Bitcoin Forum

Hi

i got a Sp20 running of 2 corsair cx850.
I got free power, and was wondering if it's possible to get i to run higere then 1610ghs by adjusting the volt setup.

the setup is:

Temp Front / Back T,B
15 °C / 67,72 °C
Fan Speed
Auto
Start Voltage
0.67 / 0.67 / 0.67 / 0.67
Max Voltage
0.74
Max Watts
280 / 280 / 280 / 280

Is there a guide to adjust so that it is running faster?

Thx.

You might be able to if you really push it hard.  But I would suggest not doing it. Most don't like being pushed to the 1.7T.

I would go for the 1.6 and not put all the strain on the miner.  They are amazing at under clocking, but over clocking is not their strong suit.

I have 20x sp20 and only 1 can make stable 1.63th/s...... for 1.6th+ you must live in a very cold place with low temperatures or you will burn the sp20's chips very fast.

ps : 1.7th is just a spondootech marketing trick i can say sp20 must be advertised as 1,4-1,5th/s miner, not really good miner for overclocking.



 

is this to high?

Temp Front / Back T,B
15 °C / 67,72 °C

or what is the max?

Yes you are pushing it.  72C is higher then I would like mine going.  I'm sure some do go that high. 

Honestly you are better off going for 1.5-1.6 and not pushing it to it's limits.  Some just will not do the 1.7.  They were not made for over clocking.

its better to look under asic stats page -->chip temps  ;)


Totally agree with @notlist3d.

2.6.14
Uptime:2399 | FPGA ver:100 | BIST in 144
-----BOARD-0-----
PSU[UNKNOWN]: 0->(266w/266w)[266 266 266] (->266w[266 266 266]) (lim=280) 0c 407GH cooling:0/0xc8
-----BOARD-1-----
PSU[UNKNOWN]: 0->(258w/258w)[258 258 258] (->258w[258 258 258]) (lim=280) 0c 399GH cooling:0/0xc8
-----BOARD-2-----
PSU[UNKNOWN]: 0->(269w/269w)[269 269 269] (->269w[269 269 269]) (lim=280) 0c 409GH cooling:0/0xc8
-----BOARD-3-----
PSU[UNKNOWN]: 0->(249w/249w)[249 249 249] (->249w[249 249 249]) (lim=280) 0c 391GH cooling:0/0xc8
LOOP[0] ON TO:0 (w:227)
 0: DC2DC/1/:[vlt1:734 vlt2:738(DCl:794 Tl:794 Ul:738) 105W 143A  53c] ASIC:[ 85c (125c) 1030hz(BL:1030)  111 (E:193) F:0 L:0]
 1: DC2DC/1/:[vlt1:730 vlt2:738(DCl:794 Tl:794 Ul:738) 110W 149A  77c] ASIC:[ 85c (125c) 1080hz(BL:1080)  116 (E:193) F:0 L:0]
LOOP[1] ON TO:0 (w:216)
 2: DC2DC/1/:[vlt1:734 vlt2:738(DCl:794 Tl:794 Ul:738) 111W 150A  83c] ASIC:[105c (125c) 1070hz(BL:1070)  110 (E:193) F:0 L:0]
 3: DC2DC/1/:[vlt1:714 vlt2:719(DCl:794 Tl:719 Ul:738) 98W 136A  85c] ASIC:[120c (125c) 1000hz(BL:1000)  106 (E:193) F:0 L:0]
LOOP[2] ON TO:0 (w:229)
 4: DC2DC/1/:[vlt1:734 vlt2:738(DCl:794 Tl:794 Ul:738) 108W 146A  55c] ASIC:[ 85c (125c) 1050hz(BL:1050)  117 (E:193) F:0 L:0]
 5: DC2DC/1/:[vlt1:734 vlt2:738(DCl:794 Tl:794 Ul:738) 110W 149A  73c] ASIC:[ 85c (125c) 1070hz(BL:1070)  112 (E:193) F:0 L:0]
LOOP[3] ON TO:0 (w:190)
 6: DC2DC/1/:[vlt1:722 vlt2:727(DCl:794 Tl:727 Ul:738) 99W 136A  78c] ASIC:[120c (125c) 980hz(BL: 980)   97 (E:193) F:0 L:0]
 7: DC2DC/1/:[vlt1:703 vlt2:709(DCl:794 Tl:709 Ul:738) 102W 144A  92c] ASIC:[120c (125c) 1050hz(BL:1050)   93 (E:193) F:0 L:0]

