Topic: The Habanero - 650GH/s - OOS - page 4. (Read 96043 times)
No, still a work in progress. I need to make a good manifold on the radiator so I can pipe other stuff in there too. I cleaned out some of the tubing and added a good inlet filter this morning, regreased and rebalanced the waterblock. I fired it back up on full clock (850) and one of the cores was wigging out a bit more than expected, so I figured to restart on stepped reduced clocks until it evened out. Restart on 840 and one of the dies isn't hashing at all. Running cold and the hashrate is 3/4 expected. That's gonna be fun to diagnose.
[...]
~700LPH 12VDC pump
12x25" aluminum car radiator, 5-gallon bucket resorvoir
12VDC 6.5A 12" fan on one half and 6x 12V 0.68A 120mm fans on the other
[...]
~700LPH 12VDC pump
12x25" aluminum car radiator, 5-gallon bucket resorvoir
12VDC 6.5A 12" fan on one half and 6x 12V 0.68A 120mm fans on the other
[...]
Got any pics of that cooling setup?
How about overvolting? Any drawbacks for (extreme) example: 800@970 ?
Edit: Apart from temps and power draw?
I believe it's volts@freq, but the basic idea is fine. 970mV for 800MHz is extreme overkill though, I think I run ~940-950mV at 900MHz for most of mine.Edit: Apart from temps and power draw?
Correct for the HFTool. jaketri-cgminer is the other way around.
One thing I noticed with the new firmware: The hfa-temp-overheat switch in cgminer does not seem to work. I set it to 105, but the boards happily go to 107. Looks like the firmware overrides that switch in any case (as it did before).
Edit: Actually it seems "a limit" gets hit and the temp drops to 96 for example. The hashrate drops by about a quarter and the cycle starts again. There is no more "Temp failure" message in cgminer. The cycle goes for a few times and then the board restarts with the "HFB failure" message.
I finally got around to lighting up the Habanero I fetched secondhand from someone back in early August, and daggum I like that thing. Figured out there was something gumming up my waterblock at one point and it was overheating; after a full flush of the lines (including blowing out a big fat dead spider) and fetching new clean water, heck that thing's running cold. It was pushing 800MHz at 880mV and 80C when I left the shop. Darn sexy piece of hardware. I know the radiator I got set up is super overkill, it's just what I had (originally spec'd for a 3kw adjustable dummy load, should probably finish that thing sometime). Eventually gonna put together a manifold for it. Got another Hashfast board I need to figure out, and when those pesky technobit minion boards ever arrive I'll probably put them in the loop.
Raystorm waterblock (I forget the model, same as QG used I think)
Arctic Silver Ceramic thermal compound
~700LPH 12VDC pump
12x25" aluminum car radiator, 5-gallon bucket resorvoir
12VDC 6.5A 12" fan on one half and 6x 12V 0.68A 120mm fans on the other
Dell 750W PSU kit
Raystorm waterblock (I forget the model, same as QG used I think)
Arctic Silver Ceramic thermal compound
~700LPH 12VDC pump
12x25" aluminum car radiator, 5-gallon bucket resorvoir
12VDC 6.5A 12" fan on one half and 6x 12V 0.68A 120mm fans on the other
Dell 750W PSU kit
How about overvolting? Any drawbacks for (extreme) example: 800@970 ?
Edit: Apart from temps and power draw?
I believe it's volts@freq, but the basic idea is fine. 970mV for 800MHz is extreme overkill though, I think I run ~940-950mV at 900MHz for most of mine. Edit: Apart from temps and power draw?
How about overvolting? Any drawbacks for (extreme) example: 800@970 ?
Edit: Apart from temps and power draw?
Edit: Apart from temps and power draw?
I have the exact same problem with mine too. Any help would also be appreciated. It acts like it wants to start, all the lights light up and then it goes into that mode and keeps losing and regaining contact with the board, but never does start hashing and just goes into an endless loop. Used it on a Win7 computer and a Ubuntu PC, same results.
Unfortunately that error is a pretty generic one when it won't communicate, so it's hard to isolate just based on that.
Hi all,
one of my best boards died today and I'm clueless how to get it up and running again. I have tried to turn off board, removed usb and all that.
Would someone know what can be wrong ?
Thank you very much
one of my best boards died today and I'm clueless how to get it up and running again. I have tried to turn off board, removed usb and all that.
Would someone know what can be wrong ?
Thank you very much
I have the exact same problem with mine too. Any help would also be appreciated. It acts like it wants to start, all the lights light up and then it goes into that mode and keeps losing and regaining contact with the board, but never does start hashing and just goes into an endless loop. Used it on a Win7 computer and a Ubuntu PC, same results.
The new firmware revived my zombies, thank you very much!
I've run a board with a piece of cardboard over one fan for a couple days to get temps up over 105, and nothing untoward has happened. For those that are interested, this is the 110C firmware.
https://www.dropbox.com/s/w8n6u26qedns9lj/firmware-20141018.tar.gz?dl=0
https://www.dropbox.com/s/w8n6u26qedns9lj/firmware-20141018.tar.gz?dl=0
Thanks as always Mr. Teal.
