The Habanero - 650GH/s - OOS

[eightcylinders]

Looking forward to see what you can do!  Also note the back-plate can get quite hot and depending on how much you push the board you might need to cool them a bit.

Well, I was thinking about that but I could not find any details about what the backplate looks like (size, placement, etc. for mounting some kind of cooling). 

Yes the Raystorm does support the LGA115x spacing.  FYI, on the peppermining.com site the cooling compatibility is listed as "standard 75mm Intel (lga 1366 socket b)" but in any case the Raystorm supports both.

[xjack]

Just a quick update now that my miners are off the test bench and into a "production" environment.

Nepton 280L - one 140mm stock fan, one 125cfm screamer (Same fan as the S1 runs, except I clipped off the PWM/RPM wires).

These are running at ambient temp of 90F/33C.



A few notes:

My windows XP test bench squelches cores at anything over 90C.  My xubuntu 12.04 hasn't clocked a single squelch in nearly 24 hours.  Same cgminer version.

Someone mentioned above that you can go too tight/uneven on the cooling.  I chipped the corner of one die on my second unit - incidentally I got very lucky, since "Chip" doesn't show any performance detriment.

If anyone is posting their speeds and temps, can you please post the ambient temp as well?  It's hard to make operating comparisons without ambient temps.

These ran stable for 8 hours yesterday 900Mhz(675gh/s) at 100C.  If I can get the ambient temps down (or add another screamer fan), I'm sure I can push farther.  MrTeal, at what clock speed do these trend towards implosion?  Can you recommend an "ideal" operating speed(or temp) for 24/7 hashing?

Awesome boards, congratulations to you and your team for putting out a quality product!

[MrTeal]

Just a quick update now that my miners are off the test bench and into a "production" environment.

Nepton 280L - one 140mm stock fan, one 125cfm screamer (Same fan as the S1 runs, except I clipped off the PWM/RPM wires).

These are running at ambient temp of 90F/33C.

A few notes:

My windows XP test bench squelches cores at anything over 90C.  My xubuntu 12.04 hasn't clocked a single squelch in nearly 24 hours.  Same cgminer version.

Someone mentioned above that you can go too tight/uneven on the cooling.  I chipped the corner of one die on my second unit - incidentally I got very lucky, since "Chip" doesn't show any performance detriment.

If anyone is posting their speeds and temps, can you please post the ambient temp as well?  It's hard to make operating comparisons without ambient temps.

These ran stable for 8 hours yesterday 900Mhz(675gh/s) at 100C.  If I can get the ambient temps down (or add another screamer fan), I'm sure I can push farther.  MrTeal, at what clock speed do these trend towards implosion?  Can you recommend an "ideal" operating speed(or temp) for 24/7 hashing?

Awesome boards, congratulations to you and your team for putting out a quality product!

On the one I have with the Nepton 280L with high CFM fans I use the following command line arguments
cgminer.exe --hfa-temp-overheat 103 --hfa-fan 100 --hfa-hash-clock XXX --hfa-noshed --hfa-temp-target 0 --api-network --api-listen
Generally that board is topped out around 950MHz on one of the dies, but it's pretty chip specific.
Hashfast claims the chips can run quite hot (and recently changed the FW on their Evo boards to run them up to 110C, and I've never had an issue with a chip getting damaged due to thermals. The chip itself will shut down at 105C, so I set the limit a little below that so cgminer will lower clocks if it gets too hot.

[gateway]

Looking forward to see what you can do!  Also note the back-plate can get quite hot and depending on how much you push the board you might need to cool them a bit.

Well, I was thinking about that but I could not find any details about what the backplate looks like (size, placement, etc. for mounting some kind of cooling). 

Yes the Raystorm does support the LGA115x spacing.  FYI, on the peppermining.com site the cooling compatibility is listed as "standard 75mm Intel (lga 1366 socket b)" but in any case the Raystorm supports both.

[gateway]

Just a quick update now that my miners are off the test bench and into a "production" environment.

Nepton 280L - one 140mm stock fan, one 125cfm screamer (Same fan as the S1 runs, except I clipped off the PWM/RPM wires).

These are running at ambient temp of 90F/33C.

