.kncminer.com/forum/main-category/hardware/5640-verify-correct-fan-installation |
|
|
|
cgminer keep show me accepted by FPGA 21 works, but onlt 0 submitted
it casue only 10% accept on pool
firstly which firmware are you running? secondlly join kncminer channel at irc.freeenode.net... you'll find a lot of skilled people there. as last thing, look at this thread https://bitcointalk.org/index.php?topic=306969.0particulary the 5th page it happen on 91 92 93 firmware running on 0.90 with only 280ghs atm i will read your post now , thanks for that mate you know that? it just happened to me on one go my jups. with fw 0.91. the strange thing is that this miner was running for almost 4 days without problem. restarting cgminer didn't help. i'll try to do hard reset as soon as someone will do it for me at the colo. |
|
|
|
cgminer keep show me accepted by FPGA 21 works, but onlt 0 submitted
it casue only 10% accept on pool
firstly which firmware are you running? secondlly join kncminer channel at irc.freeenode.net... you'll find a lot of skilled people there. as last thing, look at this thread https://bitcointalk.org/index.php?topic=306969.0particulary the 5th page |
|
|
|
|
Kudos Sitarow. Very well done report!
Could please share type/brand of the fan you put on top of the jup?
|
|
|
|
@Sitarow any news, did you replace the front fans with the avalon type you mentioned before?
Since I had upgraded to v0.93 from v0.92 Yesterday, I have chosen to do a baseline on performance prior to swapping out the fans and putting the case cover back on. Once 0.94 is released I will be swapping out the fans as I am sure it will be a far more solid build. fair I hope to be receiving my last shipment of hardware some time this week. So in order for me to be ready I need these issues all ironed out with a solid solution going forward.
The cooling with just a single fan pointing downwards with the case open has proven that performance does increase over the default factory cooling methods.
you're talking about this configuration, right?:  I seem to rembember that DeadTerra on irc today was saying something about problem he had with this PSU and that he was heading to KnC office with a faulty one to make them evaluate it. |
|
|
|
Well that is the risk you take on a personal level. However given my level of technical experience and background with other ASIC hardware. KnCMiner would know better then dispute any warranty claims made by me. Adding adequate cooling over and above what was given from the factory would not be the cause of overheating components. Especially if like me you have access to tools that make it crystal clear of potential issues with thermal airflow and heat. Edit: I wanted to include the instructions from KnCMiner on how to troubleshoot your unit for potential defective module. To verify if an ASIC board is broken or the performance drop is due to software issues please follow these steps:
1. Turn off and disconnect your PSU.
2. Open your miner case.
3. Unplug the flat ribben cable connecting one of the ASIC Boards (the circuit board that has the big fan on it) to the controller board.
4. Unplug the PCI-Express power cable from the ASIC board.
5. Ensure all other cables are connected properly.
6. Reconnect the PSU and power on the miner. Let the miner work for an hour to asses performance.
If the miner is now performing as expected you should install the 0.9.4 firmware.
7. Turn off the power and disconnect the PSU, reconnect all the PCI-E and ribben cables.
8. Restart the miner and install firmware 0.9.4.source https://www.kncminer.com/pages/supportthanks for pointing out this section of kncminer.com I wasn't aware of. Sure those guys have to improve their communication skills |
|
|
|
Update on the 0.93 cgminer restarts (Saturn):
I have the Saturn miner running with Bitminter as primary pool. Minimum difficulty had been
manually set (on the pool web interface) to 128.
With this setup, the miner and the pool often switch up to difficulty 256 and back to 128.
Today I reconfigured the Bitminter pool for a minimum difficulty of 256.
Since then, I did not see any cgminer restart. cgminer is now running for more than 8h.
In the past the max. I could observe had been 3h.
trepex thanks for sharing this info. Did you disclose your finding with KnC, it could be a useful info for them. did you notice any an increase in performance ? I'm running my jups with slush pool. slush has dynamic mechanism to set diff (Vardiff) based on the hash rate of your boxes, and despite the fact they mint almost at the same rate, one is setted to 395 and the other to 453. the former has higher hw errors ratio, thou (4.1% vs 3.1%). if KnC's new firmware will behave as the last two i'll try to switch to a pool that let me set diff manually and i'll try higher values (something like 512) thanks again |
|
|
|
You can setup backup pools the same way you do normally in CGminer.
There's no GUI for the OS (angstrom linux) in this case so no teamviewer type app. SSH is the way to go.
Has anyone tried cgwatcher ? https://bitcointalk.org/index.php?topic=185553.0Does it work remotely? I never tried it, but it says it does. The "CGRemote" at least. After a quick glance, It seems cgremote is a different program developed by the same person behind cgwatcher. The prob is that i don't have any windows workstation. So no way to use it. |
|
|
|
What is the power consumption for those of you who have received miners?
With 0.90 on my jup i got almost 850w. With 0.91 a little lower then 500w. |
|
|
|
Had you bridge the pins of 24pin ATX of psu? (Specified in page 1 of the manual)
thanks
I did it. It works. Search for "paper clip trick" on google you'll get plenty of resources. |
|
|
|
You can setup backup pools the same way you do normally in CGminer.
There's no GUI for the OS (angstrom linux) in this case so no teamviewer type app. SSH is the way to go.
Has anyone tried cgwatcher ? https://bitcointalk.org/index.php?topic=185553.0Does it work remotely? |
|
|
|
To the folks that are having issues with the .93 and .92 releases.. how stable is your internet connection? I have found that any hickups in internet connectivity causes a CGminer reboot.
mmm it seems quite stable to me but I didn't monitored closely. the only thing I could says is that cgminer restart once every 30-60s with both 0.92/93. If 0.94 won't solve the issue at hand I will monitor colo internet connection properly. so cgminer restarts and not just single cores?? I'm not sure cause screen kicks me out, i barely see something on the screen. Next time i will record the session. |
|
|
|
|
Show Posts
