Topic: Hacking The KNC Firmware: Overclocking - page 15. (Read 144351 times)
Semi-offtopic. Are there any tests for different settings and results for under-clocking Nov Jupiters?
thank you crashoveride54902!
Just by changing the ASIC 1 and 2 die's to 775Mhz in the Advanced tab I got 7% more
Saturn @ 300Ghs/s.. : )
Just by changing the ASIC 1 and 2 die's to 775Mhz in the Advanced tab I got 7% more
Saturn @ 300Ghs/s.. : )
Agree the standard case setup is sub optimal.
My best air cooled configuration involves cooling the top and BOTTOM of the PCB.
http://i.imgur.com/9PZ95M8.jpg
I created a 3 sided tunnel just for the PCBs out of 2 modules.
(note on the back of the heatsink I inserted thin cardboard in bottom heatsink fins for side of tunnel)
I was gonna add a top to the tunnel but temps were so low I never did.
Before this configuration I added 15mm M3 male/female standoffs to the case for better bottom airflow.
This alone dropped my temps a few degrees.
Even with watercooled ASIC you still need lots of air on the PCB and VRM.
LOTS of air!
YMMV
My best air cooled configuration involves cooling the top and BOTTOM of the PCB.
http://i.imgur.com/9PZ95M8.jpg
I created a 3 sided tunnel just for the PCBs out of 2 modules.
(note on the back of the heatsink I inserted thin cardboard in bottom heatsink fins for side of tunnel)
I was gonna add a top to the tunnel but temps were so low I never did.
Before this configuration I added 15mm M3 male/female standoffs to the case for better bottom airflow.
This alone dropped my temps a few degrees.
Even with watercooled ASIC you still need lots of air on the PCB and VRM.
LOTS of air!
YMMV
Well the mystery of my super hot board has been solved.
When i took the lid off to fit the VRM heatsinks everything was working fine, then about halfway through doing them I noticed the fan stop, as I observed it over the course of an hour it was stopping/starting quite frequently, this was leading to the hot temps and dies shutting down. New fan fitted temps way lower, no dies shutting off and time to start raising the clocks
When i took the lid off to fit the VRM heatsinks everything was working fine, then about halfway through doing them I noticed the fan stop, as I observed it over the course of an hour it was stopping/starting quite frequently, this was leading to the hot temps and dies shutting down. New fan fitted temps way lower, no dies shutting off and time to start raising the clocks
Hey all,
I've recently got a 750 Jupiter and was hoping to squeeze a little more processing power out of it. It's running of a 950W PSU and runs around the 45 degree mark so theres room there to get a bit more out of it.
I've had a look through this forum but there is a lot of info opinion flying about, I was hoping to get some idea of voltages and frequencies to try? I'd like to get a couple hundred Gh/s more out of it if possible.
http://i.imgur.com/y6kytAa.jpg
Cheers fellas
Yes, I agree with you crashoveride.
He needs to get everything out.
But still is very hot with such low frequency+voltage.
Whats your ambient, tigggger?
He needs to get everything out.
But still is very hot with such low frequency+voltage.
Whats your ambient, tigggger?
Thanks to you both
I went mad ordering Antminers a while ago so despite having 2 air conditioning units temps are higher than I would like in the room
Got some stick on heatsinks for the vrm's and some of the antminers will be going soon which will hopefully help
you can do 231 or 241 oc with 4 vrm boards for 320 or so...just don't push the vrms over 60a tops i'd say...and that itself is pushing it and if you don't have extra cooling..if you get heatsinks n extra fans you can use 251 max i'd say for 8vrm with extra cooling you can do 291 no problem
Thank you!
Noob question, what is the 231 or 241 value?
its the oc setting ya use...either with padrino FW or ssh and editing a file
I've been reading through and fiddling around but maybe I can save some time here..
Are there definite/optimal settings for Oct Jupes?
5 Total 2 with die issues.
Are there definite/optimal settings for Oct Jupes?
5 Total 2 with die issues.
ya pretty much max settings for oct jups...if ya try 251 w/o extra cooling you'll get temp warning on vrms
you can do 231 or 241 oc with 4 vrm boards for 320 or so...just don't push the vrms over 60a tops i'd say...and that itself is pushing it and if you don't have extra cooling..if you get heatsinks n extra fans you can use 251 max i'd say for 8vrm with extra cooling you can do 291 no problem
Thank you!
Noob question, what is the 231 or 241 value?
