1. That Zalman cooler wont fit the TR4/SP3 socket. not applicable.
2. saying you "dont think 240W on air is possible" is arbitrary without context. 240W on a tiny intel or AM4 die? sure, difficult. 240W on a huge TR/EPYC package spread across 4-8 dies? easy. it's really not that hard to keep EPYC CPUs cool, at least the CPU package since they have such a large surface area to dissipate.
1. That Zalman cooler wont fit the TR4/SP3 socket. not applicable.
Ah, I thought it fitted anything, the box comes full of hundreds of mounting brackets. I guess TR4 is unusual. Ok, here's a list of coolers which will fit TR4, half are twice as substantial as the Noctua. As in two heatsinks and two fans. https://graphicscardhub.com/socket-tr4-lga-2066-cpu-cooler/
Ian&Steve C. wrote:
saying you "dont think 240W on air is possible" is arbitrary without context. 240W on a tiny intel or AM4 die? sure, difficult. 240W on a huge TR/EPYC package spread across 4-8 dies? easy.
The cooling is nothing to do with the contact area between CPU and cooler (that's metal to metal and conducts a lot), but how big the cooler is and it's fans (conducting to air is way harder, hence the huge surface area they have to create with fins). My cooler is 6 inches by 6 inches by 6 inches, with two 6 inch fans. And that only just keeps a 120W CPU at 80C with 20C ambient. With 240W, that would be 140C, a little on the high side.
Ian&Steve C. wrote:
it's really not that hard to keep EPYC CPUs cool
Then your cooler must have much faster/noisier fans than mine. Unless it's twice as big as 6x6x6 inches.
Odd how AMD say a water cooler is necessary on even my 120W CPU.
I guess if you managed to strap a GPU heatsink and fan onto your CPU, that might cool it. That's the only heatsink I've ever seen that can shift that much heat, even then, GPUs tend to run at 180W.
As for your claim of 60C, impossible. Is your room sub-zero or something? I've never seen any CPU of any kind stay that cool with any cooler, even water cooling. 240W is two and a half 100W tungsten lightbulbs worth of heat. An absolutely colossal amount of heat to shift, simply not possible in anything that size.
If this page takes an hour to load, reduce posts per page to 20 in your settings, then the tinpot 486 Einstein uses can handle it.
I don't know what possibly could have made the AIO a "very difficult" install.
All SP3 socket installations use the 4 threaded holes provided by the socket.
Installation involves nothing more than threading 4 screws into the holes.
My first Zalman, the thread busted on one of the screws and I had to get another from the supplier. Turn it slightly too tight and it's gone. Must have been made of chocolate.
And as for backplates.... some cases that means taking the damn motherboard back out.
If this page takes an hour to load, reduce posts per page to 20 in your settings, then the tinpot 486 Einstein uses can handle it.
1. That Zalman cooler wont fit the TR4/SP3 socket. not applicable.
Ah, I thought it fitted anything, the box comes full of hundreds of mounting brackets. I guess TR4 is unusual. Ok, here's a list of coolers which will fit TR4, half are twice as substantial as the Noctua. As in two heatsinks and two fans. https://graphicscardhub.com/socket-tr4-lga-2066-cpu-cooler/
Ian&Steve C. wrote:
saying you "dont think 240W on air is possible" is arbitrary without context. 240W on a tiny intel or AM4 die? sure, difficult. 240W on a huge TR/EPYC package spread across 4-8 dies? easy.
The cooling is nothing to do with the contact area between CPU and cooler (that's metal to metal and conducts a lot), but how big the cooler is and it's fans (conducting to air is way harder, hence the huge surface area they have to create with fins). My cooler is 6 inches by 6 inches by 6 inches, with two 6 inch fans. And that only just keeps a 120W CPU at 80C with 20C ambient. With 240W, that would be 140C, a little on the high side.
Ian&Steve C. wrote:
it's really not that hard to keep EPYC CPUs cool
Then your cooler must have much faster/noisier fans than mine. Unless it's twice as big as 6x6x6 inches.
Odd how AMD say a water cooler is necessary on even my 120W CPU.
I guess if you managed to strap a GPU heatsink and fan onto your CPU, that might cool it. That's the only heatsink I've ever seen that can shift that much heat, even then, GPUs tend to run at 180W.
