The Habanero - 650GH/s - OOS

[gateway]

Can you guys help with some board comparisons?

My board gets:

MHz: 600
GH/s: ~450
Temp:  100,102,102,98

What are you guys seeing at 600mhz?

As for now a higher MHz get me 105 and an overheat.

(I do want to figure out how to get the board up to 650 gh/s.)

Also make sure to use fan plug 3 or 4, fan plug 2 is set to 50% right now until we release a firmware update

[nexus99]

Can you guys help with some board comparisons?

My board gets:

MHz: 600
GH/s: ~450
Temp:  100,102,102,98

What are you guys seeing at 600mhz?

As for now a higher MHz get me 105 and an overheat.

(I do want to figure out how to get the board up to 650 gh/s.)

Also make sure to use fan plug 3 or 4, fan plug 2 is set to 50% right now until we release a firmware update

tweaking a bit more I am at

MHz: 800 mhz
GH/s: ~603 gh/s
Temp:  99, 100, 102, 99

I will pull the pump head off and reset it again in the near future and see how that goes.
If I can pin the cooling down just a bit more I think I can get this guy up to where he needs to be.

[MrTeal]

Is a batch 3 likely ?

Batch 2 progress...

Batch 2 is progressing nicely, all the boards are built and are on their way through testing now. We don't expect to be late at all with starting shipping early next week.

[MrTeal]

Can you guys help with some board comparisons?

My board gets:

MHz: 600
GH/s: ~450
Temp:  100,102,102,98

What are you guys seeing at 600mhz?

As for now a higher MHz get me 105 and an overheat.

(I do want to figure out how to get the board up to 650 gh/s.)

Also make sure to use fan plug 3 or 4, fan plug 2 is set to 50% right now until we release a firmware update

tweaking a bit more I am at

MHz: 800 mhz
GH/s: ~603 gh/s
Temp:  99, 100, 102, 99

I will pull the pump head off and reset it again in the near future and see how that goes.
If I can pin the cooling down just a bit more I think I can get this guy up to where he needs to be.

Yeah, try giving it a good cleaning. Make sure there's not lint there or anything else that could compromise the interface. I'm seeing mid 90s at 800MHz/910mV with the stock 55CFM fans and the free included thermal paste on a Swiftech H220, so your temps should be better.

[Airwhale]

Is a batch 3 likely ?

Batch 2 progress...

Batch 2 is progressing nicely, all the boards are built and are on their way through testing now. We don't expect to be late at all with starting shipping early next week.

On thursday you said that you expected to be opening batch two back up once you had a good idea of yields and such.  Do you still plan on doing this?  If not,  let me (us) know so we can stop repeatedly checking the site... really hoping I can get some boards in this batch =)

[GenTarkin]

Can you guys help with some board comparisons?

My board gets:

MHz: 600
GH/s: ~450
Temp:  100,102,102,98

What are you guys seeing at 600mhz?

As for now a higher MHz get me 105 and an overheat.

(I do want to figure out how to get the board up to 650 gh/s.)

Also make sure to use fan plug 3 or 4, fan plug 2 is set to 50% right now until we release a firmware update

tweaking a bit more I am at

MHz: 800 mhz
GH/s: ~603 gh/s
Temp:  99, 100, 102, 99

I will pull the pump head off and reset it again in the near future and see how that goes.
If I can pin the cooling down just a bit more I think I can get this guy up to where he needs to be.

Yeah, try giving it a good cleaning. Make sure there's not lint there or anything else that could compromise the interface. I'm seeing mid 90s at 800MHz/910mV with the stock 55CFM fans and the free included thermal paste on a Swiftech H220, so your temps should be better.

I wonder if the temp variance is simply just the particular die ASIC quality. Traditionally, higher quality dies ramp up heat really quick as you go over their nominal voltage ceiling, but are able to reach very high clockspeeds at lower voltages. While, lower quality dies are very easy to feed higher voltages w/o such huge increase in heat, but also, normally requiring higher voltage to reach higher speeds.

[gateway]

I wonder if the temp variance is simply just the particular die ASIC quality. Traditionally, higher quality dies ramp up heat really quick as you go over their nominal voltage ceiling, but are able to reach very high clockspeeds at lower voltages. While, lower quality dies are very easy to feed higher voltages w/o such huge increase in heat, but also, normally requiring higher voltage to reach higher speeds.

Their is deff variances in these chips, some will perform better than others..