Let's say we want to underclock/undervolt these...right now I'm pulling around 700W for 600GHs...but this includes the fans and pump...
Would voltage/clock settings like 800@800 cause any long term problems...or even allow enough volts for the clock? Anyone have any working numbers for down-clocking?
Awesome. Thank you.
I have my hab running at 787Mhz now. Set all dies to 880mV at 800MHz. Getting around 580GH/s now, with cgminer displaying 0.82V. Setting it at 800MH was starving the chip and would cause cgminer to crash sometimes....787MHz hasn't crashed it once.
For me, the best part about the lower settings are the lower fan speeds I can set now. I can go as low as 15 for the hfa-fan setting, and temps are 95C. Defiantly saving some watts and much more silent. Thanks Mr. Teal.
I've run a board with a piece of cardboard over one fan for a couple days to get temps up over 105, and nothing untoward has happened. For those that are interested, this is the 110C firmware.
https://www.dropbox.com/s/w8n6u26qedns9lj/firmware-20141018.tar.gz?dl=0
https://www.dropbox.com/s/w8n6u26qedns9lj/firmware-20141018.tar.gz?dl=0
Thanks for the FW MrTeal. Worked perfect for me!
I've run a board with a piece of cardboard over one fan for a couple days to get temps up over 105, and nothing untoward has happened. For those that are interested, this is the 110C firmware.
https://www.dropbox.com/s/w8n6u26qedns9lj/firmware-20141018.tar.gz?dl=0
https://www.dropbox.com/s/w8n6u26qedns9lj/firmware-20141018.tar.gz?dl=0
Thanks as always Mr. Teal.
Let's say we want to underclock/undervolt these...right now I'm pulling around 700W for 600GHs...but this includes the fans and pump...
Would voltage/clock settings like 800@800 cause any long term problems...or even allow enough volts for the clock? Anyone have any working numbers for down-clocking?
I've run a board with a piece of cardboard over one fan for a couple days to get temps up over 105, and nothing untoward has happened. For those that are interested, this is the 110C firmware.
https://www.dropbox.com/s/w8n6u26qedns9lj/firmware-20141018.tar.gz?dl=0
https://www.dropbox.com/s/w8n6u26qedns9lj/firmware-20141018.tar.gz?dl=0
Thanks as always Mr. Teal.
Let's say we want to underclock/undervolt these...right now I'm pulling around 700W for 600GHs...but this includes the fans and pump...
Would voltage/clock settings like 800@800 cause any long term problems...or even allow enough volts for the clock? Anyone have any working numbers for down-clocking?
Hi all,
one of my best boards died today and I'm clueless how to get it up and running again. I have tried to turn off board, removed usb and all that.
Would someone know what can be wrong ?
Thank you very much
Have you tried fully powered it down, removing the 12V and USB connection?one of my best boards died today and I'm clueless how to get it up and running again. I have tried to turn off board, removed usb and all that.
Would someone know what can be wrong ?
Thank you very much
Yes, that was very first thing I have done.
Hi all,
one of my best boards died today and I'm clueless how to get it up and running again. I have tried to turn off board, removed usb and all that.
Would someone know what can be wrong ?
Thank you very much
Have you tried fully powered it down, removing the 12V and USB connection? one of my best boards died today and I'm clueless how to get it up and running again. I have tried to turn off board, removed usb and all that.
Would someone know what can be wrong ?
Thank you very much
I've run a board with a piece of cardboard over one fan for a couple days to get temps up over 105, and nothing untoward has happened. For those that are interested, this is the 110C firmware.
https://www.dropbox.com/s/w8n6u26qedns9lj/firmware-20141018.tar.gz?dl=0
https://www.dropbox.com/s/w8n6u26qedns9lj/firmware-20141018.tar.gz?dl=0
Hi all,
one of my best boards died today and I'm clueless how to get it up and running again. I have tried to turn off board, removed usb and all that.
Would someone know what can be wrong ?
Thank you very much
one of my best boards died today and I'm clueless how to get it up and running again. I have tried to turn off board, removed usb and all that.
Would someone know what can be wrong ?
Thank you very much
This has been asked before in this thread, but it wasn't fully answered: Does the Chain UP/DOWN work out of the box on the Habanero?
This has been asked before in this thread, but it wasn't fully answered: Does the Chain UP/DOWN work out of the box on the Habanero?
Decided not to pursue my problem board - 12v input is definitely shorted. That leaves two working dies on this board. If anyone is interested in an as-is tinker toy board - PM me.
Offering one Nepton 280 for sale in the Marketplace - https://bitcointalk.org/index.php?topic=823639.new#new
Offering one Nepton 280 for sale in the Marketplace - https://bitcointalk.org/index.php?topic=823639.new#new
Jump to: