Actually, no, but this is a good demonstration of bad temperature readings.
I have/had an MSI K7T Turbo (KT133A) with a 1.2 TB and a Swifty MC462A. After many joyous/frustrating nights spent tinkering, with the setup, adding powerful fans, replacing those fans with quieter fans, etc, I finally had a cool, and more importantly quiet computer.
Along the way, I placed a thermistor (a very small custom thermistor used where I work) next to the core, and verified that the socket thermistor was actually fairly good - it stayed within 1C of the top side thermistor from 35C all the way to 65C during testing (the socket thermistor on the K7T Turbo is on a flex cable, and makes good contact with the CPU). So I have a fairly good comfort level with my temp readings, which I managed to get down to 39C idle and 44C load (Prime95 for 3 hours). Note this is at stock speed and slightly reduced Vcore (not my sig speeds
).
Last night, I received and installed a new Epox 8K3A+, and installed it with everything else the same except using an AX-7 (same fan) instead of the Swifty (traces too near the MB holes for my taste with the Swifty metal lugs). Previously, I measured the AX-7 as performing a couple degrees warmer than the Swifty. I was curious that the 8K3A+ uses a socket thermistor that sits in the socket, and does not seem to make contact with the CPU, and what that would do to temp readings.
I was somewhat amused to see my readings sit at 34C 'idle' (not really idle, was installing and rebooting a lot) with stock speed and voltage. I was even more surprised when I was still at only 37C at 1430MHz and +0.05 on the voltage.
I think now, for the first time ever, I understand the claims of '32C idle, 35C load, TB@1450 1.85V' that I see around the boards.
I know none of this is new to most - bad socket readings are a thing of much discussion here. But I wanted to post this just to emphasize two points:
1) Not all socket thermistor readings are bad - the K7T Turbo as a case in point.
2) Not all socket thermistor readings are good - the 8K3A+ as a case in point.
If you made it this far, thanks for listening
P.S. Even though I know they are not even remotely accurate, I feel better overclocking with the new temps.
I have/had an MSI K7T Turbo (KT133A) with a 1.2 TB and a Swifty MC462A. After many joyous/frustrating nights spent tinkering, with the setup, adding powerful fans, replacing those fans with quieter fans, etc, I finally had a cool, and more importantly quiet computer.
Along the way, I placed a thermistor (a very small custom thermistor used where I work) next to the core, and verified that the socket thermistor was actually fairly good - it stayed within 1C of the top side thermistor from 35C all the way to 65C during testing (the socket thermistor on the K7T Turbo is on a flex cable, and makes good contact with the CPU). So I have a fairly good comfort level with my temp readings, which I managed to get down to 39C idle and 44C load (Prime95 for 3 hours). Note this is at stock speed and slightly reduced Vcore (not my sig speeds
).Last night, I received and installed a new Epox 8K3A+, and installed it with everything else the same except using an AX-7 (same fan) instead of the Swifty (traces too near the MB holes for my taste with the Swifty metal lugs). Previously, I measured the AX-7 as performing a couple degrees warmer than the Swifty. I was curious that the 8K3A+ uses a socket thermistor that sits in the socket, and does not seem to make contact with the CPU, and what that would do to temp readings.
I was somewhat amused to see my readings sit at 34C 'idle' (not really idle, was installing and rebooting a lot) with stock speed and voltage. I was even more surprised when I was still at only 37C at 1430MHz and +0.05 on the voltage.
I think now, for the first time ever, I understand the claims of '32C idle, 35C load, TB@1450 1.85V' that I see around the boards.
I know none of this is new to most - bad socket readings are a thing of much discussion here. But I wanted to post this just to emphasize two points:
1) Not all socket thermistor readings are bad - the K7T Turbo as a case in point.
2) Not all socket thermistor readings are good - the 8K3A+ as a case in point.
If you made it this far, thanks for listening
P.S. Even though I know they are not even remotely accurate, I feel better overclocking with the new temps.