Yes, I agree with you crashoveride.
He needs to get everything out.
But still is very hot with such low frequency+voltage.
Whats your ambient, tigggger?
He needs to get everything out.
But still is very hot with such low frequency+voltage.
Whats your ambient, tigggger?
was running my nov rig at 0.000v n 335...but one die would drop out..so i slowly started droping the volts n oc til it was happy staying on all the time but you have 3 vrms that aren't happy and on different dies so idk if that is your problem or not...also those vrms are little toasty...I would go that high if it was my rig...you'll melt the solder off it 
but you have 3 vrms that aren't happy and on different dies so idk if that is your problem or not...also those vrms are little toasty...I would go that high if it was my rig...you'll melt the solder off it Thanks for the advice, dropped the clocks and the voltages and they seem to be more stable now and are staying on.
However contrary to what I would expect the VRM's are getting even hotter than my other boards
Lower your frequency, or heatsink/cool your vrm's.
You will all have them at zero if you keep them at +95C each.
or just take'em outta the case and get some motherboard standoffs to hold them up and then a fan in front...easy setup and cools them off way better then having them in a case...but ya what tzortz said...its really hot in there
Just got saturn back online ... directly from hosting
Current firmware revision: 1.01
SPI: 1.95v
SPI Frequency: 201442 Hz
die 1 -0.1099v 750Mhz
die 2 -0.1099v 750Mhz
die 3 -0.1099v 750Mhz
die 4 -0.1099v 750Mhz
C/DC Voltage (V) Current (A) Power (W) Temperature (°C)
0 0.7764 46.3750 36.006 0.000
2 0.7783 47.0000 36.580 0.000
4 0.7988 47.6250 38.043 0.000
7 0.7822 48.1875 37.692 0.000
temp : 58.0 °C
Power : 147.791 W
286h/s @ pool ...
What are others getting from a Saturn Overclocked?
I should try first but .... If changing the die# to 775Mhz does it make a difference or I need to change as well the voltage?
thx
Current firmware revision: 1.01
SPI: 1.95v
SPI Frequency: 201442 Hz
die 1 -0.1099v 750Mhz
die 2 -0.1099v 750Mhz
die 3 -0.1099v 750Mhz
die 4 -0.1099v 750Mhz
C/DC Voltage (V) Current (A) Power (W) Temperature (°C)
0 0.7764 46.3750 36.006 0.000
2 0.7783 47.0000 36.580 0.000
4 0.7988 47.6250 38.043 0.000
7 0.7822 48.1875 37.692 0.000
temp : 58.0 °C
Power : 147.791 W
286h/s @ pool ...
What are others getting from a Saturn Overclocked?
I should try first but .... If changing the die# to 775Mhz does it make a difference or I need to change as well the voltage?
thx
you can do 231 or 241 oc with 4 vrm boards for 320 or so...just don't push the vrms over 60a tops i'd say...and that itself is pushing it and if you don't have extra cooling..if you get heatsinks n extra fans you can use 251 max i'd say for 8vrm with extra cooling you can do 291 no problem
was running my nov rig at 0.000v n 335...but one die would drop out..so i slowly started droping the volts n oc til it was happy staying on all the time but you have 3 vrms that aren't happy and on different dies so idk if that is your problem or not...also those vrms are little toasty...I would go that high if it was my rig...you'll melt the solder off it
but you have 3 vrms that aren't happy and on different dies so idk if that is your problem or not...also those vrms are little toasty...I would go that high if it was my rig...you'll melt the solder off it Thanks for the advice, dropped the clocks and the voltages and they seem to be more stable now and are staying on.
However contrary to what I would expect the VRM's are getting even hotter than my other boards
Lower your frequency, or heatsink/cool your vrm's.
You will all have them at zero if you keep them at +95C each.
Just got saturn back online ... directly from hosting
Current firmware revision: 1.01
SPI: 1.95v
SPI Frequency: 201442 Hz
die 1 -0.1099v 750Mhz
die 2 -0.1099v 750Mhz
die 3 -0.1099v 750Mhz
die 4 -0.1099v 750Mhz
C/DC Voltage (V) Current (A) Power (W) Temperature (°C)
0 0.7764 46.3750 36.006 0.000
2 0.7783 47.0000 36.580 0.000
4 0.7988 47.6250 38.043 0.000
7 0.7822 48.1875 37.692 0.000
temp : 58.0 °C
Power : 147.791 W
286h/s @ pool ...
