Iray and Cooling

areg5

Well, I was having a heck of a time with my system.  I had to throttle down the power on my cards or they would crash and it went to CPU rendering.  See my specs below.  Anyway, it was all crammed into an old Altec Mid tower case with 2 fans.  At 70 % power to the cards, my temps were 76 degrees on the 780 and 70 degrees on the 970, cpu at 75.  These are below the max by enough, or so I thought.  Well I got a Coolermaster Haf X full tower with 4 fans, and a coolermaster CPU fan with heatsink.  I put all of the components from my Altec into it.  Changed nothing else except the CPU fan is new.  I was using the stock one.  The result?  It runs flawlessly at 100%, with no crashes.  Both GPU's are running 10 degrees cooler (64 and 59 degrees).  Huge difference on the CPU with a better fan.  Under load, 54 degrees.

I had no idea that cooling is that important, but in my case it made an enormous difference.  My rendering times are much faster.  Scenes that used to take 40 minutes now take 25.  And all by cooling it off.  Who knew?  Lots of you guys, no doubt, but not me.  I thought the case was just an afterthought so it would look cooler.  Nope.  Don't underestimate the importance of cooling.  Rendering generates a lot of heat.

nicstt

Yeh, stock coolers on CPUs are usually a bad idea - at least for those using the computer for anything other than basic stuff.

Rendering is one of the most intesive processes a computer does; much more so than highend gaming.

... And it can go on for hours.

Inadequate cooling will seriously shorten the life of your computer.

areg5

nicstt said:

Yeh, stock coolers on CPUs are usually a bad idea - at least for those using the computer for anything other than basic stuff.

Rendering is one of the most intesive processes a computer does; much more so than highend gaming.

... And it can go on for hours.

Inadequate cooling will seriously shorten the life of your computer.

I've actually had the Coolermaster d92 CPU fan in my garage for months.  Ordered it when I ordered the motherboard and CPU.  It looked too big and clunky for the case I was using, and I just blew it off.  Shows what I knew.   I'm thinking of watercooling my next build.  Doing some research.  My next card will be a 1080 ti when it hits the market.  I know they have hybid 980's out now, so I may look into a hybrid one.  It'll cost bucks, but I hear they run between 20 and 30 degrees.

Kendall Sears

When dealing with GPUs, the PSU is the most important concern followed by cooling.  If the PSU is marginal or providing noisy power then the GPU performance will suffer.  Noisy power will increase heat as well as cause the circuitry to throttle back power usage to get a stable source.  It isn't necessarily about the size of the PSU, but how clean the output is.  Also a good PSU will create less ambient heat of its own and may actually pull excess heat from the system if its cooling setup is good.

Fluid cooling will not guarantee better cooling than forced air.  There are lots of variables.  The amount of time that the fluid spends in the exchanger versus the amount of airflow past the exchanger and other concerns.  Too much air velocity past the exchanger can actually ADD heat to the system (from friction and the air not having enough time to absorb enough heat) instead of removing it.  Too little airflow and not enough heat is removed.  The velocity of the air needs to change based on the amount of heat being generated as well as the ambient temperature outside the system.  Read the design specs CAREFULLY to determine if you can maintain the ambient room temperature necessary for the fluid cooler to actually function efficiently.  For every degree above nominal that the room's (or computer's enclosure -- ie under desk) temperature rises the amount of removed heat will decrease on a curve.  As the temperature rises in the environment the amount of heat retained in the fluid system increases non-linearly (the exact curve is determined by the design) causing the fluid to cumulatively gather heat to the point where the coolant starts to boil.  This creates voids in the system that can cause the pump to fail.

It is also entirely possible to draw heat INTO your system FROM the environment.  Once the ambient temp crosses a specific point (which is different for each design) the fluid will draw heat from the environment and introduce it into the computer's case.  This happens when the additive airflow from other fans in the case cause a draw approaching the output fans on the exchanger.  The temperature in the case becomes lower than the temperature in the external environment.  This leads the exchanger to gather heat rather than exhaust it.  The heat is then transferred into your components for the airflow in the case to try to remove.

The above is a very abbreviated explanation of the situation. There is a reason that datacenters containing racks and racks of huge servers use air cooling (and are extremely loud).  It is more efficient to maintain the environment at a temperature where direct air cooling can be used than to try to compensate for all of the variables introduced by fluid cooling.

Kendall

samurle

If you're worried about cooling, and have two monster graphics cards like the GTX 1080, I probably wouldn't place them next to each other.  One card would just be sucking in the hot air of the other.  If your motherboard has 3 or more x16 slots, then you could space them apart, giving them some air.

areg5

samurle said:

If you're worried about cooling, and have two monster graphics cards like the GTX 1080, I probably wouldn't place them next to each other.  One card would just be sucking in the hot air of the other.  If your motherboard has 3 or more x16 slots, then you could space them apart, giving them some air.

That's a really good idea.  I notice that  one of my cards runs hotter than the other, and they are next to each other. I don't know it that's becasue the 970 is newer and faster than the 780. Now, what is the deal when one slot is PCIexX16, the next is X8, the last X4?  are they differences that effect performance?

Kendall Sears said:

When dealing with GPUs, the PSU is the most important concern followed by cooling.  If the PSU is marginal or providing noisy power then the GPU performance will suffer.  Noisy power will increase heat as well as cause the circuitry to throttle back power usage to get a stable source.  It isn't necessarily about the size of the PSU, but how clean the output is.  Also a good PSU will create less ambient heat of its own and may actually pull excess heat from the system if its cooling setup is good.

Fluid cooling will not guarantee better cooling than forced air.  There are lots of variables.  The amount of time that the fluid spends in the exchanger versus the amount of airflow past the exchanger and other concerns.  Too much air velocity past the exchanger can actually ADD heat to the system (from friction and the air not having enough time to absorb enough heat) instead of removing it.  Too little airflow and not enough heat is removed.  The velocity of the air needs to change based on the amount of heat being generated as well as the ambient temperature outside the system.  Read the design specs CAREFULLY to determine if you can maintain the ambient room temperature necessary for the fluid cooler to actually function efficiently.  For every degree above nominal that the room's (or computer's enclosure -- ie under desk) temperature rises the amount of removed heat will decrease on a curve.  As the temperature rises in the environment the amount of heat retained in the fluid system increases non-linearly (the exact curve is determined by the design) causing the fluid to cumulatively gather heat to the point where the coolant starts to boil.  This creates voids in the system that can cause the pump to fail.

It is also entirely possible to draw heat INTO your system FROM the environment.  Once the ambient temp crosses a specific point (which is different for each design) the fluid will draw heat from the environment and introduce it into the computer's case.  This happens when the additive airflow from other fans in the case cause a draw approaching the output fans on the exchanger.  The temperature in the case becomes lower than the temperature in the external environment.  This leads the exchanger to gather heat rather than exhaust it.  The heat is then transferred into your components for the airflow in the case to try to remove.

The above is a very abbreviated explanation of the situation. There is a reason that datacenters containing racks and racks of huge servers use air cooling (and are extremely loud).  It is more efficient to maintain the environment at a temperature where direct air cooling can be used than to try to compensate for all of the variables introduced by fluid cooling.

Kendall

Makes sense. My rig isn't in an enclosure.  So if I plan on doing an animation, that looks like it would take 3 days, if water cooled the water would get hotter and hotter?  I would have thought a radiator would take care of that, in a custom water cooled rig.  Of course, the hybrid cards don't have a radiator, just what looks like a self contained resevoir and a fan.

Kendall Sears

areg5 said:

samurle said:

If you're worried about cooling, and have two monster graphics cards like the GTX 1080, I probably wouldn't place them next to each other.  One card would just be sucking in the hot air of the other.  If your motherboard has 3 or more x16 slots, then you could space them apart, giving them some air.

That's a really good idea.  I notice that  one of my cards runs hotter than the other, and they are next to each other. I don't know it that's becasue the 970 is newer and faster than the 780. Now, what is the deal when one slot is PCIexX16, the next is X8, the last X4?  are they differences that effect performance?

Kendall Sears said:

When dealing with GPUs, the PSU is the most important concern followed by cooling.  If the PSU is marginal or providing noisy power then the GPU performance will suffer.  Noisy power will increase heat as well as cause the circuitry to throttle back power usage to get a stable source.  It isn't necessarily about the size of the PSU, but how clean the output is.  Also a good PSU will create less ambient heat of its own and may actually pull excess heat from the system if its cooling setup is good.

Fluid cooling will not guarantee better cooling than forced air.  There are lots of variables.  The amount of time that the fluid spends in the exchanger versus the amount of airflow past the exchanger and other concerns.  Too much air velocity past the exchanger can actually ADD heat to the system (from friction and the air not having enough time to absorb enough heat) instead of removing it.  Too little airflow and not enough heat is removed.  The velocity of the air needs to change based on the amount of heat being generated as well as the ambient temperature outside the system.  Read the design specs CAREFULLY to determine if you can maintain the ambient room temperature necessary for the fluid cooler to actually function efficiently.  For every degree above nominal that the room's (or computer's enclosure -- ie under desk) temperature rises the amount of removed heat will decrease on a curve.  As the temperature rises in the environment the amount of heat retained in the fluid system increases non-linearly (the exact curve is determined by the design) causing the fluid to cumulatively gather heat to the point where the coolant starts to boil.  This creates voids in the system that can cause the pump to fail.

It is also entirely possible to draw heat INTO your system FROM the environment.  Once the ambient temp crosses a specific point (which is different for each design) the fluid will draw heat from the environment and introduce it into the computer's case.  This happens when the additive airflow from other fans in the case cause a draw approaching the output fans on the exchanger.  The temperature in the case becomes lower than the temperature in the external environment.  This leads the exchanger to gather heat rather than exhaust it.  The heat is then transferred into your components for the airflow in the case to try to remove.

The above is a very abbreviated explanation of the situation. There is a reason that datacenters containing racks and racks of huge servers use air cooling (and are extremely loud).  It is more efficient to maintain the environment at a temperature where direct air cooling can be used than to try to compensate for all of the variables introduced by fluid cooling.

Kendall

Makes sense. My rig isn't in an enclosure.  So if I plan on doing an animation, that looks like it would take 3 days, if water cooled the water would get hotter and hotter?  I would have thought a radiator would take care of that, in a custom water cooled rig.  Of course, the hybrid cards don't have a radiator, just what looks like a self contained resevoir and a fan.