A few notes:

My windows XP test bench squelches cores at anything over 90C.  My xubuntu 12.04 hasn't clocked a single squelch in nearly 24 hours.  Same cgminer version.

Someone mentioned above that you can go too tight/uneven on the cooling.  I chipped the corner of one die on my second unit - incidentally I got very lucky, since "Chip" doesn't show any performance detriment.

If anyone is posting their speeds and temps, can you please post the ambient temp as well?  It's hard to make operating comparisons without ambient temps.

These ran stable for 8 hours yesterday 900Mhz(675gh/s) at 100C.  If I can get the ambient temps down (or add another screamer fan), I'm sure I can push farther.  MrTeal, at what clock speed do these trend towards implosion?  Can you recommend an "ideal" operating speed(or temp) for 24/7 hashing?

Awesome boards, congratulations to you and your team for putting out a quality product!

On the one I have with the Nepton 280L with high CFM fans I use the following command line arguments
cgminer.exe --hfa-temp-overheat 103 --hfa-fan 100 --hfa-hash-clock XXX --hfa-noshed --hfa-temp-target 0 --api-network --api-listen
Generally that board is topped out around 950MHz on one of the dies, but it's pretty chip specific.
Hashfast claims the chips can run quite hot (and recently changed the FW on their Evo boards to run them up to 110C, and I've never had an issue with a chip getting damaged due to thermals. The chip itself will shut down at 105C, so I set the limit a little below that so cgminer will lower clocks if it gets too hot.

Also at some point you need more voltage to push the boards.. we programmed them at .93, but you need a tool which we are trying to get HF to allow us to release which allows you to bump up the voltages because at some point Mhz wise the chip needs more power.  You can see this sometimes in the Pepper Mining app when you watch the die temps if every you see small dips randomly (not to be confused with a board clocking it self down or squelching a core).

I don't get you windows people j/k.. anyhow.. good to see things are up and running! keep that back plate cool or fans blowing air on the board as as well.

[eightcylinders]

Looking forward to see what you can do!  Also note the back-plate can get quite hot and depending on how much you push the board you might need to cool them a bit.

Well, I was thinking about that but I could not find any details about what the backplate looks like (size, placement, etc. for mounting some kind of cooling). 

Yes the Raystorm does support the LGA115x spacing.  FYI, on the peppermining.com site the cooling compatibility is listed as "standard 75mm Intel (lga 1366 socket b)" but in any case the Raystorm supports both.

Thanks gateway, that helps.  Though I am not sure how I could run any further cooling on that plate given the indentations.  What is the plate size?  Can it be removed easily and replaced (assuming I could find a liquid cooling block big enough to replace it ..) 

[xjack]

I don't get you windows people j/k.. anyhow.. good to see things are up and running! keep that back plate cool or fans blowing air on the board as as well.

Please note my reliable mining controllers are Linux based.   

[gateway]

I don't get you windows people j/k.. anyhow.. good to see things are up and running! keep that back plate cool or fans blowing air on the board as as well.

Please note my reliable mining controllers are Linux based.   

+1 

[eightcylinders]

Looking at the close-ups just posted, I see some interesting connectors on the board.  Are any of these "live" and usable? If so, details please!!!

• FAN2, FAN3 and FAN4 - are with these control PWM fans based on the chip temps?  Are there setpoints that can be software-configured in the firmware?
• PUMP connection?  To control flow rate for liquid cooling pump??  Same questions as for fan controller re what the setpoints are and if live
• Chain Down and Chain Up?  Obviously would need to solder a connector, but are these part of the circuit?  If "chained" would there be any benefit versus simply making separate USB connections?

[jasemoney]

Looking at the close-ups just posted, I see some interesting connectors on the board.  Are any of these "live" and usable? If so, details please!!!

• FAN2, FAN3 and FAN4 - are with these control PWM fans based on the chip temps?  Are there setpoints that can be software-configured in the firmware?
• PUMP connection?  To control flow rate for liquid cooling pump??  Same questions as for fan controller re what the setpoints are and if live
• Chain Down and Chain Up?  Obviously would need to solder a connector, but are these part of the circuit?  If "chained" would there be any benefit versus simply making separate USB connections?

i would be very interested if they could be chained...

[nexus99]

Finally got my 2 units up and running. They blow up at anything faster than 550 mhz... so I need to put some tweaking time in.

