Topic: Post your Temps! (Read 3911 times)

5870 and 5830 water cooled.

My 7850 runs at 63C, a few degrees lower during winter. Fun times 

BGFMiner labels devices like that, I guess OCL = OpenCL device.

my 5830, before fan/sink clean and new paste = 75C @ 90% fan
             after                                           = 65C @ 30% fan

ambient around 25C.

I guess it was worth it

just using some ATI 5970s

using bfgminer on ubuntu, i think it might be labeling them as OCL because of the AMD drivers i am using maybe? not sure really.

All i know is it's working and stable, so it can call em UFOs for all i care

Hmm i think you are going to have some hard times in summer...

OCL what kind of hardware is that? 

 OCL 0:  41.5C  960RPM | 326.5/311.6/305.0Mh/s | A:16746 R:261 HW:0 U: 4.17/m
 OCL 1:  44.0C  960RPM | 329.2/311.7/304.4Mh/s | A:16803 R:230 HW:0 U: 4.18/m
 OCL 2:  51.0C  960RPM | 330.2/311.3/306.6Mh/s | A:16900 R:263 HW:0 U: 4.21/m
 OCL 3:  52.5C  960RPM | 329.3/311.3/306.8Mh/s | A:16938 R:279 HW:0 U: 4.21/m
 OCL 4:  45.5C  960RPM | 328.8/311.3/308.7Mh/s | A:16954 R:255 HW:0 U: 4.22/m
 OCL 5:  50.0C  960RPM | 328.2/311.3/303.7Mh/s | A:16750 R:215 HW:0 U: 4.17/m
--------------------------------------------------------------------------------

finally got my cards semi-stable, need to add the 4th, then i'll try to maybe OC a bit

My rig is 3x 7950s, the temps are usually ~84 and then the one in the middle is like 90 because it receives less air flow.

 cgminer version 3.1.0 - Started: [2013-05-28 07:51:32]
--------------------------------------------------------------------------------
 (5s):283.8M (avg):292.6Mh/s | A:14740  R:50  HW:0  U:4.0/m  WU:4.1/m
 ST: 2  SS: 0  NB: 423  LW: 27951  GF: 10  RF: 1
 Connected to stratum.bitcoin.cz diff 1 with stratum as user XXXXXXXXXX
 Block: 00dee6b00eaca825...  Diff:12.2M  Started: [20:23:35]  Best share: 40.7K
--------------------------------------------------------------------------------
 [P]ool management [G]PU management [S ]ettings [D]isplay options [Q]uit
 GPU 0:  74.0C  74%    | 135.0M/144.6Mh/s | A:7274 R:28 HW:0 U:2.00/m I: 5
 GPU 1:  66.0C  66%    | 146.4M/148.0Mh/s | A:7466 R:22 HW:0 U:2.05/m I: 5
--------------------------------------------------------------------------------

 [2013-05-30 20:29:53] Accepted 0546af3d Diff 48/1 GPU 0
 [2013-05-30 20:30:13] Accepted be41b1c1 Diff 1/1 GPU 0
 [2013-05-30 20:30:16] Accepted 825685f3 Diff 1/1 GPU 1
 [2013-05-30 20:30:19] Accepted 6deb3d52 Diff 2/1 GPU 0
 [2013-05-30 20:30:23] Accepted cfbeae14 Diff 1/1 GPU 0
 [2013-05-30 20:31:00] Accepted 9e9e05d4 Diff 1/1 GPU 0
 [2013-05-30 20:31:00] Accepted 7409e3da Diff 2/1 GPU 1
 [2013-05-30 20:31:20] Accepted 336cf4f6 Diff 4/1 GPU 1
 [2013-05-30 20:31:21] Accepted e2c83b28 Diff 1/1 GPU 1
 [2013-05-30 20:31:41] Accepted 42f385c6 Diff 3/1 GPU 1
 [2013-05-30 20:31:43] Accepted e2e340f3 Diff 1/1 GPU 1
 [2013-05-30 20:32:11] Accepted f63c0558 Diff 1/1 GPU 1

Running non-stop for 2 days

When mining litecoins my 3 stacked cards on gpu run upto 76c with outside temperature reaching 30.
With bitcoin downclocked its usually 69, but fans are on really low settings then.