What are others getting from a Saturn Overclocked?
I should try first but .... If changing the die# to 775Mhz does it make a difference or I need to change as well the voltage?
thx
Current firmware revision: 1.01
SPI: 1.95v
SPI Frequency: 201442 Hz
die 1 -0.1099v 750Mhz
die 2 -0.1099v 750Mhz
die 3 -0.1099v 750Mhz
die 4 -0.1099v 750Mhz
C/DC Voltage (V) Current (A) Power (W) Temperature (°C)
0 0.7764 46.3750 36.006 0.000
2 0.7783 47.0000 36.580 0.000
4 0.7988 47.6250 38.043 0.000
7 0.7822 48.1875 37.692 0.000
temp : 58.0 °C
Power : 147.791 W
286h/s @ pool ...
What are others getting from a Saturn Overclocked?
I should try first but .... If changing the die# to 775Mhz does it make a difference or I need to change as well the voltage?
thx
was running my nov rig at 0.000v n 335...but one die would drop out..so i slowly started droping the volts n oc til it was happy staying on all the time but you have 3 vrms that aren't happy and on different dies so idk if that is your problem or not...also those vrms are little toasty...I would go that high if it was my rig...you'll melt the solder off it
but you have 3 vrms that aren't happy and on different dies so idk if that is your problem or not...also those vrms are little toasty...I would go that high if it was my rig...you'll melt the solder off it Thanks for the advice, dropped the clocks and the voltages and they seem to be more stable now and are staying on.
However contrary to what I would expect the VRM's are getting even hotter than my other boards
My november jupiter had been running along nicely around 750GH, but recently the speeds have dropped.
Looking under advanced I see the following
Die 1 and Die 4 are both showing 0 voltage, if I leave it alone in a few minutes it starts up again and hashes normally, runs for a few minutes and then back to 0 again.
Before I start messing with clocks and voltages thought I'd ask if anyone else has had this and the best way to fix it.
Looking under advanced I see the following
Die 1 and Die 4 are both showing 0 voltage, if I leave it alone in a few minutes it starts up again and hashes normally, runs for a few minutes and then back to 0 again.
Before I start messing with clocks and voltages thought I'd ask if anyone else has had this and the best way to fix it.
was running my nov rig at 0.000v n 335...but one die would drop out..so i slowly started droping the volts n oc til it was happy staying on all the time
but you have 3 vrms that aren't happy and on different dies so idk if that is your problem or not...also those vrms are little toasty...I would go that high if it was my rig...you'll melt the solder off it 
Bumped my clocks up some and now I am getting 1.5 TH/s at 1450W. HW errors are .002% and all the temps look good. Thanks padrino!
What cooling are you using? You seem to have it very nicely cooled.
Also, what did you make change the die voltages?
The i30 coolers, but I removed the thermal paste and replaced it with PK1. Taking the modules out of the case helped a lot. The modules need airflow on the bottom of the PCB. Actually flipped one on its side and pointed a 120 mm fan at it, temps dropped 10C.
No voltage changes. HW error % is .003.
My november jupiter had been running along nicely around 750GH, but recently the speeds have dropped.
Looking under advanced I see the following
Die 1 and Die 4 are both showing 0 voltage, if I leave it alone in a few minutes it starts up again and hashes normally, runs for a few minutes and then back to 0 again.
Before I start messing with clocks and voltages thought I'd ask if anyone else has had this and the best way to fix it.
Looking under advanced I see the following
Die 1 and Die 4 are both showing 0 voltage, if I leave it alone in a few minutes it starts up again and hashes normally, runs for a few minutes and then back to 0 again.
Before I start messing with clocks and voltages thought I'd ask if anyone else has had this and the best way to fix it.
What is a recommended max temp for the VRMs? I don't want to over do it. :-)
Thanks!
Thanks!
I think you must not exceed 95C at GUI.
At 93C appears an ERROR.
Thanks!
What is a recommended max temp for the VRMs? I don't want to over do it. :-)
Thanks!
Thanks!
I think you must not exceed 95C at GUI.
At 93C appears an ERROR.
What is a recommended max temp for the VRMs? I don't want to over do it. :-)
Thanks!
Thanks!
Bumped my clocks up some and now I am getting 1.5 TH/s at 1450W. HW errors are .002% and all the temps look good. Thanks padrino!
What cooling are you using? You seem to have it very nicely cooled.
Also, what did you make change the die voltages?
Jump to: