Max Temp?

[markabark]

I'm interested in people's opinion on the max temp to run a GPU at? To save on A/C, my miner is in the garage in the texas heat, and to get a good return I want to run the cards as hot as possible without killing them. My 5770 is the hottest of my cards and currently I throttle it automatically to 82C. Any opinions? Should I run it to where the fan is at 100%?

[1713866311]

1713866311

[1713866311]

1713866311

[1713866311]

1713866311

[Fiyasko]

Dual6870BE's(gaming rig) Very clusterfucked tight packed Both cards run @ 85° for the past three weeks mining, WHILE gaming, No issues
6990 Open case/20inch box fan= 87° running 1month no issues
5830 same case 78° 5months no issues

Dual 6870's overheat at 90° and auto shutdown, I wanna run them at dual 1ghz but i can only run them at dual 981mhz due to heat

[Elder III]

I am pleased as punch if they run under 80 C, I shut them down if they run over 90 C (sustained).  Those are my rules of thumb, some people push it more than others though.  How fast you have to run your fan depends on your ambient, your airflow (either in case or open air), and the actual cooler on the GPU itself.  Each one is different and only you know how much noise you can take. hehe  I don't like to run the fan at 100% fulltime, crappier fans do die out faster that way, I prefer 80-85% or less when possible.

[Fiyasko]

I am pleased as punch if they run under 80 C, I shut them down if they run over 90 C (sustained).  Those are my rules of thumb, some people push it more than others though.  How fast you have to run your fan depends on your ambient, your airflow (either in case or open air), and the actual cooler on the GPU itself.  Each one is different and only you know how much noise you can take. hehe  I don't like to run the fan at 100% fulltime, crappier fans do die out faster that way, I prefer 80-85% or less when possible.

Yeah, If you own A.)a "silent card" like i do, Both my 6870BE's are DualFan'd then noise is no issue, So 100% speed is almost always active.
If you B.)Own a "cold card" like i Used to, (shappire IceQ5tech dual 5770's) then noise is a MAJOR issue but temps are not.

Otherwise, Yeah, it's all about your case. How much noise it can dampen and how much air it can abuse

[markabark]

Otherwise, Yeah, it's all about your case. How much noise it can dampen and how much air it can abuse

My miner is in the garage so it's not an issue. I've decided to let the 5770 run up to 85C, which puts the fan at about 92% (it's about 40C in my garage right now) That lets both the 5770 and the 5830 run at their max stable overclock.

I'm planning on a 2nd 5830 (if anyone ever has them in stock) and I think at that point I'll have to go to an open case.

[CanaryInTheMine]

Dual6870BE's(gaming rig) Very clusterfucked tight packed Both cards run @ 85° for the past three weeks mining, WHILE gaming, No issues
6990 Open case/20inch box fan= 87° running 1month no issues
5830 same case 78° 5months no issues

Dual 6870's overheat at 90° and auto shutdown, I wanna run them at dual 1ghz but i can only run them at dual 981mhz due to heat

Which 6870 are you running?  are you able to underclock the memory to low setting?

[Fiyasko]

Dual6870BE's(gaming rig) Very clusterfucked tight packed Both cards run @ 85° for the past three weeks mining, WHILE gaming, No issues
6990 Open case/20inch box fan= 87° running 1month no issues
5830 same case 78° 5months no issues

Dual 6870's overheat at 90° and auto shutdown, I wanna run them at dual 1ghz but i can only run them at dual 981mhz due to heat

Which 6870 are you running?  are you able to underclock the memory to low setting?

http://www.newegg.ca/Product/Product.aspx?Item=N82E16814150520&Tpk=zdbc
I have not attempted to drop the memory clock on these cards since i use the Often for gaming. I will go try to drop the mem and start mining

Dropping the memory clock to 575 cause graphical errors, But no crash, Nor any Sign of crashing, just Blatent Card Display failure due to too slow of a mem clock im assuming
Both cards are running and linked crossfired

I am going to raise the memory clock to see if i can remove the "damages" since i feel that these "skews" are actaully Hurting my cards

[Fiyasko]

as i am typing the graphic "glitches' seems to be reduced, and they are Fixated in play where they are skewed,

Like in GUIminer theres a green 'glitch skew pixel spasm' that moves with the window and reacts asif it's One with the window and asif it was origianlly there by the creators, Moving windows over it and minimizing etc do not affect the 'skew'
the skew clears and reappears Perfectly when the window reloads

[Fiyasko]

As of right now, both 6870BE's are running at 1020mhz core and 700mhz mem, The temperature on both cards appears to be a stable 78°

Holy fucking failure, Cards crashed and reset, and one card is running at 1020 the other at 300, And i cant change the clocks, Uh oh

Clocks are responding again, I am going to attempt to run at 982core and 1000mem for a temp drop, Then try to mine while gaming.