As I wrote, it depends on the design.  A radiator is a type of exchanger.  Very few systems use straight H20, with a majority using a coolant like Ethylene Glycol or similar.  Whether the coolant gets 'hotter' will depend on a number of factors (variables).  Each additive system and/or heat source added to the cooling system adds more variables.  For instance, an exchanger external to the enclosure is more likely to perform better than one mounted inside the enclosure, but is not guaranteed.  For instance, if the heat from the exchanger somehow gets recycled into the airflow into the enclosure and/or back through the exchanger then residual heat will accumulate into the system.  An example:  Lets say that the radiator is in the same room as the system itself.  If the room is small or not well ventilated, then the heat from the exchanger will increase the temperature of the air in the room, thus causing the system to draw in that heated air.  Now the air pulled into the system is (nominal+x degrees).  This air is subsequently used to push through the exchanger and heated another y degrees which heats the room another x degrees.  Eventually, the heat in the room will exceed the point where the fluid can exude heat and the fluid starts to approach the boiling point.  This will happen first in the tubing carrying the coolant, followed by the exchanger.  Even if the coolant doesn't boil causing voids, the residual heat remains in the system causing the heat in the components to not be removed.  There are also issues of pressure rises and degraded performance of mechanical parts (fans for the exchanger slowing down due to age) that must be monitored.

This is VERY much like an automobile overheating on a hot day.  When the temperature outdoors gets into the 90's F, vehicles start overheating even though 90F is far below the boiling point of the coolant.  However, the heat produced by the engine accumulates to the point where the coolant can no longer transfer the heat to the air and the coolant boils, the pump loses prime, and things go horribly bad.  The same thing can happen in a computer using fluid cooling especially in a situation where there are multiple sources of heat.

Soooooo..... what this says is that if you intend on using fluid cooling, then you must take into consideration that the coolant adds an additional complexity to the planning.  In the end, you are still using forced air from the environment to remove the heat.  What fluid cooling is SUPPOSED to do is allow the removal of the heat from the environment that the equipment is in by embedding it into a more dense meduim that can be more easily transported to a location away from the equipment.  If, however, you are dumping the heat right back into the same environment that the equipment is in, then you are doing no better than using the air to cool the parts directly.  Mainframes that are fluid cooled have the fluid transported to the outside of the building where the heat is dumped into the air via convection (ponds) or large evaporative exchangers.  The main advantage to using fluid cooling in a PC/Workstation is to allow the heat to be drawn away from the components to where a more efficient heat removal setup can be utilized.  If you can maintain the environment at nominal temps and the system has no occlusions that would prevent the use of efficient heat sinks, then the additional costs/headaches of a fluid cooling system may not be worth it.  However, if you have a situation where the position of your CPU/GPU's is not conducive to using "large" sinks and/or fans, or you have a desire to try to reduce the noise created by multiple fans, then the additional cost/maintenance may be worth the effort.

Kendall

areg5

Kendall Sears said:

areg5 said:

samurle said:

If you're worried about cooling, and have two monster graphics cards like the GTX 1080, I probably wouldn't place them next to each other.  One card would just be sucking in the hot air of the other.  If your motherboard has 3 or more x16 slots, then you could space them apart, giving them some air.

That's a really good idea.  I notice that  one of my cards runs hotter than the other, and they are next to each other. I don't know it that's becasue the 970 is newer and faster than the 780. Now, what is the deal when one slot is PCIexX16, the next is X8, the last X4?  are they differences that effect performance?

Kendall Sears said:

When dealing with GPUs, the PSU is the most important concern followed by cooling.  If the PSU is marginal or providing noisy power then the GPU performance will suffer.  Noisy power will increase heat as well as cause the circuitry to throttle back power usage to get a stable source.  It isn't necessarily about the size of the PSU, but how clean the output is.  Also a good PSU will create less ambient heat of its own and may actually pull excess heat from the system if its cooling setup is good.

Fluid cooling will not guarantee better cooling than forced air.  There are lots of variables.  The amount of time that the fluid spends in the exchanger versus the amount of airflow past the exchanger and other concerns.  Too much air velocity past the exchanger can actually ADD heat to the system (from friction and the air not having enough time to absorb enough heat) instead of removing it.  Too little airflow and not enough heat is removed.  The velocity of the air needs to change based on the amount of heat being generated as well as the ambient temperature outside the system.  Read the design specs CAREFULLY to determine if you can maintain the ambient room temperature necessary for the fluid cooler to actually function efficiently.  For every degree above nominal that the room's (or computer's enclosure -- ie under desk) temperature rises the amount of removed heat will decrease on a curve.  As the temperature rises in the environment the amount of heat retained in the fluid system increases non-linearly (the exact curve is determined by the design) causing the fluid to cumulatively gather heat to the point where the coolant starts to boil.  This creates voids in the system that can cause the pump to fail.

It is also entirely possible to draw heat INTO your system FROM the environment.  Once the ambient temp crosses a specific point (which is different for each design) the fluid will draw heat from the environment and introduce it into the computer's case.  This happens when the additive airflow from other fans in the case cause a draw approaching the output fans on the exchanger.  The temperature in the case becomes lower than the temperature in the external environment.  This leads the exchanger to gather heat rather than exhaust it.  The heat is then transferred into your components for the airflow in the case to try to remove.

The above is a very abbreviated explanation of the situation. There is a reason that datacenters containing racks and racks of huge servers use air cooling (and are extremely loud).  It is more efficient to maintain the environment at a temperature where direct air cooling can be used than to try to compensate for all of the variables introduced by fluid cooling.

Kendall

Makes sense. My rig isn't in an enclosure.  So if I plan on doing an animation, that looks like it would take 3 days, if water cooled the water would get hotter and hotter?  I would have thought a radiator would take care of that, in a custom water cooled rig.  Of course, the hybrid cards don't have a radiator, just what looks like a self contained resevoir and a fan.

As I wrote, it depends on the design.  A radiator is a type of exchanger.  Very few systems use straight H20, with a majority using a coolant like Ethylene Glycol or similar.  Whether the coolant gets 'hotter' will depend on a number of factors (variables).  Each additive system and/or heat source added to the cooling system adds more variables.  For instance, an exchanger external to the enclosure is more likely to perform better than one mounted inside the enclosure, but is not guaranteed.  For instance, if the heat from the exchanger somehow gets recycled into the airflow into the enclosure and/or back through the exchanger then residual heat will accumulate into the system.  An example:  Lets say that the radiator is in the same room as the system itself.  If the room is small or not well ventilated, then the heat from the exchanger will increase the temperature of the air in the room, thus causing the system to draw in that heated air.  Now the air pulled into the system is (nominal+x degrees).  This air is subsequently used to push through the exchanger and heated another y degrees which heats the room another x degrees.  Eventually, the heat in the room will exceed the point where the fluid can exude heat and the fluid starts to approach the boiling point.  This will happen first in the tubing carrying the coolant, followed by the exchanger.  Even if the coolant doesn't boil causing voids, the residual heat remains in the system causing the heat in the components to not be removed.  There are also issues of pressure rises and degraded performance of mechanical parts (fans for the exchanger slowing down due to age) that must be monitored.

This is VERY much like an automobile overheating on a hot day.  When the temperature outdoors gets into the 90's F, vehicles start overheating even though 90F is far below the boiling point of the coolant.  However, the heat produced by the engine accumulates to the point where the coolant can no longer transfer the heat to the air and the coolant boils, the pump loses prime, and things go horribly bad.  The same thing can happen in a computer using fluid cooling especially in a situation where there are multiple sources of heat.

Soooooo..... what this says is that if you intend on using fluid cooling, then you must take into consideration that the coolant adds an additional complexity to the planning.  In the end, you are still using forced air from the environment to remove the heat.  What fluid cooling is SUPPOSED to do is allow the removal of the heat from the environment that the equipment is in by embedding it into a more dense meduim that can be more easily transported to a location away from the equipment.  If, however, you are dumping the heat right back into the same environment that the equipment is in, then you are doing no better than using the air to cool the parts directly.  Mainframes that are fluid cooled have the fluid transported to the outside of the building where the heat is dumped into the air via convection (ponds) or large evaporative exchangers.  The main advantage to using fluid cooling in a PC/Workstation is to allow the heat to be drawn away from the components to where a more efficient heat removal setup can be utilized.  If you can maintain the environment at nominal temps and the system has no occlusions that would prevent the use of efficient heat sinks, then the additional costs/headaches of a fluid cooling system may not be worth it.  However, if you have a situation where the position of your CPU/GPU's is not conducive to using "large" sinks and/or fans, or you have a desire to try to reduce the noise created by multiple fans, then the additional cost/maintenance may be worth the effort.

Kendall

I had thought about ethylene glycol, but the how-to videos I've seen all talk about using distilled water.  My temp readings inside the box tend to hover around 40 c under load, so that seems pretty good.  I'm thinking that as the system is stable, and from what you're telling me, air cooling is the way to go.  Now, what effect would having a faster processor have?  I mean, if the rendering is going through the GPU's, does upgrading form a 4790 give me any significantly faster rendering times?  Also, when I chose the motherboard, I basically was looking for the cheapest one that would hold the processor and cards.  Is there any advantage to a costlier one?  I currently have a Gigabyte Z97.

 

I know those aren't cooling questions, but you seem pretty well versed.

Kendall Sears

areg5 said:

Kendall Sears said:

areg5 said:

samurle said:

If you're worried about cooling, and have two monster graphics cards like the GTX 1080, I probably wouldn't place them next to each other.  One card would just be sucking in the hot air of the other.  If your motherboard has 3 or more x16 slots, then you could space them apart, giving them some air.

That's a really good idea.  I notice that  one of my cards runs hotter than the other, and they are next to each other. I don't know it that's becasue the 970 is newer and faster than the 780. Now, what is the deal when one slot is PCIexX16, the next is X8, the last X4?  are they differences that effect performance?

Kendall Sears said:

When dealing with GPUs, the PSU is the most important concern followed by cooling.  If the PSU is marginal or providing noisy power then the GPU performance will suffer.  Noisy power will increase heat as well as cause the circuitry to throttle back power usage to get a stable source.  It isn't necessarily about the size of the PSU, but how clean the output is.  Also a good PSU will create less ambient heat of its own and may actually pull excess heat from the system if its cooling setup is good.

Fluid cooling will not guarantee better cooling than forced air.  There are lots of variables.  The amount of time that the fluid spends in the exchanger versus the amount of airflow past the exchanger and other concerns.  Too much air velocity past the exchanger can actually ADD heat to the system (from friction and the air not having enough time to absorb enough heat) instead of removing it.  Too little airflow and not enough heat is removed.  The velocity of the air needs to change based on the amount of heat being generated as well as the ambient temperature outside the system.  Read the design specs CAREFULLY to determine if you can maintain the ambient room temperature necessary for the fluid cooler to actually function efficiently.  For every degree above nominal that the room's (or computer's enclosure -- ie under desk) temperature rises the amount of removed heat will decrease on a curve.  As the temperature rises in the environment the amount of heat retained in the fluid system increases non-linearly (the exact curve is determined by the design) causing the fluid to cumulatively gather heat to the point where the coolant starts to boil.  This creates voids in the system that can cause the pump to fail.

It is also entirely possible to draw heat INTO your system FROM the environment.  Once the ambient temp crosses a specific point (which is different for each design) the fluid will draw heat from the environment and introduce it into the computer's case.  This happens when the additive airflow from other fans in the case cause a draw approaching the output fans on the exchanger.  The temperature in the case becomes lower than the temperature in the external environment.  This leads the exchanger to gather heat rather than exhaust it.  The heat is then transferred into your components for the airflow in the case to try to remove.

The above is a very abbreviated explanation of the situation. There is a reason that datacenters containing racks and racks of huge servers use air cooling (and are extremely loud).  It is more efficient to maintain the environment at a temperature where direct air cooling can be used than to try to compensate for all of the variables introduced by fluid cooling.

Kendall

Makes sense. My rig isn't in an enclosure.  So if I plan on doing an animation, that looks like it would take 3 days, if water cooled the water would get hotter and hotter?  I would have thought a radiator would take care of that, in a custom water cooled rig.  Of course, the hybrid cards don't have a radiator, just what looks like a self contained resevoir and a fan.

As I wrote, it depends on the design.  A radiator is a type of exchanger.  Very few systems use straight H20, with a majority using a coolant like Ethylene Glycol or similar.  Whether the coolant gets 'hotter' will depend on a number of factors (variables).  Each additive system and/or heat source added to the cooling system adds more variables.  For instance, an exchanger external to the enclosure is more likely to perform better than one mounted inside the enclosure, but is not guaranteed.  For instance, if the heat from the exchanger somehow gets recycled into the airflow into the enclosure and/or back through the exchanger then residual heat will accumulate into the system.  An example:  Lets say that the radiator is in the same room as the system itself.  If the room is small or not well ventilated, then the heat from the exchanger will increase the temperature of the air in the room, thus causing the system to draw in that heated air.  Now the air pulled into the system is (nominal+x degrees).  This air is subsequently used to push through the exchanger and heated another y degrees which heats the room another x degrees.  Eventually, the heat in the room will exceed the point where the fluid can exude heat and the fluid starts to approach the boiling point.  This will happen first in the tubing carrying the coolant, followed by the exchanger.  Even if the coolant doesn't boil causing voids, the residual heat remains in the system causing the heat in the components to not be removed.  There are also issues of pressure rises and degraded performance of mechanical parts (fans for the exchanger slowing down due to age) that must be monitored.

This is VERY much like an automobile overheating on a hot day.  When the temperature outdoors gets into the 90's F, vehicles start overheating even though 90F is far below the boiling point of the coolant.  However, the heat produced by the engine accumulates to the point where the coolant can no longer transfer the heat to the air and the coolant boils, the pump loses prime, and things go horribly bad.  The same thing can happen in a computer using fluid cooling especially in a situation where there are multiple sources of heat.

Soooooo..... what this says is that if you intend on using fluid cooling, then you must take into consideration that the coolant adds an additional complexity to the planning.  In the end, you are still using forced air from the environment to remove the heat.  What fluid cooling is SUPPOSED to do is allow the removal of the heat from the environment that the equipment is in by embedding it into a more dense meduim that can be more easily transported to a location away from the equipment.  If, however, you are dumping the heat right back into the same environment that the equipment is in, then you are doing no better than using the air to cool the parts directly.  Mainframes that are fluid cooled have the fluid transported to the outside of the building where the heat is dumped into the air via convection (ponds) or large evaporative exchangers.  The main advantage to using fluid cooling in a PC/Workstation is to allow the heat to be drawn away from the components to where a more efficient heat removal setup can be utilized.  If you can maintain the environment at nominal temps and the system has no occlusions that would prevent the use of efficient heat sinks, then the additional costs/headaches of a fluid cooling system may not be worth it.  However, if you have a situation where the position of your CPU/GPU's is not conducive to using "large" sinks and/or fans, or you have a desire to try to reduce the noise created by multiple fans, then the additional cost/maintenance may be worth the effort.

Kendall

I had thought about ethylene glycol, but the how-to videos I've seen all talk about using distilled water.  My temp readings inside the box tend to hover around 40 c under load, so that seems pretty good.  I'm thinking that as the system is stable, and from what you're telling me, air cooling is the way to go.  Now, what effect would having a faster processor have?  I mean, if the rendering is going through the GPU's, does upgrading form a 4790 give me any significantly faster rendering times?  Also, when I chose the motherboard, I basically was looking for the cheapest one that would hold the processor and cards.  Is there any advantage to a costlier one?  I currently have a Gigabyte Z97.

 

I know those aren't cooling questions, but you seem pretty well versed.

If you're using Iray exclusively and you have lots of CUDA cores and more VRAM than you will need, then a "higher performance" CPU will be of limited utility.  HOWEVER, if you are going to run into a situation where a scene will exceed the VRAM on your GPUs, then you'll be glad to have a more powerful CPU when the render falls back to it.  There are still plenty of computations that occur on the CPU, but if you are already happy with the responsiveness of your system then the Return-On-Investment for a newer CPU may be low to zero.

It sounds like you've already provided a good cooling situation.  If your PSU is good, then I'd say spend your money on other things.

Kendall

namffuak

areg5 said:

Kendall Sears said:

areg5 said:

samurle said:

If you're worried about cooling, and have two monster graphics cards like the GTX 1080, I probably wouldn't place them next to each other.  One card would just be sucking in the hot air of the other.  If your motherboard has 3 or more x16 slots, then you could space them apart, giving them some air.

That's a really good idea.  I notice that  one of my cards runs hotter than the other, and they are next to each other. I don't know it that's becasue the 970 is newer and faster than the 780. Now, what is the deal when one slot is PCIexX16, the next is X8, the last X4?  are they differences that effect performance?

Kendall Sears said:

When dealing with GPUs, the PSU is the most important concern followed by cooling.  If the PSU is marginal or providing noisy power then the GPU performance will suffer.  Noisy power will increase heat as well as cause the circuitry to throttle back power usage to get a stable source.  It isn't necessarily about the size of the PSU, but how clean the output is.  Also a good PSU will create less ambient heat of its own and may actually pull excess heat from the system if its cooling setup is good.

Fluid cooling will not guarantee better cooling than forced air.  There are lots of variables.  The amount of time that the fluid spends in the exchanger versus the amount of airflow past the exchanger and other concerns.  Too much air velocity past the exchanger can actually ADD heat to the system (from friction and the air not having enough time to absorb enough heat) instead of removing it.  Too little airflow and not enough heat is removed.  The velocity of the air needs to change based on the amount of heat being generated as well as the ambient temperature outside the system.  Read the design specs CAREFULLY to determine if you can maintain the ambient room temperature necessary for the fluid cooler to actually function efficiently.  For every degree above nominal that the room's (or computer's enclosure -- ie under desk) temperature rises the amount of removed heat will decrease on a curve.  As the temperature rises in the environment the amount of heat retained in the fluid system increases non-linearly (the exact curve is determined by the design) causing the fluid to cumulatively gather heat to the point where the coolant starts to boil.  This creates voids in the system that can cause the pump to fail.

It is also entirely possible to draw heat INTO your system FROM the environment.  Once the ambient temp crosses a specific point (which is different for each design) the fluid will draw heat from the environment and introduce it into the computer's case.  This happens when the additive airflow from other fans in the case cause a draw approaching the output fans on the exchanger.  The temperature in the case becomes lower than the temperature in the external environment.  This leads the exchanger to gather heat rather than exhaust it.  The heat is then transferred into your components for the airflow in the case to try to remove.

The above is a very abbreviated explanation of the situation. There is a reason that datacenters containing racks and racks of huge servers use air cooling (and are extremely loud).  It is more efficient to maintain the environment at a temperature where direct air cooling can be used than to try to compensate for all of the variables introduced by fluid cooling.

Kendall

Makes sense. My rig isn't in an enclosure.  So if I plan on doing an animation, that looks like it would take 3 days, if water cooled the water would get hotter and hotter?  I would have thought a radiator would take care of that, in a custom water cooled rig.  Of course, the hybrid cards don't have a radiator, just what looks like a self contained resevoir and a fan.

As I wrote, it depends on the design.  A radiator is a type of exchanger.  Very few systems use straight H20, with a majority using a coolant like Ethylene Glycol or similar.  Whether the coolant gets 'hotter' will depend on a number of factors (variables).  Each additive system and/or heat source added to the cooling system adds more variables.  For instance, an exchanger external to the enclosure is more likely to perform better than one mounted inside the enclosure, but is not guaranteed.  For instance, if the heat from the exchanger somehow gets recycled into the airflow into the enclosure and/or back through the exchanger then residual heat will accumulate into the system.  An example:  Lets say that the radiator is in the same room as the system itself.  If the room is small or not well ventilated, then the heat from the exchanger will increase the temperature of the air in the room, thus causing the system to draw in that heated air.  Now the air pulled into the system is (nominal+x degrees).  This air is subsequently used to push through the exchanger and heated another y degrees which heats the room another x degrees.  Eventually, the heat in the room will exceed the point where the fluid can exude heat and the fluid starts to approach the boiling point.  This will happen first in the tubing carrying the coolant, followed by the exchanger.  Even if the coolant doesn't boil causing voids, the residual heat remains in the system causing the heat in the components to not be removed.  There are also issues of pressure rises and degraded performance of mechanical parts (fans for the exchanger slowing down due to age) that must be monitored.