One of my fans start to make noises, I fixed it but, I started thinking... why not replace the original fans and store them, so we can sell the card later on when we no longer use it! and put two Arctic cooling F PWM F8 or F9 CO fans screwed to the heatsink directly? these are designed for 24/7 operation.

I will try this in a couple of days and post pics, the fans are 4 wire, PWM with speed sensor, and, removing the plastic cover, will not be too difficult to bolt them to the heat-sink at all, I will use F8 for 78xx cards and F9 for 79xx cards (F9 are 9,2 mm, more CFM )

Haha zipties ftw 

That poll needs tighter values...

Between 70c-90c = 158F-194F

30C is THE-UNICORN
35C is THE-GUY-PAYED-TO-GROOM-THE-UNICORN
40C is THE-GUY-WHO-RIDES-A-UNICORN-TO-WORK
45C is A-FREAKING-LIAR
50C is DAMN-NEAR-A-FANTASY
55C is PERFECT
60C is GREAT
65C is IDEAL *** (My temps)
70C is NORMAL *** (My temps)
75C is OK
80C is WARM
85C is HOT
90C is NEAR-LIMIT
95C is AT-LIMIT
100C is SOON-TO-BE-DEAD
105C is DEAD

My temps, as reported by the GPU... (In order)
75, 75, 74, 73, 69, 69, 69, 69, 65, 63, 61, 61, 61, 61, 61, 61, 61, 61, 61, 61, 61, 61, 61, 60

Actual temp of all of them... (Majority of the day) These match the above cards ordered.
67, 67, 67, 66, 67, 67, 67, 62, 66, 66, 67, 67, 66, 65, 65, 65, 65, 65, 65, 65, 65, 65, 65, 66

Ambient room temp being 29c (86F) {outside air, I live in Florida, it is 25c(77F) at night, and humid.}
Nights the cards run about 2c-3c cooler.

This is "running for days" temps. The cards do NOT report the actual temperature. They are "ballpark" at best. They are not calibrated or high-quality temperature sensors. The most important thing you get from them, is "increase over ambient", once you know the "ambient" temperature of the running card. The higher the temp, the more inaccurate the sensor, but it is consistently inaccurate.

I wish someone made a temp-program that let you "calibrate" the output of the sensors. (Even if it only shows correctly in that program).

I write the "ambient" and "running" temp on each card, and give them an ID, to tell when they are drifting into dangerous grounds. (Along with a cheat-sheet, at my keyboard.)

P.S. These are all 7970 reference-design (HIS series {the hot ones}), and PowerColor (The cheap 2-fan models {the cooler ones})

When the stock fan dies you just remove it and put a 80 or 120mm case fan on it with zipties lol

Mostly 80s and 90s. I have a couple 6950s that chill out around 102 and dont seem to mind.

They might be worth $30 each 30 months from now so whatever...

PSA: if you're mining bitcoins, you can lower your temps significantly by lowering your memory clocks. Your mining speed will not be affected. In some cases, you will actually gain speed.

Why would I put my my $400 GPU at risk of overheating, just to save a $5 fan from burning out quicker? Crank those fans for as long as they'll go, and you can always replace them later. It's a lot more expensive to replace a GPU.

Do you guys think that it is better to sacrifice a few temp degrees to lower the fan speed?

I dont know why but i have the impression that my fan is much more vulnerable than my card.

So I guess I am killing my poor 6870's

first is at 965Mhz slight voltage underclocke to 1.104

95
98
101

The second is at 945 same voltage

95
96
94


Running 24/7 flawlessly stable even at those temps.

Ill take the advice and down clock them till I get the chance to move it to the AC Cooled server room.


edit
dropped to 850Mhz, voltage at 1.000

82
84
88

and

85
83
81


Stupid blower card fan higher then 48% get's loud. past 60% is brutal.

hehehe yeah!, same here with my 7850 when I was in windows, If I lower the ram below 700 Mhz I start to see all kind of crappy blocky shit on the screen heehe.
that didn't happens in linux thought, also in linux , I get an extra 25MH per card at same settings and speeds.

two 7850's bottom card 61 top card 68

stock fan settings.  side of case removed and stand alone room fan kind of blowing into the open case.

1250mhz over clock on bottom ram at 600 mhz stable 350Mhash

1100mhz oc on top card ram at 1200 mhz just barely stable 305Mhash   This card has my monitor on it and video is constantly crapping out with display blocks and stuff just messed up. Kind of like someone is playing an old atari asteroids game in the background. 

both sapphire dual fan 2gb cards

Also a 5770 that ran around 85. OC to about 940Mhz. 225 mhash. fan at 100%. WRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRR. Ran for six months straight like this stable no problems. Pulled the plastic cover and removed the hairball jamming the heatsink and redid the thermal paste and back to 70 deg. Just saying that a little maintenance can go a long way to cooler temperatures.  This has a single turbine style fan and is LOUD.

Using a combination of Trixx for the sapphire cards and AB for the 5770.  I like how AB put a little temp widget in my taskbar.

That's all

Sappire 5870, running at 900 memory at 300.  Temperature is 76°.

5 C Higher in the GPU, the VRMs are easy 10 to 15C above the GPU temperature, be very careful in LTC mining, specially if you OC the cards.

That is because LTC uses A LOT the memory and the internal bandwidth between memory and GPU engine, while BTC only uses the GPU, not memory.
also you will have 20 to 30% more power consumption in LTC than in BTC depending on settings.

SM411: check your VRM temperature, it's 85C, a little high, try to keep that one not more than 80C for safety, it is easier to replace the fan than blow the VRM and have to send the entire card to RMA so, if I have to choose between FAN life and CARD life, I will choose CARD life , you can even remove the plastic cover, both fans and buy 2 artctic cooling F8 CO fans and screw them to the heatsink directly

check the 80mm 4wire PWM fan: http://www.arctic.ac/en/p/cooling/case-fans/556/arctic-f-pwm-co-.html

I ordered a couple of them because one of my older cards have fan problems, I already fixed them but it's a short term solution, will have to replace them soon, two of these will do

Does anyone know why mining LTC GPU temps are much higher than mining BTC?

For instance my temps stay 80'ish Celcius when mining BTC.  When mining LTC my temps rise about 5C higher.

Im mining bitcoins. I lowered the temp to 85 by increasing fan speed. Can increase more, but fan is then running >80% and I guess that will ruin the fan over time.

http://gpuz.techpowerup.com/13/05/27/c3c.png

You are using scrypt right? (LTC) in scrypt there is usually 10 to 15 C higher in VRM than core temps, so you must get this value down at least 10 degrees, increase fan speed, relocate board to a cooler area or lower engine Mhz. (860 maybe)

I think that max temp for vrm are 100C (somebody correct me if I am wrong), but even if you are not, at these temps you are cooking the ram chips too!, which usually share the heatsink with the VRMs.

What is the maximum GPU#2 temp i can run?

Too high fix it somehow!

http://gpuz.techpowerup.com/13/05/26/dp9.png

Im a bit worried about GPU #2 temp, but doenst know what it is 

after reading your post, paranoia hit me again 
I will check all the screws of my hottest cards just in case.

My two Gigabite 7970 run 65-87 C.  One runs hotter than the other one.

The one that runs cool is consistently 72 C at stock speed.  The hot one runs consistently 80C at stock speed.  I swapped the two cards to different PCIe slots and get the same result. 

So I removed the hot 7970 yesterday and checked it carefully.  Long and behold, the screws for the VRM heatsink were loose.  What the h*ll?  It's a new one month old card.  So I tightened the thing down to finger tight.  (Not very scientific, I know.)

