I recently bought a Xeon E3-1230, running on a Supermicro X9SCM-F motherboard, as seems to be a pretty common setup around here.
As my usual practice with a new system, I proceeded to run some stress tests under 100% load to make sure the system does not have a fault that will appear later.
I loaded a recent version of Linux (Linux Mint Debian Edition) and did "modprobe w83627ehf" and "modprobe coretemp", and then ran "sensors" to get the CPU temperature information.
I then ran prime95, 8 threads, to fully load the CPU. With this running, the Core 0-4 temperatures range from 75-80C, and the "PECI Agent 0" temperature goes up to a max of about 86.0C. I'm not exactly sure why the PECI Agent 0 temperature is a few degrees higher than the Core 0-4 temperatures, but they do seem to change together.
I'm using the stock Intel CPU heatsink/fan, and I have it set to Full Speed in the BIOS, which seems to be working as "sensors" also shows it running around 2000 rpm.
I double-checked that the heatsink/fan was locked in place, and idle temperatures range from 24-35C so that definitely seems reasonable.
When fully loaded and the PECI Agent 0 temperature shows in the low 80s, the Supermicro IPMI interface shows "Medium" for temperature (not sure why they don't show the actual temperature here). But when it gets up to the max of about 86.0C then it goes to high, so even if the Linux sensors are reading slightly off, Supermicro's BIOS does confirm that it is running too hot.
My case has plenty of airflow, 3 other fans turned up all the way.
Has anyone else done a similar test and seen the same/different results? I thought that the stock Intel heatsink would be good enough to handle a full load as long as I wasn't overclocking, but it doesn't seem to be able to. Or is something not working right on my setup?
The Intel datasheet for the E3-1230 shows a TCASE of 69.1C, so that's why I'm concerned about these temperatures?
Should I buy an aftermarket cooler?
As my usual practice with a new system, I proceeded to run some stress tests under 100% load to make sure the system does not have a fault that will appear later.
I loaded a recent version of Linux (Linux Mint Debian Edition) and did "modprobe w83627ehf" and "modprobe coretemp", and then ran "sensors" to get the CPU temperature information.
I then ran prime95, 8 threads, to fully load the CPU. With this running, the Core 0-4 temperatures range from 75-80C, and the "PECI Agent 0" temperature goes up to a max of about 86.0C. I'm not exactly sure why the PECI Agent 0 temperature is a few degrees higher than the Core 0-4 temperatures, but they do seem to change together.
I'm using the stock Intel CPU heatsink/fan, and I have it set to Full Speed in the BIOS, which seems to be working as "sensors" also shows it running around 2000 rpm.
I double-checked that the heatsink/fan was locked in place, and idle temperatures range from 24-35C so that definitely seems reasonable.
When fully loaded and the PECI Agent 0 temperature shows in the low 80s, the Supermicro IPMI interface shows "Medium" for temperature (not sure why they don't show the actual temperature here). But when it gets up to the max of about 86.0C then it goes to high, so even if the Linux sensors are reading slightly off, Supermicro's BIOS does confirm that it is running too hot.
My case has plenty of airflow, 3 other fans turned up all the way.
Has anyone else done a similar test and seen the same/different results? I thought that the stock Intel heatsink would be good enough to handle a full load as long as I wasn't overclocking, but it doesn't seem to be able to. Or is something not working right on my setup?
The Intel datasheet for the E3-1230 shows a TCASE of 69.1C, so that's why I'm concerned about these temperatures?
Should I buy an aftermarket cooler?