This is VERY much like an automobile overheating on a hot day.  When the temperature outdoors gets into the 90's F, vehicles start overheating even though 90F is far below the boiling point of the coolant.  However, the heat produced by the engine accumulates to the point where the coolant can no longer transfer the heat to the air and the coolant boils, the pump loses prime, and things go horribly bad.  The same thing can happen in a computer using fluid cooling especially in a situation where there are multiple sources of heat.

Soooooo..... what this says is that if you intend on using fluid cooling, then you must take into consideration that the coolant adds an additional complexity to the planning.  In the end, you are still using forced air from the environment to remove the heat.  What fluid cooling is SUPPOSED to do is allow the removal of the heat from the environment that the equipment is in by embedding it into a more dense meduim that can be more easily transported to a location away from the equipment.  If, however, you are dumping the heat right back into the same environment that the equipment is in, then you are doing no better than using the air to cool the parts directly.  Mainframes that are fluid cooled have the fluid transported to the outside of the building where the heat is dumped into the air via convection (ponds) or large evaporative exchangers.  The main advantage to using fluid cooling in a PC/Workstation is to allow the heat to be drawn away from the components to where a more efficient heat removal setup can be utilized.  If you can maintain the environment at nominal temps and the system has no occlusions that would prevent the use of efficient heat sinks, then the additional costs/headaches of a fluid cooling system may not be worth it.  However, if you have a situation where the position of your CPU/GPU's is not conducive to using "large" sinks and/or fans, or you have a desire to try to reduce the noise created by multiple fans, then the additional cost/maintenance may be worth the effort.

Kendall

I had thought about ethylene glycol, but the how-to videos I've seen all talk about using distilled water.  My temp readings inside the box tend to hover around 40 c under load, so that seems pretty good.  I'm thinking that as the system is stable, and from what you're telling me, air cooling is the way to go.  Now, what effect would having a faster processor have?  I mean, if the rendering is going through the GPU's, does upgrading form a 4790 give me any significantly faster rendering times?  Also, when I chose the motherboard, I basically was looking for the cheapest one that would hold the processor and cards.  Is there any advantage to a costlier one?  I currently have a Gigabyte Z97.

 

I know those aren't cooling questions, but you seem pretty well versed.

I have an I7-5930; 6 core 3.5 GHz cpu. I also have a scene that takes 57 minutes to render with my 980ti. If I add the cpu the render time drops to 52 minutes and all 12 cores are 100% busy for that 52 minutes. You may draw your own conclusion. My system is completely air cooled and during that render the cpu temp peaks at 56 C and the 980ti at 59 C.

areg5

Kendall Sears said:

areg5 said:

Kendall Sears said:

areg5 said:

samurle said:

If you're worried about cooling, and have two monster graphics cards like the GTX 1080, I probably wouldn't place them next to each other.  One card would just be sucking in the hot air of the other.  If your motherboard has 3 or more x16 slots, then you could space them apart, giving them some air.

That's a really good idea.  I notice that  one of my cards runs hotter than the other, and they are next to each other. I don't know it that's becasue the 970 is newer and faster than the 780. Now, what is the deal when one slot is PCIexX16, the next is X8, the last X4?  are they differences that effect performance?

Kendall Sears said:

When dealing with GPUs, the PSU is the most important concern followed by cooling.  If the PSU is marginal or providing noisy power then the GPU performance will suffer.  Noisy power will increase heat as well as cause the circuitry to throttle back power usage to get a stable source.  It isn't necessarily about the size of the PSU, but how clean the output is.  Also a good PSU will create less ambient heat of its own and may actually pull excess heat from the system if its cooling setup is good.

Fluid cooling will not guarantee better cooling than forced air.  There are lots of variables.  The amount of time that the fluid spends in the exchanger versus the amount of airflow past the exchanger and other concerns.  Too much air velocity past the exchanger can actually ADD heat to the system (from friction and the air not having enough time to absorb enough heat) instead of removing it.  Too little airflow and not enough heat is removed.  The velocity of the air needs to change based on the amount of heat being generated as well as the ambient temperature outside the system.  Read the design specs CAREFULLY to determine if you can maintain the ambient room temperature necessary for the fluid cooler to actually function efficiently.  For every degree above nominal that the room's (or computer's enclosure -- ie under desk) temperature rises the amount of removed heat will decrease on a curve.  As the temperature rises in the environment the amount of heat retained in the fluid system increases non-linearly (the exact curve is determined by the design) causing the fluid to cumulatively gather heat to the point where the coolant starts to boil.  This creates voids in the system that can cause the pump to fail.

It is also entirely possible to draw heat INTO your system FROM the environment.  Once the ambient temp crosses a specific point (which is different for each design) the fluid will draw heat from the environment and introduce it into the computer's case.  This happens when the additive airflow from other fans in the case cause a draw approaching the output fans on the exchanger.  The temperature in the case becomes lower than the temperature in the external environment.  This leads the exchanger to gather heat rather than exhaust it.  The heat is then transferred into your components for the airflow in the case to try to remove.

The above is a very abbreviated explanation of the situation. There is a reason that datacenters containing racks and racks of huge servers use air cooling (and are extremely loud).  It is more efficient to maintain the environment at a temperature where direct air cooling can be used than to try to compensate for all of the variables introduced by fluid cooling.

Kendall

Makes sense. My rig isn't in an enclosure.  So if I plan on doing an animation, that looks like it would take 3 days, if water cooled the water would get hotter and hotter?  I would have thought a radiator would take care of that, in a custom water cooled rig.  Of course, the hybrid cards don't have a radiator, just what looks like a self contained resevoir and a fan.

As I wrote, it depends on the design.  A radiator is a type of exchanger.  Very few systems use straight H20, with a majority using a coolant like Ethylene Glycol or similar.  Whether the coolant gets 'hotter' will depend on a number of factors (variables).  Each additive system and/or heat source added to the cooling system adds more variables.  For instance, an exchanger external to the enclosure is more likely to perform better than one mounted inside the enclosure, but is not guaranteed.  For instance, if the heat from the exchanger somehow gets recycled into the airflow into the enclosure and/or back through the exchanger then residual heat will accumulate into the system.  An example:  Lets say that the radiator is in the same room as the system itself.  If the room is small or not well ventilated, then the heat from the exchanger will increase the temperature of the air in the room, thus causing the system to draw in that heated air.  Now the air pulled into the system is (nominal+x degrees).  This air is subsequently used to push through the exchanger and heated another y degrees which heats the room another x degrees.  Eventually, the heat in the room will exceed the point where the fluid can exude heat and the fluid starts to approach the boiling point.  This will happen first in the tubing carrying the coolant, followed by the exchanger.  Even if the coolant doesn't boil causing voids, the residual heat remains in the system causing the heat in the components to not be removed.  There are also issues of pressure rises and degraded performance of mechanical parts (fans for the exchanger slowing down due to age) that must be monitored.

This is VERY much like an automobile overheating on a hot day.  When the temperature outdoors gets into the 90's F, vehicles start overheating even though 90F is far below the boiling point of the coolant.  However, the heat produced by the engine accumulates to the point where the coolant can no longer transfer the heat to the air and the coolant boils, the pump loses prime, and things go horribly bad.  The same thing can happen in a computer using fluid cooling especially in a situation where there are multiple sources of heat.

Soooooo..... what this says is that if you intend on using fluid cooling, then you must take into consideration that the coolant adds an additional complexity to the planning.  In the end, you are still using forced air from the environment to remove the heat.  What fluid cooling is SUPPOSED to do is allow the removal of the heat from the environment that the equipment is in by embedding it into a more dense meduim that can be more easily transported to a location away from the equipment.  If, however, you are dumping the heat right back into the same environment that the equipment is in, then you are doing no better than using the air to cool the parts directly.  Mainframes that are fluid cooled have the fluid transported to the outside of the building where the heat is dumped into the air via convection (ponds) or large evaporative exchangers.  The main advantage to using fluid cooling in a PC/Workstation is to allow the heat to be drawn away from the components to where a more efficient heat removal setup can be utilized.  If you can maintain the environment at nominal temps and the system has no occlusions that would prevent the use of efficient heat sinks, then the additional costs/headaches of a fluid cooling system may not be worth it.  However, if you have a situation where the position of your CPU/GPU's is not conducive to using "large" sinks and/or fans, or you have a desire to try to reduce the noise created by multiple fans, then the additional cost/maintenance may be worth the effort.

Kendall

I had thought about ethylene glycol, but the how-to videos I've seen all talk about using distilled water.  My temp readings inside the box tend to hover around 40 c under load, so that seems pretty good.  I'm thinking that as the system is stable, and from what you're telling me, air cooling is the way to go.  Now, what effect would having a faster processor have?  I mean, if the rendering is going through the GPU's, does upgrading form a 4790 give me any significantly faster rendering times?  Also, when I chose the motherboard, I basically was looking for the cheapest one that would hold the processor and cards.  Is there any advantage to a costlier one?  I currently have a Gigabyte Z97.

 

I know those aren't cooling questions, but you seem pretty well versed.

If you're using Iray exclusively and you have lots of CUDA cores and more VRAM than you will need, then a "higher performance" CPU will be of limited utility.  HOWEVER, if you are going to run into a situation where a scene will exceed the VRAM on your GPUs, then you'll be glad to have a more powerful CPU when the render falls back to it.  There are still plenty of computations that occur on the CPU, but if you are already happy with the responsiveness of your system then the Return-On-Investment for a newer CPU may be low to zero.

It sounds like you've already provided a good cooling situation.  If your PSU is good, then I'd say spend your money on other things.

Kendall

Well, I have a 1200 w PSU.  I struggle to keep my scenes small enough to go through the cards and not the CPU.  If thy're too big, I cut all non-essential objects.  My next investment will be a GTX 1080 ti when it comes out.  Then another one.  It's projected to have 12 gig of ram, and that should be big enough for any scene.  My biggest card now only has 4 gig, but since the smaller one has 3 gig, that's my scene limit.  Typically, that's no more than 4 characters plus scenery.

areg5

namffuak said:

areg5 said:

Kendall Sears said:

areg5 said:

samurle said:

If you're worried about cooling, and have two monster graphics cards like the GTX 1080, I probably wouldn't place them next to each other.  One card would just be sucking in the hot air of the other.  If your motherboard has 3 or more x16 slots, then you could space them apart, giving them some air.