As for your claim of 60C, impossible. Is your room sub-zero or something? I've never seen any CPU of any kind stay that cool with any cooler, even water cooling. 240W is two and a half 100W tungsten lightbulbs worth of heat. An absolutely colossal amount of heat to shift, simply not possible in anything that size.
you have a lot to learn. the contact area absolutely affects how much heat can be dissipated because it influences how quickly heat can be transferred from the CPU to the cooler, it is not perfect metal to metal contact (TIM) and there are several millimeters of conducting material before heat makes it's way to the fins for dissipation. there simply isnt enough temperature gradient to drive the heat transfer faster, so more surface area = more heat transfer. I don't think you realize how big they are. the IHS is like 4x the size of a Ryzen CPU.
that system is at low ambient in my garage, about 10-12C at the moment, not nearly sub-zero. in the summer when ambient is ~30-35C, the CPU still stays below 80C, still at 240W, still with this rather modest cooler and good case airflow. I don't worry too much about the noise as it's in my garage, I wouldn't consider it loud even if it were in the house.
EPYCs are easier to cool than smaller consumer chips at the same power levels due to the larger IHS and distribution of heat over a larger area. basic thermodynamics.
Peter, you need to review basic thermodynamic laws and restrictions. Area of thermal transfer interface is integral to calculations of heat transfer. As Ian points out the speed of transfer is directly related.It is easier to remove more heat per unit of time with a larger surface area. To get an idea of the relative size of
the IHS of Epyc/Threadripper compared to a Ryzen IHS view this short YT video.
I am feeling exceptionally stupid. Is that the model of Epyc Air Cooler that you sold me? If not, how loud is it? I would be perfectly happy to buy that model if it will keep my 7742 QS cpu that cool. Honest :)
Tom M
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!
I don't know what possibly could have made the AIO a "very difficult" install.
All SP3 socket installations use the 4 threaded holes provided by the socket.
Installation involves nothing more than threading 4 screws into the holes.
Keith,
I ended up with very hard to torque hose curl's on my first install. I was under the weather (I think) and rushing at the time.
My 2nd attempt will include removing a Cd-Rom drive to see if I can re-arrange the hoses basically straight. And finish turning over the fans to see if I can reduce the bearing rattle noises till I get motivated enough to buy some more Noctura Fans.
I will also investigate trying to turn the pump-heat exchange 180 degrees like you proposed. As far as I can tell that combination looks like it correctly fits in only one direction.
Since "it" is puttering along and not losing a lot of ground on the u@h project with the cpu running power restricted I have time to try to do it "right" this time.
Tom M
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!
Since "it" is puttering along and not losing a lot of ground on the u@h project with the cpu running power restricted I have time to try to do it "right" this time.
Tom, I once had a teacher who said: "If you don't have the time to do something right the first time, then don't do it at all".
Hence, I have taken that statement to heart and literally always made the correct choices in whatever I have done. It may take me a while longer, but I do get the job done... correctly.
Tom, the SP3 socket hole spacing is different on each end so if the coldplate/pump does not have a removable attachment plate that simply allows swapping the fixing 180 degrees, then the pump can only mount in one orientation and where the hoses exit is what you are stuck with.
But every TR4/Epyc AIO solution I have come across has a simple method of attaching the mounting ring or plate that allows swapping end for end or even at 90 degrees for the undersized coldplate solutions with a square or round coldplate.
The full coverage, rectangular coldplates will only fit at 0 or 180° depending on which way you install the attachment plate.
1. That Zalman cooler wont
)
1. That Zalman cooler wont fit the TR4/SP3 socket. not applicable.
2. saying you "dont think 240W on air is possible" is arbitrary without context. 240W on a tiny intel or AM4 die? sure, difficult. 240W on a huge TR/EPYC package spread across 4-8 dies? easy. it's really not that hard to keep EPYC CPUs cool, at least the CPU package since they have such a large surface area to dissipate.
my 7V12, at 240W, with a totally standard Supermicro 4U Air cooler with a single 92mm fan, sits at 55-60C while crunching 125 threads of Universe.
_________________________________________________________________________
I don't know what possibly
)
I don't know what possibly could have made the AIO a "very difficult" install.
All SP3 socket installations use the 4 threaded holes provided by the socket.
Installation involves nothing more than threading 4 screws into the holes.
Ian&Steve C. wrote:1. That
)
Ah, I thought it fitted anything, the box comes full of hundreds of mounting brackets. I guess TR4 is unusual. Ok, here's a list of coolers which will fit TR4, half are twice as substantial as the Noctua. As in two heatsinks and two fans. https://graphicscardhub.com/socket-tr4-lga-2066-cpu-cooler/
The cooling is nothing to do with the contact area between CPU and cooler (that's metal to metal and conducts a lot), but how big the cooler is and it's fans (conducting to air is way harder, hence the huge surface area they have to create with fins). My cooler is 6 inches by 6 inches by 6 inches, with two 6 inch fans. And that only just keeps a 120W CPU at 80C with 20C ambient. With 240W, that would be 140C, a little on the high side.
