Fan Speed vs Heat | Bitcointalksearch.org

Xephan

newbie

Activity: 42

Merit: 0

Quote from: SgtSpike on July 19, 2011, 02:23:19 PM

I would love to see proof of said "rule of thumb" actually being true in real life.

I don't have stats for GPU but back when overclocking CPU was the craze, there were quite a few users who had their highly overclocked and hot CPU die after a few months. If you dig up electronics and semicon datasheets, there are manufacturer charts for expected lifespan vs temperature delta.

Quote

IMO, heat is way overblown. GPU's are safe up to 90c EASILY, and I've only ever heard of a GPU dying from heat over 110c. Keep it below that, and in all likelyhood your card will be absolutely fine.

Which is why I said it might not matter if you intend to replace the GPU with a faster one as soon as they are available. The manufacturers usually set the shutdown/throttle point at some temperature they expect is the max allowable that doesn't degrade the expected lifespan near their warranty timespan.

However, the risk is that if you push it to that limit, the delay between thermal protections kicking in could still kill your GPU within a few months like those dead CPUs.

So while I think it's silly to spend money on massive cooling equipment to keep temperatures as low as possible, I also don't think it's good to run the GPU right up to the throttling/shutdown point. 10C is a healthy buffer so I try to keep things to 80C max.

SgtSpike

legendary

Activity: 1400

Merit: 1005

Quote from: Morebitcoinsplease on July 19, 2011, 03:18:19 PM

Quote from: SgtSpike on July 19, 2011, 02:23:19 PM

Quote from: Xephan on July 19, 2011, 01:53:30 PM

Quote from: gigabytecoin on July 19, 2011, 01:37:46 PM

The fact is, that running your computer 24/7 actually makes it last longer than turning it off and on.

The reason being is that heating and cooling of materials degrades them over time (just like bridges, roads, etc up here in Canada) ... if your rig simply stays at 80 degrees celcius forever it will hardly even know it was turned on.

The high speed of the fan could eventually wear down the ball bearings and stop spinning. Therefore it may be beneficial to run your fan at a lower speed and simply accept the higher temps.

Sorry but the video comment is misleading. While it's quite true that maintaining the chips at some constant temperature is likely to make it last longer than daily on/off cycles (due to expension/contraction cycles as well as inrush stress), the qualifier here is thermal damage at the temperature does not exceed the damage of the on/off cycles. Rule of thumb is every 10C halves the life of a component so at some point, the GPU despite being at a constant temperature, will die really fast. So it's always about finding a good balance point and if necessary, sacrifice some of the cheaper component's life in favour of the more expensive item.

Of course, if they do keep up improvements to the GPU, then all this simply may not matter. You just to keep it alive long enough to be profitable to replace it with a better unit Cheesy

I would love to see proof of said "rule of thumb" actually being true in real life.

IMO, heat is way overblown. GPU's are safe up to 90c EASILY, and I've only ever heard of a GPU dying from heat over 110c. Keep it below that, and in all likelyhood your card will be absolutely fine.

I have seen & tested with my GPUs when they even reach near 100C the PC auto shutdowns because it knows it over heating.

The range my cards go is 65C to 82C I don't even like the cards going into the 80's but thats how hard they are working. I think it depends on how long you hope to keep the cards running 24/7 for a year. At 90C or even 80C, the temperatures just don't sit well with me.

But I don't know how many hours these ATI cards were built to last. Guess that's why its important to keep that box and warranty Cheesy

And at 90c it will start clocking down the core and memory clocks to preserve the card as well, which is my point entirely. A GPU manufacturer would start it clocking down sooner if they thought it would result in more RMAs running at 90c.

CanaryInTheMine

donator

Activity: 2352

Merit: 1060

between a rock and a block!

Quote from: Bitcoin_Silver_Supply on July 15, 2011, 01:58:22 PM

I'm wondering whether one should go for higher fan speed or higher heat in trying to get the longest life span out of a miner. I currently have two 5830s running at 67C with fans at 76% and I'm wondering if it's better to run the fan harder or let the cards go a little hotter... Any opinions?

you're not squeezing as much as you can out of this. OC to the point where the temp is ~75C at least with same fan speed.

Before this card fails, you are likely to stop mining anyways, so get as much out of it as possible.

Morebitcoinsplease

member

Activity: 84

Merit: 10

Quote from: SgtSpike on July 19, 2011, 02:23:19 PM

Quote from: Xephan on July 19, 2011, 01:53:30 PM

Quote from: gigabytecoin on July 19, 2011, 01:37:46 PM

The fact is, that running your computer 24/7 actually makes it last longer than turning it off and on.

The reason being is that heating and cooling of materials degrades them over time (just like bridges, roads, etc up here in Canada) ... if your rig simply stays at 80 degrees celcius forever it will hardly even know it was turned on.

The high speed of the fan could eventually wear down the ball bearings and stop spinning. Therefore it may be beneficial to run your fan at a lower speed and simply accept the higher temps.