That's a really good idea.  I notice that  one of my cards runs hotter than the other, and they are next to each other. I don't know it that's becasue the 970 is newer and faster than the 780. Now, what is the deal when one slot is PCIexX16, the next is X8, the last X4?  are they differences that effect performance?

Kendall Sears said:

When dealing with GPUs, the PSU is the most important concern followed by cooling.  If the PSU is marginal or providing noisy power then the GPU performance will suffer.  Noisy power will increase heat as well as cause the circuitry to throttle back power usage to get a stable source.  It isn't necessarily about the size of the PSU, but how clean the output is.  Also a good PSU will create less ambient heat of its own and may actually pull excess heat from the system if its cooling setup is good.

Fluid cooling will not guarantee better cooling than forced air.  There are lots of variables.  The amount of time that the fluid spends in the exchanger versus the amount of airflow past the exchanger and other concerns.  Too much air velocity past the exchanger can actually ADD heat to the system (from friction and the air not having enough time to absorb enough heat) instead of removing it.  Too little airflow and not enough heat is removed.  The velocity of the air needs to change based on the amount of heat being generated as well as the ambient temperature outside the system.  Read the design specs CAREFULLY to determine if you can maintain the ambient room temperature necessary for the fluid cooler to actually function efficiently.  For every degree above nominal that the room's (or computer's enclosure -- ie under desk) temperature rises the amount of removed heat will decrease on a curve.  As the temperature rises in the environment the amount of heat retained in the fluid system increases non-linearly (the exact curve is determined by the design) causing the fluid to cumulatively gather heat to the point where the coolant starts to boil.  This creates voids in the system that can cause the pump to fail.

It is also entirely possible to draw heat INTO your system FROM the environment.  Once the ambient temp crosses a specific point (which is different for each design) the fluid will draw heat from the environment and introduce it into the computer's case.  This happens when the additive airflow from other fans in the case cause a draw approaching the output fans on the exchanger.  The temperature in the case becomes lower than the temperature in the external environment.  This leads the exchanger to gather heat rather than exhaust it.  The heat is then transferred into your components for the airflow in the case to try to remove.

The above is a very abbreviated explanation of the situation. There is a reason that datacenters containing racks and racks of huge servers use air cooling (and are extremely loud).  It is more efficient to maintain the environment at a temperature where direct air cooling can be used than to try to compensate for all of the variables introduced by fluid cooling.

Kendall

Makes sense. My rig isn't in an enclosure.  So if I plan on doing an animation, that looks like it would take 3 days, if water cooled the water would get hotter and hotter?  I would have thought a radiator would take care of that, in a custom water cooled rig.  Of course, the hybrid cards don't have a radiator, just what looks like a self contained resevoir and a fan.

As I wrote, it depends on the design.  A radiator is a type of exchanger.  Very few systems use straight H20, with a majority using a coolant like Ethylene Glycol or similar.  Whether the coolant gets 'hotter' will depend on a number of factors (variables).  Each additive system and/or heat source added to the cooling system adds more variables.  For instance, an exchanger external to the enclosure is more likely to perform better than one mounted inside the enclosure, but is not guaranteed.  For instance, if the heat from the exchanger somehow gets recycled into the airflow into the enclosure and/or back through the exchanger then residual heat will accumulate into the system.  An example:  Lets say that the radiator is in the same room as the system itself.  If the room is small or not well ventilated, then the heat from the exchanger will increase the temperature of the air in the room, thus causing the system to draw in that heated air.  Now the air pulled into the system is (nominal+x degrees).  This air is subsequently used to push through the exchanger and heated another y degrees which heats the room another x degrees.  Eventually, the heat in the room will exceed the point where the fluid can exude heat and the fluid starts to approach the boiling point.  This will happen first in the tubing carrying the coolant, followed by the exchanger.  Even if the coolant doesn't boil causing voids, the residual heat remains in the system causing the heat in the components to not be removed.  There are also issues of pressure rises and degraded performance of mechanical parts (fans for the exchanger slowing down due to age) that must be monitored.

This is VERY much like an automobile overheating on a hot day.  When the temperature outdoors gets into the 90's F, vehicles start overheating even though 90F is far below the boiling point of the coolant.  However, the heat produced by the engine accumulates to the point where the coolant can no longer transfer the heat to the air and the coolant boils, the pump loses prime, and things go horribly bad.  The same thing can happen in a computer using fluid cooling especially in a situation where there are multiple sources of heat.

Soooooo..... what this says is that if you intend on using fluid cooling, then you must take into consideration that the coolant adds an additional complexity to the planning.  In the end, you are still using forced air from the environment to remove the heat.  What fluid cooling is SUPPOSED to do is allow the removal of the heat from the environment that the equipment is in by embedding it into a more dense meduim that can be more easily transported to a location away from the equipment.  If, however, you are dumping the heat right back into the same environment that the equipment is in, then you are doing no better than using the air to cool the parts directly.  Mainframes that are fluid cooled have the fluid transported to the outside of the building where the heat is dumped into the air via convection (ponds) or large evaporative exchangers.  The main advantage to using fluid cooling in a PC/Workstation is to allow the heat to be drawn away from the components to where a more efficient heat removal setup can be utilized.  If you can maintain the environment at nominal temps and the system has no occlusions that would prevent the use of efficient heat sinks, then the additional costs/headaches of a fluid cooling system may not be worth it.  However, if you have a situation where the position of your CPU/GPU's is not conducive to using "large" sinks and/or fans, or you have a desire to try to reduce the noise created by multiple fans, then the additional cost/maintenance may be worth the effort.

Kendall

I had thought about ethylene glycol, but the how-to videos I've seen all talk about using distilled water.  My temp readings inside the box tend to hover around 40 c under load, so that seems pretty good.  I'm thinking that as the system is stable, and from what you're telling me, air cooling is the way to go.  Now, what effect would having a faster processor have?  I mean, if the rendering is going through the GPU's, does upgrading form a 4790 give me any significantly faster rendering times?  Also, when I chose the motherboard, I basically was looking for the cheapest one that would hold the processor and cards.  Is there any advantage to a costlier one?  I currently have a Gigabyte Z97.

 

I know those aren't cooling questions, but you seem pretty well versed.

I have an I7-5930; 6 core 3.5 GHz cpu. I also have a scene that takes 57 minutes to render with my 980ti. If I add the cpu the render time drops to 52 minutes and all 12 cores are 100% busy for that 52 minutes. You may draw your own conclusion. My system is completely air cooled and during that render the cpu temp peaks at 56 C and the 980ti at 59 C.

57 minutes?  I thought it would go a little faster with a 980 Ti.  The cooling is about the same as my 970.  The 780 runs a bit hotter.

Kendall Sears

areg5 said:

Kendall Sears said:

areg5 said:

Kendall Sears said:

areg5 said:

samurle said:

If you're worried about cooling, and have two monster graphics cards like the GTX 1080, I probably wouldn't place them next to each other.  One card would just be sucking in the hot air of the other.  If your motherboard has 3 or more x16 slots, then you could space them apart, giving them some air.

That's a really good idea.  I notice that  one of my cards runs hotter than the other, and they are next to each other. I don't know it that's becasue the 970 is newer and faster than the 780. Now, what is the deal when one slot is PCIexX16, the next is X8, the last X4?  are they differences that effect performance?

Kendall Sears said:

When dealing with GPUs, the PSU is the most important concern followed by cooling.  If the PSU is marginal or providing noisy power then the GPU performance will suffer.  Noisy power will increase heat as well as cause the circuitry to throttle back power usage to get a stable source.  It isn't necessarily about the size of the PSU, but how clean the output is.  Also a good PSU will create less ambient heat of its own and may actually pull excess heat from the system if its cooling setup is good.

Fluid cooling will not guarantee better cooling than forced air.  There are lots of variables.  The amount of time that the fluid spends in the exchanger versus the amount of airflow past the exchanger and other concerns.  Too much air velocity past the exchanger can actually ADD heat to the system (from friction and the air not having enough time to absorb enough heat) instead of removing it.  Too little airflow and not enough heat is removed.  The velocity of the air needs to change based on the amount of heat being generated as well as the ambient temperature outside the system.  Read the design specs CAREFULLY to determine if you can maintain the ambient room temperature necessary for the fluid cooler to actually function efficiently.  For every degree above nominal that the room's (or computer's enclosure -- ie under desk) temperature rises the amount of removed heat will decrease on a curve.  As the temperature rises in the environment the amount of heat retained in the fluid system increases non-linearly (the exact curve is determined by the design) causing the fluid to cumulatively gather heat to the point where the coolant starts to boil.  This creates voids in the system that can cause the pump to fail.

It is also entirely possible to draw heat INTO your system FROM the environment.  Once the ambient temp crosses a specific point (which is different for each design) the fluid will draw heat from the environment and introduce it into the computer's case.  This happens when the additive airflow from other fans in the case cause a draw approaching the output fans on the exchanger.  The temperature in the case becomes lower than the temperature in the external environment.  This leads the exchanger to gather heat rather than exhaust it.  The heat is then transferred into your components for the airflow in the case to try to remove.

The above is a very abbreviated explanation of the situation. There is a reason that datacenters containing racks and racks of huge servers use air cooling (and are extremely loud).  It is more efficient to maintain the environment at a temperature where direct air cooling can be used than to try to compensate for all of the variables introduced by fluid cooling.

Kendall

Makes sense. My rig isn't in an enclosure.  So if I plan on doing an animation, that looks like it would take 3 days, if water cooled the water would get hotter and hotter?  I would have thought a radiator would take care of that, in a custom water cooled rig.  Of course, the hybrid cards don't have a radiator, just what looks like a self contained resevoir and a fan.