Then your cooler must have much faster/noisier fans than mine. Unless it's twice as big as 6x6x6 inches.
Odd how AMD say a water cooler is necessary on even my 120W CPU.
I guess if you managed to strap a GPU heatsink and fan onto your CPU, that might cool it. That's the only heatsink I've ever seen that can shift that much heat, even then, GPUs tend to run at 180W.
As for your claim of 60C, impossible. Is your room sub-zero or something? I've never seen any CPU of any kind stay that cool with any cooler, even water cooling. 240W is two and a half 100W tungsten lightbulbs worth of heat. An absolutely colossal amount of heat to shift, simply not possible in anything that size.
If this page takes an hour to load, reduce posts per page to 20 in your settings, then the tinpot 486 Einstein uses can handle it.
Keith Myers wrote: I don't
)
My first Zalman, the thread busted on one of the screws and I had to get another from the supplier. Turn it slightly too tight and it's gone. Must have been made of chocolate.
And as for backplates.... some cases that means taking the damn motherboard back out.
If this page takes an hour to load, reduce posts per page to 20 in your settings, then the tinpot 486 Einstein uses can handle it.
Peter Hucker
)
you have a lot to learn. the contact area absolutely affects how much heat can be dissipated because it influences how quickly heat can be transferred from the CPU to the cooler, it is not perfect metal to metal contact (TIM) and there are several millimeters of conducting material before heat makes it's way to the fins for dissipation. there simply isnt enough temperature gradient to drive the heat transfer faster, so more surface area = more heat transfer. I don't think you realize how big they are. the IHS is like 4x the size of a Ryzen CPU.
that system is at low ambient in my garage, about 10-12C at the moment, not nearly sub-zero. in the summer when ambient is ~30-35C, the CPU still stays below 80C, still at 240W, still with this rather modest cooler and good case airflow. I don't worry too much about the noise as it's in my garage, I wouldn't consider it loud even if it were in the house.
EPYCs are easier to cool than smaller consumer chips at the same power levels due to the larger IHS and distribution of heat over a larger area. basic thermodynamics.
opps, I was wrong: only 54C
_________________________________________________________________________
Peter, you need to review
)
Peter, you need to review basic thermodynamic laws and restrictions. Area of thermal transfer interface is integral to calculations of heat transfer. As Ian points out the speed of transfer is directly related.It is easier to remove more heat per unit of time with a larger surface area. To get an idea of the relative size of
the IHS of Epyc/Threadripper compared to a Ryzen IHS view this short YT video.
How Big is AMD's Threadripper CPU?
Ian&Steve C. wrote:my 7V12,
)
Ian&SteveC,
I am feeling exceptionally stupid. Is that the model of Epyc Air Cooler that you sold me? If not, how loud is it? I would be perfectly happy to buy that model if it will keep my 7742 QS cpu that cool. Honest :)
Tom M
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!
Keith Myers wrote: I don't
)
Keith,
I ended up with very hard to torque hose curl's on my first install. I was under the weather (I think) and rushing at the time.
My 2nd attempt will include removing a Cd-Rom drive to see if I can re-arrange the hoses basically straight. And finish turning over the fans to see if I can reduce the bearing rattle noises till I get motivated enough to buy some more Noctura Fans.
I will also investigate trying to turn the pump-heat exchange 180 degrees like you proposed. As far as I can tell that combination looks like it correctly fits in only one direction.
Since "it" is puttering along and not losing a lot of ground on the u@h project with the cpu running power restricted I have time to try to do it "right" this time.
Tom M
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!
Tom M wrote: Since "it" is
)
Tom, I once had a teacher who said: "If you don't have the time to do something right the first time, then don't do it at all".
Hence, I have taken that statement to heart and literally always made the correct choices in whatever I have done. It may take me a while longer, but I do get the job done... correctly.
Proud member of the Old Farts Association
Tom, the SP3 socket hole
)
Tom, the SP3 socket hole spacing is different on each end so if the coldplate/pump does not have a removable attachment plate that simply allows swapping the fixing 180 degrees, then the pump can only mount in one orientation and where the hoses exit is what you are stuck with.
But every TR4/Epyc AIO solution I have come across has a simple method of attaching the mounting ring or plate that allows swapping end for end or even at 90 degrees for the undersized coldplate solutions with a square or round coldplate.
The full coverage, rectangular coldplates will only fit at 0 or 180° depending on which way you install the attachment plate.
Which TR4 AIO are you using?