[H:HW:1607Gh (500),DC-W:845,L:0,A:8,MMtmp:0 TMP:(15/15)=>=>=>(66/66 , 69/69)]
Pushed 11 jobs , in HW queue 4 jobs (sw:0, hw:0)!
min:55 wins:862[this/last min:24/20] bist-fail:248, hw-err:0
leading-zeroes:42 idle promils[s/m]:0/0, rate:1398gh/s asic-count:328 (wins:9+15)
wall watts:1143
Fan:90, conseq:200
AC2DC BAD: 0 0
R/NR: 1607/0
RTF asics: 0
FET: 0:9 1:9
 0 restarted      0 reset          0 reset2         0 fake_wins
 0 stuck_bist     0 low_power      0 stuck_pll      0 runtime_dsble
 0 purge_queue    0 read_timeouts  0 dc2dc_i2c       0 read_tmout2    0 read_crptn
 0 purge_queue3   0 bad_idle
 0 err_murata
Adapter queues: rsp=1, req=24

The incoming air temp of 15C is just great. Since you have already got 3 of the 8 chips running at 120C, there probably really isn't much more to be had. While I personally think you should stop here, or even dial back a bit, there a couple of small adjustments that can be made. I don't recommend them though:

The fan can go to 100. Change the Auto setting to 100. I doubt this will help much. It probably won't sound much louder since it's already at 90 (selected by Auto).

The Max Volts can also go as high as .79 I think.

I think the power limit for each loop can actually go to 288 rather than 280. Right now you don't appear to be hitting 280 anyway, so this probably won't matter.

As has been said, 1600+ Ghash is an excellent "real world" number for an SP20. Seeking more really isn't wise.

Use Voltage/Max Voltage  0.790

FAN 90

you are running it too fast  see the 3 in bold at 120c = bad

lower the settings.

also this miner sucks power like mad at the speed you are doing.

drop the miner down to 1350gh if your power is not free

if your power is free try to get it to run at 1475-1525gh

Philipma has the best SP20 thread - https://bitcointalk.org/index.php?topic=872014.0

In that thread you can find all kinds of good information.  If you pay electricity I highly suggest looking at under clocking.   Even if you don't I would not push it as hard as you are.  It's going to hurt it one day if you do it continued at extreme heat.

I reach >1.7th only with under 0c air temp(temp front was stuck at 0c as it couldn't read under 0c) and max voltage
not really worth!

if you really really want to pick-up a bit more gigahash(is not worth it!) you can try to mod the miner for better temps
some ideas: change the air flow direction and adapt more fans, or bigger fan in place of the original, or outside the case with some kind of adapter!
                  or do the above big fan, fans, adapter thing on the front by somehow insulating the cables...
                  place the left board so that it has the radiators to the outside, now add cut the sides and add fans!
                  ....

good luck!

Is 120 bad for the chips?  I thought they throttle themselves to stay under 125 to prevent damage?  Of course generally speaking chips last longer if they are run cooler but I thought <125 is still within spec, correct me if I am wrong.

In addition to the temperature issues, the SP20, as do many miners, become grossly inefficient at their very top speed. By that I mean that last 100 Gh/s requires well over 100 watts to achieve. Everybody that claims to have free power really means that they have a limited amount of power they have already paid for in one way or another.