[daddyfatsax]

Amazon dropped the price of the H110i, only $5 more than the H80i.
http://www.amazon.com/dp/B009ZN2NH6/

[bit_NINj4]

Amazon dropped the price of the H110i, only $5 more than the H80i.
http://www.amazon.com/dp/B009ZN2NH6/

Because of it's copper radiator, the Swiftech H220 will destroy any other all-in-one systems once you replace the fans. By ~10C

[daddyfatsax]

Amazon dropped the price of the H110i, only $5 more than the H80i.
http://www.amazon.com/dp/B009ZN2NH6/

Because of it's copper radiator, the Swiftech H220 will destroy any other all-in-one systems once you replace the fans. By ~10C

I have the H220 in my personal computer, and I agree it is awesome. Problem is the H220 is not available in the US anymore and the comparable H2O-X20 system is $240. So $180 for 2 complete water cooling loops is not a bad deal.

[Hippie Tech]

Can you guys help with some board comparisons?

My board gets:

MHz: 600
GH/s: ~450
Temp:  100,102,102,98

What are you guys seeing at 600mhz?

As for now a higher MHz get me 105 and an overheat.

(I do want to figure out how to get the board up to 650 gh/s.)

Also make sure to use fan plug 3 or 4, fan plug 2 is set to 50% right now until we release a firmware update

tweaking a bit more I am at

MHz: 800 mhz
GH/s: ~603 gh/s
Temp:  99, 100, 102, 99

I will pull the pump head off and reset it again in the near future and see how that goes.
If I can pin the cooling down just a bit more I think I can get this guy up to where he needs to be.

Yeah, try giving it a good cleaning. Make sure there's not lint there or anything else that could compromise the interface. I'm seeing mid 90s at 800MHz/910mV with the stock 55CFM fans and the free included thermal paste on a Swiftech H220, so your temps should be better.

Has anyone checked to see if the chip and/or cooler are flat ?

How safe is it to lap those chips ?

[nexus99]

Alright.

Now with the 2 monster fans on the radiator ... plus another 4 120mm 'normal' fans blowing on the board itself.

MHz: 850
GH/s: ~635
Temp:  96,96,99,96

It sounds like a hurricane.

[ZiG]

Amazon dropped the price of the H110i, only $5 more than the H80i.
http://www.amazon.com/dp/B009ZN2NH6/

Because of it's copper radiator, the Swiftech H220 will destroy any other all-in-one systems once you replace the fans. By ~10C

I have the H220 in my personal computer, and I agree it is awesome. Problem is the H220 is not available in the US anymore and the comparable H2O-X20 system is $240. So $180 for 2 complete water cooling loops is not a bad deal.

+1

I have H220 ..."Cooler Master" version...it is a BEAST...the best imho...

ZiG

[MrTeal]

Can you guys help with some board comparisons?

My board gets:

MHz: 600
GH/s: ~450
Temp:  100,102,102,98

What are you guys seeing at 600mhz?

As for now a higher MHz get me 105 and an overheat.

(I do want to figure out how to get the board up to 650 gh/s.)

Also make sure to use fan plug 3 or 4, fan plug 2 is set to 50% right now until we release a firmware update

tweaking a bit more I am at

MHz: 800 mhz
GH/s: ~603 gh/s
Temp:  99, 100, 102, 99

I will pull the pump head off and reset it again in the near future and see how that goes.
If I can pin the cooling down just a bit more I think I can get this guy up to where he needs to be.

Yeah, try giving it a good cleaning. Make sure there's not lint there or anything else that could compromise the interface. I'm seeing mid 90s at 800MHz/910mV with the stock 55CFM fans and the free included thermal paste on a Swiftech H220, so your temps should be better.

Has anyone checked to see if the chip and/or cooler are flat ?

How safe is it to lap those chips ?

On a scale between 0 and 10, where 0 is "Not safe at all" and 10 is "Totally safe, lap those chips until they shine", I would say -53.
Do not try it, you'll almost certainly break them.

[MrTeal]

Is a batch 3 likely ?

Batch 2 progress...

Batch 2 is progressing nicely, all the boards are built and are on their way through testing now. We don't expect to be late at all with starting shipping early next week.

On thursday you said that you expected to be opening batch two back up once you had a good idea of yields and such.  Do you still plan on doing this?  If not,  let me (us) know so we can stop repeatedly checking the site... really hoping I can get some boards in this batch =)

No testing happened over the weekend, but we should know early next week once we get back to it and put the rest of the units through.

[Batshark]

Finally managed to fire up one board (was making frame for it) and noticed that 850MHz gave me around 630GH/s then I have tried 900MHz and it gave me 600GH/s.?

Not sure if it is my PSU or Voltage. Voltage was 0.86.

Try running the Pepper app. What you'll see if you go too high with insufficient voltage is that dies start stalling, and the temperature will drop.
I got these from gateway last night, so I'm pilfering them but attribution.

This is a board at 950MHz, with different voltage setpoints.
930mV
https://peppermining.com/wp-content/uploads/2014/06/930mV.jpg

940mV
https://peppermining.com/wp-content/uploads/2014/06/940mV.jpg

950mV
https://peppermining.com/wp-content/uploads/2014/06/950mV.jpg

970mV
https://peppermining.com/wp-content/uploads/2014/06/970mV.jpg

You can see the temperatures drop constantly at 930mV; the chip is being starved for voltage. It's much better at 940mV, and at 950mV three of the dies are running fine. At 970mV they're all running good.
With the python script gateway posted you can even adjust the dies individually. For the above case, you could do
./hftool.py -w 0:950@950,1:950@950,2:950@950,3:970@950
That would set the first 3 dies to 950mV and the last one to 970mV

Note that cgminer overwrites the frequency value (the number after the @) with it's setting, unless you use --hfa-hash-clock 1
That can be useful if you're really trying to push the board. You might end up with a die that just can't do 950MHz even at higher voltages, but the other three can get up to 1GHz. If that's the case, you can set the defaults for that board individually and run some of the dies faster than the others.

I've tried getting this to run on windows with no luck. I get 'HFBulkDevice not found' in a loop. Does this only work on ubuntu? I'm unable to get past ~875 at the stock voltage. Is hftool the only way to adjust voltage?

Right now I'm running:

https://i.imgur.com/HOUkBDk.png

Anything above this and I get fluctuating temps like gateway demonstrated.

Thanks, great boards btw!

[gateway]

Amazon dropped the price of the H110i, only $5 more than the H80i.
http://www.amazon.com/dp/B009ZN2NH6/

Because of it's copper radiator, the Swiftech H220 will destroy any other all-in-one systems once you replace the fans. By ~10C

I have the H220 in my personal computer, and I agree it is awesome. Problem is the H220 is not available in the US anymore and the comparable H2O-X20 system is $240. So $180 for 2 complete water cooling loops is not a bad deal.

the coolmaster glacer 240l is the same one http://www.amazon.com/gp/product/B00GJYNM4M

[gateway]

I've tried getting this to run on windows with no luck. I get 'HFBulkDevice not found' in a loop. Does this only work on ubuntu? I'm unable to get past ~875 at the stock voltage. Is hftool the only way to adjust voltage?

Right now I'm running:



Anything above this and I get fluctuating temps like gateway demonstrated.

Thanks, great boards btw!

Yea.. most of the dev that hf did is on ubuntu boxes running 13.x not much in the lines of support by HF for windows.  I have been meaning to create a virtual image that you could run in virtual box that would allow you to run the tool in ubuntu but I havent had much time as of late.

[gateway]

Just a word of warning for all wanting to push their boards or if your a first timer using a liquid cooling block.

The dies on the chips can crack if your apply to much pressure on them, once cracked well that die is dead, you can use the hf-tool to disable the die but your only hashing on 3 then, which well you can get 500Gh/s out of.

I just want to caution everyone thats playing around with these things that this is a real possibility so please take your time, do not over tighten, make sure to not put more pressure on one side than the other etc.

[bit_NINj4]

Amazon dropped the price of the H110i, only $5 more than the H80i.
http://www.amazon.com/dp/B009ZN2NH6/

Because of it's copper radiator, the Swiftech H220 will destroy any other all-in-one systems once you replace the fans. By ~10C

I have the H220 in my personal computer, and I agree it is awesome. Problem is the H220 is not available in the US anymore and the comparable H2O-X20 system is $240. So $180 for 2 complete water cooling loops is not a bad deal.

the coolmaster glacer 240l is the same one http://www.amazon.com/gp/product/B00GJYNM4M

Yea my bad, I should have mentioned that. Swiftech helped design the Cooler Master Glacer 240L after it's H220. Though if you really want to go overboard and you're Canadian, I'd put my money on the H320.
Just don't forget to upgrade those fans. Noctua NF-F12 PWM would be my choice or the Corsair SP120. Fans will be the deciding factor with all your liquid cooling.
And for the argument of 'push' vs 'pull', performance wise, there's almost no difference. But in 'pull' you have access to the radiator to clean the dust and have no downtime.