As I wrote, it depends on the design.  A radiator is a type of exchanger.  Very few systems use straight H20, with a majority using a coolant like Ethylene Glycol or similar.  Whether the coolant gets 'hotter' will depend on a number of factors (variables).  Each additive system and/or heat source added to the cooling system adds more variables.  For instance, an exchanger external to the enclosure is more likely to perform better than one mounted inside the enclosure, but is not guaranteed.  For instance, if the heat from the exchanger somehow gets recycled into the airflow into the enclosure and/or back through the exchanger then residual heat will accumulate into the system.  An example:  Lets say that the radiator is in the same room as the system itself.  If the room is small or not well ventilated, then the heat from the exchanger will increase the temperature of the air in the room, thus causing the system to draw in that heated air.  Now the air pulled into the system is (nominal+x degrees).  This air is subsequently used to push through the exchanger and heated another y degrees which heats the room another x degrees.  Eventually, the heat in the room will exceed the point where the fluid can exude heat and the fluid starts to approach the boiling point.  This will happen first in the tubing carrying the coolant, followed by the exchanger.  Even if the coolant doesn't boil causing voids, the residual heat remains in the system causing the heat in the components to not be removed.  There are also issues of pressure rises and degraded performance of mechanical parts (fans for the exchanger slowing down due to age) that must be monitored.

This is VERY much like an automobile overheating on a hot day.  When the temperature outdoors gets into the 90's F, vehicles start overheating even though 90F is far below the boiling point of the coolant.  However, the heat produced by the engine accumulates to the point where the coolant can no longer transfer the heat to the air and the coolant boils, the pump loses prime, and things go horribly bad.  The same thing can happen in a computer using fluid cooling especially in a situation where there are multiple sources of heat.

Soooooo..... what this says is that if you intend on using fluid cooling, then you must take into consideration that the coolant adds an additional complexity to the planning.  In the end, you are still using forced air from the environment to remove the heat.  What fluid cooling is SUPPOSED to do is allow the removal of the heat from the environment that the equipment is in by embedding it into a more dense meduim that can be more easily transported to a location away from the equipment.  If, however, you are dumping the heat right back into the same environment that the equipment is in, then you are doing no better than using the air to cool the parts directly.  Mainframes that are fluid cooled have the fluid transported to the outside of the building where the heat is dumped into the air via convection (ponds) or large evaporative exchangers.  The main advantage to using fluid cooling in a PC/Workstation is to allow the heat to be drawn away from the components to where a more efficient heat removal setup can be utilized.  If you can maintain the environment at nominal temps and the system has no occlusions that would prevent the use of efficient heat sinks, then the additional costs/headaches of a fluid cooling system may not be worth it.  However, if you have a situation where the position of your CPU/GPU's is not conducive to using "large" sinks and/or fans, or you have a desire to try to reduce the noise created by multiple fans, then the additional cost/maintenance may be worth the effort.

Kendall

I had thought about ethylene glycol, but the how-to videos I've seen all talk about using distilled water.  My temp readings inside the box tend to hover around 40 c under load, so that seems pretty good.  I'm thinking that as the system is stable, and from what you're telling me, air cooling is the way to go.  Now, what effect would having a faster processor have?  I mean, if the rendering is going through the GPU's, does upgrading form a 4790 give me any significantly faster rendering times?  Also, when I chose the motherboard, I basically was looking for the cheapest one that would hold the processor and cards.  Is there any advantage to a costlier one?  I currently have a Gigabyte Z97.

 

I know those aren't cooling questions, but you seem pretty well versed.

If you're using Iray exclusively and you have lots of CUDA cores and more VRAM than you will need, then a "higher performance" CPU will be of limited utility.  HOWEVER, if you are going to run into a situation where a scene will exceed the VRAM on your GPUs, then you'll be glad to have a more powerful CPU when the render falls back to it.  There are still plenty of computations that occur on the CPU, but if you are already happy with the responsiveness of your system then the Return-On-Investment for a newer CPU may be low to zero.

It sounds like you've already provided a good cooling situation.  If your PSU is good, then I'd say spend your money on other things.

Kendall

Well, I have a 1200 w PSU.  I struggle to keep my scenes small enough to go through the cards and not the CPU.  If thy're too big, I cut all non-essential objects.  My next investment will be a GTX 1080 ti when it comes out.  Then another one.  It's projected to have 12 gig of ram, and that should be big enough for any scene.  My biggest card now only has 4 gig, but since the smaller one has 3 gig, that's my scene limit.  Typically, that's no more than 4 characters plus scenery.

It's not the "objects" that cause the problem, but the large textures.  The best way to stay "on card" is to lower the texture sizes on items that are not "in focus" or are not visible (skin textures covered by clothing for instance, or displacement on fabric too far from the camera to be seen).  Polygons are mostly not a problem.  There are long discussions on this subject in the commons.

Kendall

namffuak

areg5 said:

namffuak said:

areg5 said:

Kendall Sears said:

areg5 said:

samurle said:

If you're worried about cooling, and have two monster graphics cards like the GTX 1080, I probably wouldn't place them next to each other.  One card would just be sucking in the hot air of the other.  If your motherboard has 3 or more x16 slots, then you could space them apart, giving them some air.

That's a really good idea.  I notice that  one of my cards runs hotter than the other, and they are next to each other. I don't know it that's becasue the 970 is newer and faster than the 780. Now, what is the deal when one slot is PCIexX16, the next is X8, the last X4?  are they differences that effect performance?

Kendall Sears said:

When dealing with GPUs, the PSU is the most important concern followed by cooling.  If the PSU is marginal or providing noisy power then the GPU performance will suffer.  Noisy power will increase heat as well as cause the circuitry to throttle back power usage to get a stable source.  It isn't necessarily about the size of the PSU, but how clean the output is.  Also a good PSU will create less ambient heat of its own and may actually pull excess heat from the system if its cooling setup is good.

Fluid cooling will not guarantee better cooling than forced air.  There are lots of variables.  The amount of time that the fluid spends in the exchanger versus the amount of airflow past the exchanger and other concerns.  Too much air velocity past the exchanger can actually ADD heat to the system (from friction and the air not having enough time to absorb enough heat) instead of removing it.  Too little airflow and not enough heat is removed.  The velocity of the air needs to change based on the amount of heat being generated as well as the ambient temperature outside the system.  Read the design specs CAREFULLY to determine if you can maintain the ambient room temperature necessary for the fluid cooler to actually function efficiently.  For every degree above nominal that the room's (or computer's enclosure -- ie under desk) temperature rises the amount of removed heat will decrease on a curve.  As the temperature rises in the environment the amount of heat retained in the fluid system increases non-linearly (the exact curve is determined by the design) causing the fluid to cumulatively gather heat to the point where the coolant starts to boil.  This creates voids in the system that can cause the pump to fail.

It is also entirely possible to draw heat INTO your system FROM the environment.  Once the ambient temp crosses a specific point (which is different for each design) the fluid will draw heat from the environment and introduce it into the computer's case.  This happens when the additive airflow from other fans in the case cause a draw approaching the output fans on the exchanger.  The temperature in the case becomes lower than the temperature in the external environment.  This leads the exchanger to gather heat rather than exhaust it.  The heat is then transferred into your components for the airflow in the case to try to remove.

The above is a very abbreviated explanation of the situation. There is a reason that datacenters containing racks and racks of huge servers use air cooling (and are extremely loud).  It is more efficient to maintain the environment at a temperature where direct air cooling can be used than to try to compensate for all of the variables introduced by fluid cooling.

Kendall

Makes sense. My rig isn't in an enclosure.  So if I plan on doing an animation, that looks like it would take 3 days, if water cooled the water would get hotter and hotter?  I would have thought a radiator would take care of that, in a custom water cooled rig.  Of course, the hybrid cards don't have a radiator, just what looks like a self contained resevoir and a fan.

As I wrote, it depends on the design.  A radiator is a type of exchanger.  Very few systems use straight H20, with a majority using a coolant like Ethylene Glycol or similar.  Whether the coolant gets 'hotter' will depend on a number of factors (variables).  Each additive system and/or heat source added to the cooling system adds more variables.  For instance, an exchanger external to the enclosure is more likely to perform better than one mounted inside the enclosure, but is not guaranteed.  For instance, if the heat from the exchanger somehow gets recycled into the airflow into the enclosure and/or back through the exchanger then residual heat will accumulate into the system.  An example:  Lets say that the radiator is in the same room as the system itself.  If the room is small or not well ventilated, then the heat from the exchanger will increase the temperature of the air in the room, thus causing the system to draw in that heated air.  Now the air pulled into the system is (nominal+x degrees).  This air is subsequently used to push through the exchanger and heated another y degrees which heats the room another x degrees.  Eventually, the heat in the room will exceed the point where the fluid can exude heat and the fluid starts to approach the boiling point.  This will happen first in the tubing carrying the coolant, followed by the exchanger.  Even if the coolant doesn't boil causing voids, the residual heat remains in the system causing the heat in the components to not be removed.  There are also issues of pressure rises and degraded performance of mechanical parts (fans for the exchanger slowing down due to age) that must be monitored.

This is VERY much like an automobile overheating on a hot day.  When the temperature outdoors gets into the 90's F, vehicles start overheating even though 90F is far below the boiling point of the coolant.  However, the heat produced by the engine accumulates to the point where the coolant can no longer transfer the heat to the air and the coolant boils, the pump loses prime, and things go horribly bad.  The same thing can happen in a computer using fluid cooling especially in a situation where there are multiple sources of heat.

Soooooo..... what this says is that if you intend on using fluid cooling, then you must take into consideration that the coolant adds an additional complexity to the planning.  In the end, you are still using forced air from the environment to remove the heat.  What fluid cooling is SUPPOSED to do is allow the removal of the heat from the environment that the equipment is in by embedding it into a more dense meduim that can be more easily transported to a location away from the equipment.  If, however, you are dumping the heat right back into the same environment that the equipment is in, then you are doing no better than using the air to cool the parts directly.  Mainframes that are fluid cooled have the fluid transported to the outside of the building where the heat is dumped into the air via convection (ponds) or large evaporative exchangers.  The main advantage to using fluid cooling in a PC/Workstation is to allow the heat to be drawn away from the components to where a more efficient heat removal setup can be utilized.  If you can maintain the environment at nominal temps and the system has no occlusions that would prevent the use of efficient heat sinks, then the additional costs/headaches of a fluid cooling system may not be worth it.  However, if you have a situation where the position of your CPU/GPU's is not conducive to using "large" sinks and/or fans, or you have a desire to try to reduce the noise created by multiple fans, then the additional cost/maintenance may be worth the effort.