Do you realize how hot 120c is?   Using the throttle temp as a guide is not really a good idea.  You are pushing the SP20 very very hard.  Almost so hard it will throttle it. 

You would be better off electricity wise, and longer life if you don't try to push it so hard. 

you are wrong..   the miner will die faster.   If you read all the sp20 info.

 2 things kill them a shitty psu
running them past 1500gh without really good cooling.  ----------costs too much power and noisy as fuck.

even the techs from sp-tech mention  115C as the ideal max temp in more then one post.

Also I did a failure thread  asking for people's fried gear info.

Did not get a lot of replies (my guess is people don't like to say they pushed it too much)  .  The replies on the sp20 for failed machines were over 1500gh note only 2 or 3 answered for sp20's but they were over 1500gh.

This gear really should have been listed at 1500gh over that and your warranty   is void.

also on the power use end 1350gh = 800 watts    1500 gh = 1000 watts  1600gh = 1220 watts

I run mine at 10% fan, the 2 rear(loop 1 & 3) 'sections' run quite a bit hot up to 120C consistently, the firmware reliably declocks / devolts the ASICs as needed to keep them at 120C or lower. Ran this way for months w/o issues.
My exhaust temp has gotten high as 80C

One thing Ive started doing on my SP20's is setting the rear 'sections' at a lower start voltage / lower power max to keep them running cooler. The front ones Ive set higher to make up for the speed loss. This helps even out the temperatures quite a bit, between the front n back sections.

Good to know.  I'm currently getting a tad under 1.3T with a 750W PSU which sounds normal and yes loops 1 and 3 do run hotter so I will try dropping the power and voltage to see if that helps with temps.  I do have a HP DPS 1200 that I want to use instead to give me a little headroom, I just need to get a breakout board for it.

I would turn up the fan.  120C is still very very hot to run it over time. I would underclock some to get lower on those two loops.

What speed are you going for to get such high temperature?

Getting about 105 chip side @ 1.32T with these settings.  Thanks for the tips, as a new miner I find that people on this forum have been very helpful and patient.

Start Voltage 0.64 / 0.62 / 0.64 / 0.62
Max Watts 200 / 150 / 200 / 150

MY opinion is you have it in its sweet spot.
_______________________________________________________________________________ _______
_______________________________________________________________________________ _______
this type of overheating to 120c is not the same as running the fan at 90% and the gear at 1600gh
You run the fan at 10%  and I know you clock lower then 1600gh
Op is sending a lot more juice into his sp20 then you do.
The temp checks are in a few places not all over.  Thus it is far more likely his gear will burn up then your gear will.




 

guys you are crazy to run miners with 120C chip temps  ??? ???
how long do u think your miner will handle with this push ?  ;)

They run anywhere between 1300-1450GH depending on the unit & the quality of chips. 2 of my SP20's do run the fan at 30%, those were my higher clocked ones.
I really need to keep the other 2 fans at 10% cuz they are in living quarters, anything above 10% makes it unbearable lol!

It's interesting that on my one SP20, Loop #3 is the one that has a single "hot" chip on it. It is the same loop, that always exceeds it's power limit by a significant margin. The other 3 loops come in just short of their respective limits, but #3 is at 140W with a limit of 100W.

I have wondered if perhaps the one "hot" ASIC just needs some attention to it's heatsink and thermal paste. Since I am power limiting to about 600W total at the wall, everything runs just fine, and hence I don't really have a compelling reason to monkey with the hardware and risk serious damage.

Has anyone tried to tried to "fix" just one hot chip in their SP20?

I would guess most have not tried.  With SP gear if you open it you lose warranty.   So most never are even opened.

One way to make it obey the power limit is to drop its starting voltage that way it will climb up to that power limit and not go over.

Good idea on the Starting Voltage that I hadn't considered. I'll give it a try when I next decide to adjust the settings.

Thanks!!!