my two systems on the mining open chassis' are running EPYCD8 boards. they have 7402P CPUs @ 200W. I haven't really checked these. but I know the IPMI on these boards also lacks CPU VRM monitoring.
I used the ROMED8-2T board as a point of comparison because it was right in front of me, and you had the same board/CPU. my ROMED8-2T board with 7443P is in my O11D case. 3 fans pulling air into the bottom, through 360mm rad into the case, and 3 fans pushing air through the top 360mm radiator out of the case. both CPU (200W) and GPU (300W) in the cooling loop. the GPU also undoubtedly impedes airflow around the CPU socket. The case is closed, but pressure is neutral due to the rather large openings on the back panel and I/O panels of the case. this is why i mentioned the airflow as the likely reason your heatsink surface temp feels cooler.
but even though the temps are "fine" since we are not experiencing any adverse effects (like shutdowns or throttling), I still assert that these Asrock boards don't have proper CPU VRM monitoring sensors. I'm convinced the "MB" sensor is just on the board somewhere, and not on the VRM itself. I just can't believe that the weak VRMs and tiny heatsink are driving a 200W CPU at 35C without watercooling or massive airflow from server fans.
But the FLIR images of the ROMED8-2T mobo VRM heatsink under load is in the same ballpark as what I measured with my IR temp gun under load and all the sensors outputs are in the same ballpark and all change in response to the environment.
Whether any are reporting real VRM temps is debatable I guess.
But the FLIR images of the ROMED8-2T mobo VRM heatsink under load is in the same ballpark as what I measured with my IR temp gun under load and all the sensors outputs are in the same ballpark and all change in response to the environment.
and that's why i bought up the specifics on how those guns work as a potential source of error/innacuacy of the reading. with the temp gradient of the heatsink itself, the unknown IR reflectivity of the heatsink surface itself, the temperature averaging of the FOV from the gun, you can see how there could be pretty significant error stacking.
and what we care about is the VRM component temp. measuring the temperature of the heatsink is pretty irrelevant no matter what method you are using. the only way to get an accurate read is a temp sensor on the component itself. but it seems Asrock has not setup their board that way, or not exposed it to software.
It's already cranked Mikey! Epyc cpus aren't overclockable. I have mine running at its max 200W TDP and it runs 24/7 all year round.
Along with the 3080 with its 240mm radiator and a 2080 Ti it puts out plenty of heat in the wintertime
And summertime too unfortunately. Plenty of open windows and fans running all the time and A/C too.
My main computer is in the lounge, but my 6 Boinc only computers are in the garage, along with 2 Amazon parrots. I adjust the temperature to between 20 and 25C (parrot favourite) by opening a window at each end by a different amount depending on the weather. In Scotland a temp of over 25C outside is extremely rare.
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.
It's already cranked Mikey! Epyc cpus aren't overclockable. I have mine running at its max 200W TDP and it runs 24/7 all year round.
Along with the 3080 with its 240mm radiator and a 2080 Ti it puts out plenty of heat in the wintertime
And summertime too unfortunately. Plenty of open windows and fans running all the time and A/C too.
Yup it's the same here at the beach in NC, but my computer room is finally finished and as soon as I find all the hubs and network cables all the pc's will be up and running, only about 5 or 6 are right now. The problem with a room though, even though it's in the garage is no windows because of it's location so the only way to cool it off is to open the door and the side door to the garage as well. I put a storm door on it just for that so we will see how it works, I'm thinking I may still need a fan or two.
Fans work pretty well as long as you have good inflow and outflow. I cool really well when there is a north breeze that comes into the computer rooms where they are inches from the window. Then have fans to assist the incoming breeze to exit the rooms and traverse the rest of the house to exit the south exposure windows.
But reversing the breeze and have it come from the south does not work well at all. Too much impedance and restrictions for the south input breeze to make much headway to the north rooms. Even turning the exit fans around in direction does not work well. Still too many restrictions. South breeze = A/C being run.
my two systems on the mining
)
my two systems on the mining open chassis' are running EPYCD8 boards. they have 7402P CPUs @ 200W. I haven't really checked these. but I know the IPMI on these boards also lacks CPU VRM monitoring.
I used the ROMED8-2T board as a point of comparison because it was right in front of me, and you had the same board/CPU. my ROMED8-2T board with 7443P is in my O11D case. 3 fans pulling air into the bottom, through 360mm rad into the case, and 3 fans pushing air through the top 360mm radiator out of the case. both CPU (200W) and GPU (300W) in the cooling loop. the GPU also undoubtedly impedes airflow around the CPU socket. The case is closed, but pressure is neutral due to the rather large openings on the back panel and I/O panels of the case. this is why i mentioned the airflow as the likely reason your heatsink surface temp feels cooler.
https://i.imgur.com/PbUqxWO.jpg
but even though the temps are "fine" since we are not experiencing any adverse effects (like shutdowns or throttling), I still assert that these Asrock boards don't have proper CPU VRM monitoring sensors. I'm convinced the "MB" sensor is just on the board somewhere, and not on the VRM itself. I just can't believe that the weak VRMs and tiny heatsink are driving a 200W CPU at 35C without watercooling or massive airflow from server fans.
_________________________________________________________________________
But the FLIR images of the
)
But the FLIR images of the ROMED8-2T mobo VRM heatsink under load is in the same ballpark as what I measured with my IR temp gun under load and all the sensors outputs are in the same ballpark and all change in response to the environment.
Whether any are reporting real VRM temps is debatable I guess.
Keith Myers wrote: But the
)
and that's why i bought up the specifics on how those guns work as a potential source of error/innacuacy of the reading. with the temp gradient of the heatsink itself, the unknown IR reflectivity of the heatsink surface itself, the temperature averaging of the FOV from the gun, you can see how there could be pretty significant error stacking.
and what we care about is the VRM component temp. measuring the temperature of the heatsink is pretty irrelevant no matter what method you are using. the only way to get an accurate read is a temp sensor on the component itself. but it seems Asrock has not setup their board that way, or not exposed it to software.
_________________________________________________________________________
This is my VRM cooling
)
This is my VRM cooling solution for my Thermaltake Core X9 case that has the ROMED8-2T mobo and 7443P cpu.
The fan sitting on top of the drive cages is a Corsair ML140 at full 1800 rprm. Likely why I can report such low heatsink temps with my IR temp gun.
Keith Myers wrote: This is
)
If you change those tubes to an outside radiator during your winter months you could crank that thing up!!
It's aleady cranked Mikey!
)
It's already cranked Mikey!
)
It's already cranked Mikey! Epyc cpus aren't overclockable. I have mine running at its max 200W TDP and it runs 24/7 all year round.
Along with the 3080 with its 240mm radiator and a 2080 Ti it puts out plenty of heat in the wintertime
And summertime too unfortunately. Plenty of open windows and fans running all the time and A/C too.
Keith Myers wrote:It's
)
My main computer is in the lounge, but my 6 Boinc only computers are in the garage, along with 2 Amazon parrots. I adjust the temperature to between 20 and 25C (parrot favourite) by opening a window at each end by a different amount depending on the weather. In Scotland a temp of over 25C outside is extremely rare.
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: It's
)
Yup it's the same here at the beach in NC, but my computer room is finally finished and as soon as I find all the hubs and network cables all the pc's will be up and running, only about 5 or 6 are right now. The problem with a room though, even though it's in the garage is no windows because of it's location so the only way to cool it off is to open the door and the side door to the garage as well. I put a storm door on it just for that so we will see how it works, I'm thinking I may still need a fan or two.
Fans work pretty well as long
)
Fans work pretty well as long as you have good inflow and outflow. I cool really well when there is a north breeze that comes into the computer rooms where they are inches from the window. Then have fans to assist the incoming breeze to exit the rooms and traverse the rest of the house to exit the south exposure windows.
But reversing the breeze and have it come from the south does not work well at all. Too much impedance and restrictions for the south input breeze to make much headway to the north rooms. Even turning the exit fans around in direction does not work well. Still too many restrictions. South breeze = A/C being run.