Kendall

I had thought about ethylene glycol, but the how-to videos I've seen all talk about using distilled water.  My temp readings inside the box tend to hover around 40 c under load, so that seems pretty good.  I'm thinking that as the system is stable, and from what you're telling me, air cooling is the way to go.  Now, what effect would having a faster processor have?  I mean, if the rendering is going through the GPU's, does upgrading form a 4790 give me any significantly faster rendering times?  Also, when I chose the motherboard, I basically was looking for the cheapest one that would hold the processor and cards.  Is there any advantage to a costlier one?  I currently have a Gigabyte Z97.

 

I know those aren't cooling questions, but you seem pretty well versed.

I have an I7-5930; 6 core 3.5 GHz cpu. I also have a scene that takes 57 minutes to render with my 980ti. If I add the cpu the render time drops to 52 minutes and all 12 cores are 100% busy for that 52 minutes. You may draw your own conclusion. My system is completely air cooled and during that render the cpu temp peaks at 56 C and the 980ti at 59 C.

57 minutes?  I thought it would go a little faster with a 980 Ti.  The cooling is about the same as my 970.  The 780 runs a bit hotter.

I've found ways to speed it up, but I keep this version handy as a maximum stress test. FWIW, with both the 1080 and the 980ti it runs in 26 minutes.

namffuak

Argh! Stoopid double post!

Mattymanx

I too have the CoolerMaster HafX with four fans and will vouch for its ablility to keep things cool.

For anyone looking for a really good CPU cooler, I will swear by the Noctua NH-D15 (dual fan)  - http://noctua.at/en/nh-d15.html - My i7 6700K will top out at 65-70C and the second it is done, the temp drops very very rapidly.

I can concur with samurle that having two cards close to each other will cause one to suck the heat from the other.  I have two Zotac AMP Extreme 980TIs that take up 3 slots each and they main card takes in the heat from the second card.  As a result the main card always hits 80C while the second card usally caps out at 65C while rendering or gamming.

namffuak

What Mattymanx said about Noctua - I'm running a Noctua NH U-12S with a single fan in an Antec 1200 V3 case (6 fans - 3 intake, three exhaust).

areg5

Mattymanx said:

I too have the CoolerMaster HafX with four fans and will vouch for its ablility to keep things cool.

For anyone looking for a really good CPU cooler, I will swear by the Noctua NH-D15 (dual fan)  - http://noctua.at/en/nh-d15.html - My i7 6700K will top out at 65-70C and the second it is done, the temp drops very very rapidly.

I can concur with samurle that having two cards close to each other will cause one to suck the heat from the other.  I have two Zotac AMP Extreme 980TIs that take up 3 slots each and they main card takes in the heat from the second card.  As a result the main card always hits 80C while the second card usally caps out at 65C while rendering or gamming.

Do you modulate your fans?  The only fan I can't control is the one on the side panel, because that one is powered directly from the PSU.  I have my GPU fans started really kicking on at 60 degrees,   So under load, with my room temp around 68, the 780 ti runs at 64 degrees, the 970 at 53, and the CPU at 54.

jura

areg5 said:

Well, I was having a heck of a time with my system.  I had to throttle down the power on my cards or they would crash and it went to CPU rendering.  See my specs below.  Anyway, it was all crammed into an old Altec Mid tower case with 2 fans.  At 70 % power to the cards, my temps were 76 degrees on the 780 and 70 degrees on the 970, cpu at 75.  These are below the max by enough, or so I thought.  Well I got a Coolermaster Haf X full tower with 4 fans, and a coolermaster CPU fan with heatsink.  I put all of the components from my Altec into it.  Changed nothing else except the CPU fan is new.  I was using the stock one.  The result?  It runs flawlessly at 100%, with no crashes.  Both GPU's are running 10 degrees cooler (64 and 59 degrees).  Huge difference on the CPU with a better fan.  Under load, 54 degrees.

I had no idea that cooling is that important, but in my case it made an enormous difference.  My rendering times are much faster.  Scenes that used to take 40 minutes now take 25.  And all by cooling it off.  Who knew?  Lots of you guys, no doubt, but not me.  I thought the case was just an afterthought so it would look cooler.  Nope.  Don't underestimate the importance of cooling.  Rendering generates a lot of heat.

Hi there

If you are experiencing with throttling,then I would probably suggest better case like is Phanteks Enthoo Pro/ATX or if you like smaller case then EVOLV ATX Tempered glass looks awesome,those cases do have great airflow and they should be lot better with multiple GPU and should be better with cooling yours current PC

Regarding cooling the CPU,I would go with Noctua NH-D14,Thermalright True Power 140,Cryorig R1 Universal or Ultimate and there are several other CPU cooler which are awesome and will do awesome job in rendering,I'm running i7-5820k 4.5GHz OC and is cooled by NH-D15 and my temps during the rendering are in 61-63C,previouslly have run Corsair H100i v2 which has been pain,my temps at 4.4GHz OC has been in middle 70's and in stress tests I've hit easily 85C,due this I wouldn't consider any kind of AIO/CLC,if yes then have look on EK Predator or Swiftech X series,those seems are best

HAF X is great case,have run that case for several months and years,but right now I'm running Enthoo Primo and that case offers me everythinbg what I like and what I love on that case,if budget is smaller then Thermaltake X71 or X5 or X9 are great cases which offers awesome features for little money,similar cases like is CaseLabs cost few times more than those cases 

Hope this helps

Thanks,Jura

 

 

areg5

I have a Coolermaster D92 on my 4790, and it doesn't go above 54 degrees with rendering.  With the stock fan it stayed at around 78.

jura

areg5 said:

I have a Coolermaster D92 on my 4790, and it doesn't go above 54 degrees with rendering.  With the stock fan it stayed at around 78.

Yup stock CPU cooler is really poor CPU cooler,I don't use with IRAY CPU becuase won't make any noticeable difference(we are speaking about 2-3 mins as max maybe bit more),yours CPU is OC there,my is running 4.5GHz,but right now I'm running 4.6GHz as outside temps are lower

Same I'm running OC on both GPU:Titan X is running 1450MHz(stock 1360MHz) and GTX1080 running 2055Mhz(stock 1883MHz)

Here is it how it looks

But in yours case,you made big progress on temps side there

Hope this helps

Thanks,Jura

alan bard newcomer

I have dual xeons with stock fans ..  and one 980ti with 6 gigs..  Idling the cores sit at 34 to 43 (my monitor software shows them all in a rainbow) and hit in the 60s when 3ding but I use a limiter that cuts Daz to 75% of the cpu and ram so I can do other stuff
---
was just going to do a fan count.. but there's only one in top on ....
---
but I have big thermal take server cases ... and I have them sitting crossways behind the center monitors...
and I don't have the side on them...
they add about ten degrees to the base temp of the room...  and I too keep planning to hook up a couple little fans actually blowing the heat out into the room.
--- the gpu is at 64 but fan is only running a 14% so I guess the card is happy.. using 1275 mb out of 6 running 3 monitors..
---
so the room acts as a big heat sink ...
--
but if you can open up the case that might help a lot with airflow...

---
 

areg5

jura said:

areg5 said:

I have a Coolermaster D92 on my 4790, and it doesn't go above 54 degrees with rendering.  With the stock fan it stayed at around 78.

Yup stock CPU cooler is really poor CPU cooler,I don't use with IRAY CPU becuase won't make any noticeable difference(we are speaking about 2-3 mins as max maybe bit more),yours CPU is OC there,my is running 4.5GHz,but right now I'm running 4.6GHz as outside temps are lower

Same I'm running OC on both GPU:Titan X is running 1450MHz(stock 1360MHz) and GTX1080 running 2055Mhz(stock 1883MHz)

Here is it how it looks

But in yours case,you made big progress on temps side there

Hope this helps

Thanks,Jura

 

jura said:

areg5 said:

I have a Coolermaster D92 on my 4790, and it doesn't go above 54 degrees with rendering.  With the stock fan it stayed at around 78.

Yup stock CPU cooler is really poor CPU cooler,I don't use with IRAY CPU becuase won't make any noticeable difference(we are speaking about 2-3 mins as max maybe bit more),yours CPU is OC there,my is running 4.5GHz,but right now I'm running 4.6GHz as outside temps are lower

Same I'm running OC on both GPU:Titan X is running 1450MHz(stock 1360MHz) and GTX1080 running 2055Mhz(stock 1883MHz)

Here is it how it looks

But in yours case,you made big progress on temps side there

Hope this helps

Thanks,Jura

Very nice system!  So, you don't use this for Iray?  Looks like it would do it really well!  I don't overclock anything.  I'm scared to do it, and I heard it doesn't add much to rendering times.

Mattymanx

areg5 said:

Mattymanx said:

I too have the CoolerMaster HafX with four fans and will vouch for its ablility to keep things cool.

For anyone looking for a really good CPU cooler, I will swear by the Noctua NH-D15 (dual fan)  - http://noctua.at/en/nh-d15.html - My i7 6700K will top out at 65-70C and the second it is done, the temp drops very very rapidly.

I can concur with samurle that having two cards close to each other will cause one to suck the heat from the other.  I have two Zotac AMP Extreme 980TIs that take up 3 slots each and they main card takes in the heat from the second card.  As a result the main card always hits 80C while the second card usally caps out at 65C while rendering or gamming.

Do you modulate your fans?  The only fan I can't control is the one on the side panel, because that one is powered directly from the PSU.  I have my GPU fans started really kicking on at 60 degrees,   So under load, with my room temp around 68, the 780 ti runs at 64 degrees, the 970 at 53, and the CPU at 54.

I have an ASUS Z170 Deluxe that came with its own software for regulating fan speeds and I think it controls all but one in the case.

areg5

Mattymanx said:

areg5 said:

Mattymanx said:

I too have the CoolerMaster HafX with four fans and will vouch for its ablility to keep things cool.

For anyone looking for a really good CPU cooler, I will swear by the Noctua NH-D15 (dual fan)  - http://noctua.at/en/nh-d15.html - My i7 6700K will top out at 65-70C and the second it is done, the temp drops very very rapidly.