Then decided to remove the heatsink and see what's going on underneath.  There doesn't seem to be enough thermal paste between the 7970 core and the heatsink.  Removed the existing thermal paste, cleaned off with rubbing alcohol, reapplied with Arctic silver paste.  Now it runs consistently 4 degree C cooler. 

Totally agree with that!, VRM temperatures are the most dangerous ones.
Typically in 7950 / 7970 Sapphire Dual X Cards, they are 1 or 2 degrees above core temperature, so it is not dangerous, but this is on BTC mining.
In LTC is completely different, I have some cards than in LTC are 10 degrees above, others 15 degrees above core, that's why when I mine LTC y downclock the cards to 820 - 840 Mhz, and lower the voltage to 1.080, because VRM gets hot as hell in LTC! at stock speeds I have seen 86 degrees on the VRM YIKES!! (which they are also cooking the memory chips!).

I think that sapphire completely underestimated VRM cooling with the flat heatink that covers the VRMs and memory chips.

Heh 6990 on stock clocks runs upto 99 with other cards next to it

Underclocked & undervolted it and gained a few mh/s once the temps dropped to 70-80 degrees. cool hey

The VRM temps are what most people should be worried about more than the actual cores.  The VRMS almost always run hotter than the cores and they sometimes have bad cooling with the aftermarket designs.  Use GPUZ to see what they're running.  I'm OK up to 85C, beyond that is asking for trouble even in the short term.

70 and up to 80C is not dangerous for the GPU.
I've ran mine at ~70 for months.

Rig1:  3x 7950: coolest card 58C hottest 66C

Rig2:  4x 7850: coolest card 55C hottest 63C

Rig3:  2x 7970: coolest card 60C hottest 67C

Auto fan, Auto GPU clock (cgminer 311a (kano's mod), xubuntu 12.04 from persistent usb flashdrive)

Ambient temperature 25C, window open 

I won't sleep if any of my rigs touch 70C, no way GPU's are way too expensive to risk them like that IMHO.

Rig1
6950 @ 900Mhz 89c        Gigabyte tripple fan OC model
6950 @ 940Mhz 79c        Gigabyte tripple fan OC model
6570 @ 750Mhz 75c        Asus
5570 @ 690Mhz 92c        Shit (was used for bitcoin mining. Unused for litecoin due to stability)
Fans manual @ 70%

Rig2
6870 @ 960Mhz 92c       Sapphire OEM blower fan model
6870 @ 915Mhz 92c       Sapphire OEM blower fan model
Fans auto, stable around 52%

Work computer
R6570, Stock settings, fans on auto, 72c
Low aggression during work hours.

I'm doing my best to keep them under 75C.

Pansies.  Some of my multicard rigs idle at 57C on the hot card, so keeping under 90C is good....considering the ambient in the room was 103F.

That was with full AC running in a commercial building - outside ambient was 106F last week here in SoCal.

These damn voltage locked cards will the first to go 

60-75 average ranges depending on temp outside as Im drawing in cold air. Although one sunny hot afternoon my GF decided to close the door to the room my rig resides in cutting off the air circulation I had going on. When I noticed it was closed a few hours later I walked into what felt like a pre-heated oven and the temps of some of my cards were low 90s and one of my systems had already crashed. YIKES!

Living in Atlanta. In the daytime my xfire 7970s are averaging 78-82 C during the hottest part of the day. Been around 85F, 30C ambient temperature in the room where the machines are running. Temps outside also the same though very humid out there.

50 ambient??? Lucky... Mine stay pretty cool at 80c considering ambient is usually in the 90s

Sometimes it's under 70, sometimes it's over 70, depending on the ambient. When it hits 50F at night, they stay pretty chilly. When it hits 85F in the day, the fan is crankin and they get underclocked to stay below 73C.

The joys of living in New England. 

70-80 Celsius mark is fine, but mostly 70-90 if clocked and runing a miner 24/7 with extra flags