Sorry but the video comment is misleading. While it's quite true that maintaining the chips at some constant temperature is likely to make it last longer than daily on/off cycles (due to expension/contraction cycles as well as inrush stress), the qualifier here is thermal damage at the temperature does not exceed the damage of the on/off cycles. Rule of thumb is every 10C halves the life of a component so at some point, the GPU despite being at a constant temperature, will die really fast. So it's always about finding a good balance point and if necessary, sacrifice some of the cheaper component's life in favour of the more expensive item.

Of course, if they do keep up improvements to the GPU, then all this simply may not matter. You just to keep it alive long enough to be profitable to replace it with a better unit Cheesy

I would love to see proof of said "rule of thumb" actually being true in real life.

IMO, heat is way overblown. GPU's are safe up to 90c EASILY, and I've only ever heard of a GPU dying from heat over 110c. Keep it below that, and in all likelyhood your card will be absolutely fine.

I have seen & tested with my GPUs when they even reach near 100C the PC auto shutdowns because it knows it over heating.

The range my cards go is 65C to 82C I don't even like the cards going into the 80's but thats how hard they are working. I think it depends on how long you hope to keep the cards running 24/7 for a year. At 90C or even 80C, the temperatures just don't sit well with me.

But I don't know how many hours these ATI cards were built to last. Guess that's why its important to keep that box and warranty Cheesy

Bitcoin_Silver_Supply

sr. member

Activity: 258

Merit: 250

Based on the mix of comments I have gone with a moderate approach with slightly higher temps and lower fan speeds. On hot days I'm running about 73-75C with fan speeds between 62-65. I have had a few "device driver errors" causing the screen to go black and reset, but typically this has been while trying to watch a video with the GPUs going overclocked.

Also, having a mining PC beside your bed is a bitch.

SgtSpike

legendary

Activity: 1400

Merit: 1005

Quote from: Xephan on July 19, 2011, 01:53:30 PM

Quote from: gigabytecoin on July 19, 2011, 01:37:46 PM

The fact is, that running your computer 24/7 actually makes it last longer than turning it off and on.

The reason being is that heating and cooling of materials degrades them over time (just like bridges, roads, etc up here in Canada) ... if your rig simply stays at 80 degrees celcius forever it will hardly even know it was turned on.

The high speed of the fan could eventually wear down the ball bearings and stop spinning. Therefore it may be beneficial to run your fan at a lower speed and simply accept the higher temps.

Sorry but the video comment is misleading. While it's quite true that maintaining the chips at some constant temperature is likely to make it last longer than daily on/off cycles (due to expension/contraction cycles as well as inrush stress), the qualifier here is thermal damage at the temperature does not exceed the damage of the on/off cycles. Rule of thumb is every 10C halves the life of a component so at some point, the GPU despite being at a constant temperature, will die really fast. So it's always about finding a good balance point and if necessary, sacrifice some of the cheaper component's life in favour of the more expensive item.

Of course, if they do keep up improvements to the GPU, then all this simply may not matter. You just to keep it alive long enough to be profitable to replace it with a better unit Cheesy

I would love to see proof of said "rule of thumb" actually being true in real life.

IMO, heat is way overblown. GPU's are safe up to 90c EASILY, and I've only ever heard of a GPU dying from heat over 110c. Keep it below that, and in all likelyhood your card will be absolutely fine.

kokojie

legendary

Activity: 1806

Merit: 1003

Fan speed should be kept as low as possible. On the other hand, GPU don't really care about heat until it's over the threshold (somewhere around 108C)

Xephan

newbie

Activity: 42

Merit: 0

Quote from: gigabytecoin on July 19, 2011, 01:37:46 PM

The fact is, that running your computer 24/7 actually makes it last longer than turning it off and on.

The reason being is that heating and cooling of materials degrades them over time (just like bridges, roads, etc up here in Canada) ... if your rig simply stays at 80 degrees celcius forever it will hardly even know it was turned on.

The high speed of the fan could eventually wear down the ball bearings and stop spinning. Therefore it may be beneficial to run your fan at a lower speed and simply accept the higher temps.

Sorry but the video comment is misleading. While it's quite true that maintaining the chips at some constant temperature is likely to make it last longer than daily on/off cycles (due to expension/contraction cycles as well as inrush stress), the qualifier here is thermal damage at the temperature does not exceed the damage of the on/off cycles. Rule of thumb is every 10C halves the life of a component so at some point, the GPU despite being at a constant temperature, will die really fast. So it's always about finding a good balance point and if necessary, sacrifice some of the cheaper component's life in favour of the more expensive item.

Of course, if they do keep up improvements to the GPU, then all this simply may not matter. You just to keep it alive long enough to be profitable to replace it with a better unit Cheesy

gigabytecoin

sr. member

Activity: 280

Merit: 252

The fact is, that running your computer 24/7 actually makes it last longer than turning it off and on.

The reason being is that heating and cooling of materials degrades them over time (just like bridges, roads, etc up here in Canada) ... if your rig simply stays at 80 degrees celcius forever it will hardly even know it was turned on.

The high speed of the fan could eventually wear down the ball bearings and stop spinning. Therefore it may be beneficial to run your fan at a lower speed and simply accept the higher temps.

Xephan

newbie

Activity: 42

Merit: 0

Quote from: Bitcoin_Silver_Supply on July 15, 2011, 01:58:22 PM

I'm wondering whether one should go for higher fan speed or higher heat in trying to get the longest life span out of a miner. I currently have two 5830s running at 67C with fans at 76% and I'm wondering if it's better to run the fan harder or let the cards go a little hotter... Any opinions?

You can usually change a fan and/or heatsink for less than the price of the card so it's always better to trade off the fan's lifespan to extend the GPU. That said, 67C is really cool for these things. They are pretty OK to run at 70+

ensign_lee

hero member

Activity: 886

Merit: 500

I'd say leave it the way it is. You have a pretty nice balance of fan speed and temps atm.

My 5870, by comparison, bounces between 85 and 90 celcius at 96% fan speed.

being

newbie

Activity: 38

Merit: 0

Quote from: shotgun on July 18, 2011, 05:53:37 PM

Quote from: Leon on July 15, 2011, 04:28:31 PM

I use cables, keeps things way cooler and you don't even need to mess with fan speeds.

wat? cables?

yes, cables!
http://hacknmod.com/wp-content/uploads/2009/04/messy-cables.jpg

SgtSpike

legendary

Activity: 1400

Merit: 1005

Quote from: shotgun on July 18, 2011, 05:53:37 PM

Quote from: Leon on July 15, 2011, 04:28:31 PM

I use cables, keeps things way cooler and you don't even need to mess with fan speeds.

wat? cables?

:facepalm: Of course, cables! I should have thought of that for my cooling methods as well!

shotgun

member

Activity: 98

Merit: 11

Quote from: Leon on July 15, 2011, 04:28:31 PM

I use cables, keeps things way cooler and you don't even need to mess with fan speeds.

wat? cables?

Leon

newbie

Activity: 14

Merit: 0

I use cables, keeps things way cooler and you don't even need to mess with fan speeds.

SgtSpike

legendary

Activity: 1400

Merit: 1005

Quote from: Bitcoin_Silver_Supply on July 15, 2011, 03:31:51 PM

Quote from: SgtSpike on July 15, 2011, 02:58:38 PM

I just let the fan auto-adjust to temps by whatever the default settings are.... works fine for me.

I find the auto fan option on Trixx doesn't pick up fan speed enough at high temps. I came home once after my apartment heated up to see the temp at over 85C with the fan at a relatively low speed and that always scares me a bit.

The only time I'm concerned with temperature is if the GPU is clocking itself down because it got too hot (> 90c). Then I start losing Mhashes. But even an overclocked 5830 won't go past 90c at full load with the fan not spinning (it'll clock itself all the way down to 150/300 if it has to), so I wouldn't worry yourself too much. These things are good over 100c, probably up to 110 or 120c before you start seeing any hardware-killing issues crop up.

Bitcoin_Silver_Supply

sr. member

Activity: 258

Merit: 250

Quote from: SgtSpike on July 15, 2011, 02:58:38 PM

I just let the fan auto-adjust to temps by whatever the default settings are.... works fine for me.

I find the auto fan option on Trixx doesn't pick up fan speed enough at high temps. I came home once after my apartment heated up to see the temp at over 85C with the fan at a relatively low speed and that always scares me a bit.

Bitcoin_Silver_Supply

sr. member

Activity: 258

Merit: 250

Quote from: bcpokey on July 15, 2011, 02:54:16 PM

As with most things in life, the best is probably the middle ground. 67C is quite cool, and 76% fanspeed is rather high. Probably something like 75C and 40-50% fanspeed would last you quite a long time. You can reapply lubricant to the fans bearings every couple of years to increase the lifespan of the fan as well, assuming you do it properly, as that is the main issue fans running constantly run into. The MTBF for fans is typically something like 40,000 - 50,000 hours for cheapo fans like GPUs use, so expect 4-6 years of 24/7 use if properly kept up.

Good advice, thanks. I guess the main tradeoff is short term mhash vs longevity then. I can hold 74C at 61% fan speed which I assume is a tad better than where I was before.

Also, I do have a table fan blowing directly onto the cards but they are so close together that they're harder to cool.

SgtSpike

legendary

Activity: 1400

Merit: 1005

I just let the fan auto-adjust to temps by whatever the default settings are.... works fine for me.

bcpokey

hero member

Activity: 602

Merit: 500

As with most things in life, the best is probably the middle ground. 67C is quite cool, and 76% fanspeed is rather high. Probably something like 75C and 40-50% fanspeed would last you quite a long time. You can reapply lubricant to the fans bearings every couple of years to increase the lifespan of the fan as well, assuming you do it properly, as that is the main issue fans running constantly run into. The MTBF for fans is typically something like 40,000 - 50,000 hours for cheapo fans like GPUs use, so expect 4-6 years of 24/7 use if properly kept up.

Topic: Fan Speed vs Heat (Read 3263 times)