I can concur with samurle that having two cards close to each other will cause one to suck the heat from the other.  I have two Zotac AMP Extreme 980TIs that take up 3 slots each and they main card takes in the heat from the second card.  As a result the main card always hits 80C while the second card usally caps out at 65C while rendering or gamming.

Do you modulate your fans?  The only fan I can't control is the one on the side panel, because that one is powered directly from the PSU.  I have my GPU fans started really kicking on at 60 degrees,   So under load, with my room temp around 68, the 780 ti runs at 64 degrees, the 970 at 53, and the CPU at 54.

I have an ASUS Z170 Deluxe that came with its own software for regulating fan speeds and I think it controls all but one in the case.

How about the GPU fans?

 

jura

areg5 said:

jura said:

areg5 said:

I have a Coolermaster D92 on my 4790, and it doesn't go above 54 degrees with rendering.  With the stock fan it stayed at around 78.

Yup stock CPU cooler is really poor CPU cooler,I don't use with IRAY CPU becuase won't make any noticeable difference(we are speaking about 2-3 mins as max maybe bit more),yours CPU is OC there,my is running 4.5GHz,but right now I'm running 4.6GHz as outside temps are lower

Same I'm running OC on both GPU:Titan X is running 1450MHz(stock 1360MHz) and GTX1080 running 2055Mhz(stock 1883MHz)

Here is it how it looks

But in yours case,you made big progress on temps side there

Hope this helps

Thanks,Jura

 

jura said:

areg5 said:

I have a Coolermaster D92 on my 4790, and it doesn't go above 54 degrees with rendering.  With the stock fan it stayed at around 78.

Yup stock CPU cooler is really poor CPU cooler,I don't use with IRAY CPU becuase won't make any noticeable difference(we are speaking about 2-3 mins as max maybe bit more),yours CPU is OC there,my is running 4.5GHz,but right now I'm running 4.6GHz as outside temps are lower

Same I'm running OC on both GPU:Titan X is running 1450MHz(stock 1360MHz) and GTX1080 running 2055Mhz(stock 1883MHz)

Here is it how it looks

But in yours case,you made big progress on temps side there

Hope this helps

Thanks,Jura

Very nice system!  So, you don't use this for Iray?  Looks like it would do it really well!  I don't overclock anything.  I'm scared to do it, and I heard it doesn't add much to rendering times.

Hi there

Yes I do use this for IRAY and many other renderers,but mostly I render with Poser Pro 11 with SuperFly or in Blender Cycles 

Regarding the OC,what I've tested in Octane and in LuxRender or Reality there I've got better and faster results,in Poser this give me slightly better and faster times 

Hope this helps

Thanks,Jura

Gazukull

Water cooling if you are serious about keeping GPU's cool.

It's not that difficult to do and the cooling is bar none :P

Mattymanx

areg5 said:

Mattymanx said:

areg5 said:

Mattymanx said:

I too have the CoolerMaster HafX with four fans and will vouch for its ablility to keep things cool.

For anyone looking for a really good CPU cooler, I will swear by the Noctua NH-D15 (dual fan)  - http://noctua.at/en/nh-d15.html - My i7 6700K will top out at 65-70C and the second it is done, the temp drops very very rapidly.

I can concur with samurle that having two cards close to each other will cause one to suck the heat from the other.  I have two Zotac AMP Extreme 980TIs that take up 3 slots each and they main card takes in the heat from the second card.  As a result the main card always hits 80C while the second card usally caps out at 65C while rendering or gamming.

Do you modulate your fans?  The only fan I can't control is the one on the side panel, because that one is powered directly from the PSU.  I have my GPU fans started really kicking on at 60 degrees,   So under load, with my room temp around 68, the 780 ti runs at 64 degrees, the 970 at 53, and the CPU at 54.

I have an ASUS Z170 Deluxe that came with its own software for regulating fan speeds and I think it controls all but one in the case.

How about the GPU fans?

 

ITs all temperature controled.  So right now while that computer is idle and not gamming or rendering, all the fans are quiet.  The GPU fans will not run until 50C+ and will only speed up as the GPU gets hotter.  Same with my CPU fans and the rest of the case fans (the ones controled by the software anyways)  Like I said I dont know which one is not controled by the software but my case has 1 x 230mm, 3 x 200mm and a 140mm fan and they run quietly all the time.  My GPU fans drown out the case fans and even then its still quiet compared to older setups I have had.

areg5

Mattymanx said:

areg5 said:

Mattymanx said:

areg5 said:

Mattymanx said:

I too have the CoolerMaster HafX with four fans and will vouch for its ablility to keep things cool.

For anyone looking for a really good CPU cooler, I will swear by the Noctua NH-D15 (dual fan)  - http://noctua.at/en/nh-d15.html - My i7 6700K will top out at 65-70C and the second it is done, the temp drops very very rapidly.

I can concur with samurle that having two cards close to each other will cause one to suck the heat from the other.  I have two Zotac AMP Extreme 980TIs that take up 3 slots each and they main card takes in the heat from the second card.  As a result the main card always hits 80C while the second card usally caps out at 65C while rendering or gamming.

Do you modulate your fans?  The only fan I can't control is the one on the side panel, because that one is powered directly from the PSU.  I have my GPU fans started really kicking on at 60 degrees,   So under load, with my room temp around 68, the 780 ti runs at 64 degrees, the 970 at 53, and the CPU at 54.

I have an ASUS Z170 Deluxe that came with its own software for regulating fan speeds and I think it controls all but one in the case.

How about the GPU fans?

 

ITs all temperature controled.  So right now while that computer is idle and not gamming or rendering, all the fans are quiet.  The GPU fans will not run until 50C+ and will only speed up as the GPU gets hotter.  Same with my CPU fans and the rest of the case fans (the ones controled by the software anyways)  Like I said I dont know which one is not controled by the software but my case has 1 x 230mm, 3 x 200mm and a 140mm fan and they run quietly all the time.  My GPU fans drown out the case fans and even then its still quiet compared to older setups I have had.

I adjusted the fan curves on everything.  Between 50-60 degrees on the GPU's, the fans go up to near full.  The case fans go up at lower temps.

Mattymanx

areg5 said:

Mattymanx said:

areg5 said:

Mattymanx said:

areg5 said:

Mattymanx said:

I too have the CoolerMaster HafX with four fans and will vouch for its ablility to keep things cool.

For anyone looking for a really good CPU cooler, I will swear by the Noctua NH-D15 (dual fan)  - http://noctua.at/en/nh-d15.html - My i7 6700K will top out at 65-70C and the second it is done, the temp drops very very rapidly.

I can concur with samurle that having two cards close to each other will cause one to suck the heat from the other.  I have two Zotac AMP Extreme 980TIs that take up 3 slots each and they main card takes in the heat from the second card.  As a result the main card always hits 80C while the second card usally caps out at 65C while rendering or gamming.

Do you modulate your fans?  The only fan I can't control is the one on the side panel, because that one is powered directly from the PSU.  I have my GPU fans started really kicking on at 60 degrees,   So under load, with my room temp around 68, the 780 ti runs at 64 degrees, the 970 at 53, and the CPU at 54.

I have an ASUS Z170 Deluxe that came with its own software for regulating fan speeds and I think it controls all but one in the case.

How about the GPU fans?

 

ITs all temperature controled.  So right now while that computer is idle and not gamming or rendering, all the fans are quiet.  The GPU fans will not run until 50C+ and will only speed up as the GPU gets hotter.  Same with my CPU fans and the rest of the case fans (the ones controled by the software anyways)  Like I said I dont know which one is not controled by the software but my case has 1 x 230mm, 3 x 200mm and a 140mm fan and they run quietly all the time.  My GPU fans drown out the case fans and even then its still quiet compared to older setups I have had.

I adjusted the fan curves on everything.  Between 50-60 degrees on the GPU's, the fans go up to near full.  The case fans go up at lower temps.

I think mine are all set to auto.  I dont recall, Im not the one who set it up.  As for the GPU, guess it depends on what you have.  My Zotac AMP Extreme 980TIs have 3 fans on them.  I have never seen it go over 83C (main card sucking in heat from the second) and the fans at the same time sit around 92% max speed

areg5

Mattymanx said:

areg5 said:

Mattymanx said:

areg5 said:

Mattymanx said:

areg5 said:

Mattymanx said:

I too have the CoolerMaster HafX with four fans and will vouch for its ablility to keep things cool.

For anyone looking for a really good CPU cooler, I will swear by the Noctua NH-D15 (dual fan)  - http://noctua.at/en/nh-d15.html - My i7 6700K will top out at 65-70C and the second it is done, the temp drops very very rapidly.

I can concur with samurle that having two cards close to each other will cause one to suck the heat from the other.  I have two Zotac AMP Extreme 980TIs that take up 3 slots each and they main card takes in the heat from the second card.  As a result the main card always hits 80C while the second card usally caps out at 65C while rendering or gamming.

Do you modulate your fans?  The only fan I can't control is the one on the side panel, because that one is powered directly from the PSU.  I have my GPU fans started really kicking on at 60 degrees,   So under load, with my room temp around 68, the 780 ti runs at 64 degrees, the 970 at 53, and the CPU at 54.

I have an ASUS Z170 Deluxe that came with its own software for regulating fan speeds and I think it controls all but one in the case.

How about the GPU fans?

 

ITs all temperature controled.  So right now while that computer is idle and not gamming or rendering, all the fans are quiet.  The GPU fans will not run until 50C+ and will only speed up as the GPU gets hotter.  Same with my CPU fans and the rest of the case fans (the ones controled by the software anyways)  Like I said I dont know which one is not controled by the software but my case has 1 x 230mm, 3 x 200mm and a 140mm fan and they run quietly all the time.  My GPU fans drown out the case fans and even then its still quiet compared to older setups I have had.

I adjusted the fan curves on everything.  Between 50-60 degrees on the GPU's, the fans go up to near full.  The case fans go up at lower temps.

I think mine are all set to auto.  I dont recall, Im not the one who set it up.  As for the GPU, guess it depends on what you have.  My Zotac AMP Extreme 980TIs have 3 fans on them.  I have never seen it go over 83C (main card sucking in heat from the second) and the fans at the same time sit around 92% max speed

Pretty easy to do.  I use the EVGA Precision x app, with I think works on any kind of card.  You can modulate the fans on your cards.  Mine runs at 63 degrees.