On another note, My cards have gathered some dust and the overheat while gaming+mining @982core and i had to drop it to 962core

[CanaryInTheMine]

Dual6870BE's(gaming rig) Very clusterfucked tight packed Both cards run @ 85° for the past three weeks mining, WHILE gaming, No issues
6990 Open case/20inch box fan= 87° running 1month no issues
5830 same case 78° 5months no issues

Dual 6870's overheat at 90° and auto shutdown, I wanna run them at dual 1ghz but i can only run them at dual 981mhz due to heat

Which 6870 are you running?  are you able to underclock the memory to low setting?

http://www.newegg.ca/Product/Product.aspx?Item=N82E16814150520&Tpk=zdbc
I have not attempted to drop the memory clock on these cards since i use the Often for gaming. I will go try to drop the mem and start mining

Dropping the memory clock to 575 cause graphical errors, But no crash, Nor any Sign of crashing, just Blatent Card Display failure due to too slow of a mem clock im assuming
Both cards are running and linked crossfired

I am going to raise the memory clock to see if i can remove the "damages" since i feel that these "skews" are actaully Hurting my cards

I've been experimenting with this:

http://www.newegg.com/Product/Product.aspx?Item=N82E16814131446

and so far the memory clock speed can only be lowered by 100 less than clock speed.  ex: 900/800, 930/830 ect...
can't go less than by 100.

tried flashing bios.  seems some restriction is on the card and/or windows catalyst driver.  damn thing detects that bios was flashed and keeps rebooting computer.
have not tried under linux.  read some posts that implied that under linux flashed bios are working... if so, it's probably windows driver limitation then.

[Fiyasko]

.... Then how exactly am i mining at 1ghz with 650mem

Swapping to 982core and 1000mem and going to run CS:S video stress test at -f250
To test the theroy of 100+/-= no bugs

Too much heat. im expecting to crash and i dont want to deal with that crap. Flipping back to 962core 1150mem

[joulesbeef]

FYI running fans at 100% 24/7 will exponentially decrease the lifespan of the fan

[CanaryInTheMine]

.... Then how exactly am i mining at 1ghz with 650mem

Swapping to 982core and 1000mem and going to run CS:S video stress test at -f250
To test the theroy of 100+/-= no bugs

Too much heat. im expecting to crash and i dont want to deal with that crap. Flipping back to 962core 1150mem

it seems that dual-GPU cards have these limitations. 6990s have them, now this dual 6870 has it too. seems like any dual-gpu 6xxx series is limited to how low memory can be set to in comparison to the core clock speed.

i thought you meant you had a dual-gpu 6870, but i see now that you have 2 cards actually.

[Fiyasko]

FYI running fans at 100% 24/7 will exponentially decrease the lifespan of the fan

Without a doubt. And they spin slower too, So your 100% will eventually be like a 60%

.... Then how exactly am i mining at 1ghz with 650mem

Swapping to 982core and 1000mem and going to run CS:S video stress test at -f250
To test the theroy of 100+/-= no bugs

Too much heat. im expecting to crash and i dont want to deal with that crap. Flipping back to 962core 1150mem

it seems that dual-GPU cards have these limitations. 6990s have them, now this dual 6870 has it too. seems like any dual-gpu 6xxx series is limited to how low memory can be set to in comparison to the core clock speed.

i thought you meant you had a dual-gpu 6870, but i see now that you have 2 cards actually.

Uhhh buddy? I have a 6990 running at 930core 600mem, Your still wrong

982core and 1000mem showed no glitches and everything went OK, aside from Notably slower memclock loads, But the 5° drop is seemingly worth it in the mining senario

[joulesbeef]

Without a doubt. And they spin slower too, So your 100% will eventually be like a 60%

learnt the old fashion hard way. my 5000 rpms dropped to 2800 at 100%

[CanaryInTheMine]

FYI running fans at 100% 24/7 will exponentially decrease the lifespan of the fan

Without a doubt. And they spin slower too, So your 100% will eventually be like a 60%

.... Then how exactly am i mining at 1ghz with 650mem

Swapping to 982core and 1000mem and going to run CS:S video stress test at -f250
To test the theroy of 100+/-= no bugs

Too much heat. im expecting to crash and i dont want to deal with that crap. Flipping back to 962core 1150mem

it seems that dual-GPU cards have these limitations. 6990s have them, now this dual 6870 has it too. seems like any dual-gpu 6xxx series is limited to how low memory can be set to in comparison to the core clock speed.

i thought you meant you had a dual-gpu 6870, but i see now that you have 2 cards actually.

Uhhh buddy? I have a 6990 running at 930core 600mem, Your still wrong

982core and 1000mem showed no glitches and everything went OK, aside from Notably slower memclock loads, But the 5° drop is seemingly worth it in the mining senario

how did you set the 6990 to 600 memory?

[tritium]

are you all reading just the core temp? if you use something like everest/aida64 you can monitor the core temp, memory temp and shader temp. most of my cards are within 5C for the 3 temps but one of my 5850s is reading 75C core 72C mem and 95C shader, it seems fine and stable but i am a little concerned so im watching it closley

[CanaryInTheMine]

are you all reading just the core temp? if you use something like everest/aida64 you can monitor the core temp, memory temp and shader temp. most of my cards are within 5C for the 3 temps but one of my 5850s is reading 75C core 72C mem and 95C shader, it seems fine and stable but i am a little concerned so im watching it closley

dunno in their cases, but you can use GPU-Z, it's one of the favorites out there...

[Niann]

Radeon gpus in general run hot..
You should worry if you break the 85~90C barrier.

I feel safe below 80. But some ppl do run their cards hotter.

[chungenhung]

cards will run just fine as long as it is under 105C.
Know that all modern cards have auto throttle when it gets to danger zone, which is about close to 100C