[QuiveringGibbage]

Looking at the close-ups just posted, I see some interesting connectors on the board.  Are any of these "live" and usable? If so, details please!!!

• FAN2, FAN3 and FAN4 - are with these control PWM fans based on the chip temps?  Are there setpoints that can be software-configured in the firmware?
• PUMP connection?  To control flow rate for liquid cooling pump??  Same questions as for fan controller re what the setpoints are and if live
• Chain Down and Chain Up?  Obviously would need to solder a connector, but are these part of the circuit?  If "chained" would there be any benefit versus simply making separate USB connections?

i would be very interested if they could be chained...

from my experience with the Sierra/BabyJets boards, it's much more efficient to mine to each board/chip individually so you can tune your settings to each chip's unique behaviour. whereas if you were to chain the boards, setting one freq will be the same across all boards you've chained together.

you can see what i mean here: http://imgur.com/a/94wo7
the first and last pictures are the cgminer stats from when the Sierra was mining as one (clock speed at 604MHz) and after i took it apart into 3 individual boards (550MHz, 656MHz & 666MHz).
note: the last pic's hash readings a little high... oddly...

Cheers,
QG

[ZBC3]

I took my h100i off the board to attach it like the picture and now it's not working , doh.

I'm getting a libusb error pipe, libusb not found error.

Hashfast detect failed to initialize incorrect device.

The device keeps adding and disabling the device with those previous errors.

I'm using minepeon  2.4.6 with cgminer 4.3.4.

[faxfan2002]

What PSU's are people using?

[nexus99]

I took my h100i off the board to attach it like the picture and now it's not working , doh.

I'm getting a libusb error pipe, libusb not found error.

Hashfast detect failed to initialize incorrect device.

The device keeps adding and disabling the device with those previous errors.

I'm using minepeon  2.4.6 with cgminer 4.3.4.

Have you rebooted the Pi?

[nexus99]

What PSU's are people using?

I'm using a HP server PSU on one and a 1000 watt PC PSU from Rosewill on another. The HP is using a gigampz breakout card.

[xjack]

What PSU's are people using?

I had a couple Corsair RM1000 leftover from another project.

If you have to buy PSUs, look at the sidehack breakout boards, they'll save you quite a bit of money.

[ZBC3]

I even put the image on a new sd card and get the same results. I tried loosening the screws that are holding the pump on but nothing happened then either. My other Havanero works without issues on the same Pi.

The Pi also wants to recognize the device as black arrow and says the USB isn't programmed.

[jasemoney]

Looking at the close-ups just posted, I see some interesting connectors on the board.  Are any of these "live" and usable? If so, details please!!!

• FAN2, FAN3 and FAN4 - are with these control PWM fans based on the chip temps?  Are there setpoints that can be software-configured in the firmware?
• PUMP connection?  To control flow rate for liquid cooling pump??  Same questions as for fan controller re what the setpoints are and if live
• Chain Down and Chain Up?  Obviously would need to solder a connector, but are these part of the circuit?  If "chained" would there be any benefit versus simply making separate USB connections?

i would be very interested if they could be chained...

from my experience with the Sierra/BabyJets boards, it's much more efficient to mine to each board/chip individually so you can tune your settings to each chip's unique behaviour. whereas if you were to chain the boards, setting one freq will be the same across all boards you've chained together.

you can see what i mean here: http://imgur.com/a/94wo7
the first and last pictures are the cgminer stats from when the Sierra was mining as one (clock speed at 604MHz) and after i took it apart into 3 individual boards (550MHz, 656MHz & 666MHz).
note: the last pic's hash readings a little high... oddly...

Cheers,
QG

Great pics, thanks for sharing.  And thats very interesting. 

[btchedge]

I took my h100i off the board to attach it like the picture and now it's not working , doh.

I'm getting a libusb error pipe, libusb not found error.

Hashfast detect failed to initialize incorrect device.

The device keeps adding and disabling the device with those previous errors.

I'm using minepeon  2.4.6 with cgminer 4.3.4.

I had the same problem when I significantly re-routed the tubing. In my situation I clearly loosened the cooling head. Have you tried to the monitoring tool at a very low clock speed to make sure you have good contact? Reseating the cooling head, with the help of the tool fixed this for me.