Computer Hope
Hardware => Hardware => Topic started by: Accessless on May 19, 2015, 11:11:06 AM
-
I've just rebuilt my old computer system & refitted it with a faster CPU:
M3A32-MVP DELUXE/WIFI-AP
AMD Phenom 9550 [Cooler] Stock --> Phenom II X4 960 [Cooler] Scythe Shuriken Rev.B
Radeon HD 5770
4Gb DDR2 RAM
500W PSU
But I have some thermal problems. Core temps register at least 8°C higher than CPU temps. And more worryingly an unknown temperature that wildly fluctuates between 40°C and 75°C. I assume that this reading is false as it can go from 40-75 and back again in a couple of seconds but I still don't know what it refers to. My worry is that it will cause instability by means of shutdowns because the system thinks that it is melting.
I have to emphasize though that I have no idea if the reading is false or not. It does seem to increase with work load.
Anyone have any ideas on how to identify it? Labeled temp3
-
What are you using to observe the temperatures?
-
SIW & Speedfan
-
The reading is false. Temps never vary from 40 to 75 in a short time.
-
But what is it? And more importantly will it cause "protective" shutdowns?
-
But what is it? And more importantly will it cause "protective" shutdowns?
Likely Erroneous, and probably not.
The only sensors that will cause any protective shutdown will be visible in your system BIOS, furthermore, the only one ever used for such a feature is the CPU temperature. the temperature sensors are effectively part of the chipset. Speedfan/SIW/HWMonitor try to view the sensor information by basically probing certain I/O Addresses based on what chipset it thinks it sees, But the actual chipset implementations differ between motherboards and sometimes the addresses simply don't correlate with a sensor.
Your fluctuating temperature could be a value unrelated to temperatures at all, or a failed thermosistor, but either way the temperatuer value will play no part in any emergency shutdown.
-
The reading is false. Temps never vary from 40 to 75 in a short time.
I totally disagree...i've observed AMD's go even more whacko than that in less time...
-
I totally disagree...i've observed AMD's go even more whacko than that in less time...
But that was when you used the AMD CPU to fry eggs.. Right?
-
This is the sensors I have listed [Idle]:
Temp1: 44°C
Temp2: 34°C
Temp3: 43-65°C
HD0: 28°C
CPU: 44°C
MB: 34°C
GPU: 33°C
Core: 50°C
The CPU and Core values also have me worried as their can be a discrepancy of up to 10°C.
I'm pretty unhappy with the whole situation. I'm contemplating buying an i3 with similar performance and less than half the TDP to deal with.
Is the north bridge used fairly consistently or do you get times of peak load? (or south bridge I get them mixed up).
-
I would look up the operating temp ranges for the CPU before doing a wholesale change...
-
I would look up the operating temp ranges for the CPU before doing a wholesale change...
http://products.amd.com/en-us/DesktopCPUDetail.aspx?id=617&f1=AMD+Phenom%E2%84%A2+II+X4&f2=965&f3=3400&f4=512&f5=AM3&f6=&f7=&f8=&f9=4000&f10=False&f11=True&f12=True
62°C AMD's usually recommend no more than ~65°C Which is laughable as without some serious water cooling these values under load can be impossible to achieve.
I have a Bequite! air cooler about the size of my head cooling an FX-8350 that just about achieves this (that's with a good case and plenty of fans).
On a side note, Temp3 does seem to immediately correlate with CPU use which does support the argument that it may not be a thermostat.
-
Ok after watching "temp3" jump +/-50C in 2 seconds as well as seeing it peek at 127C during max CPU load has confirmed to me that it is in fact not a real sensor. Also speed fan seems to be misrepresenting the core temperatures. AMD overdrive represents core temperature as four separate values (one for each core) as opposed to Speedfan's single value, so I assume that Speedfan is averaging the values incorrectly.
I was going to put a meter across the power as well due to the strange readings Speedfan was reporting e.g. -18V on -12V rail. But I think that its safe to ignore that as well. Prime95 tests came back good (once I worked up the nerve to use it :p)
Thanks all.
-
So your saying the temps are within range...or being mis-reported ? ?
I'm confused....
-
Speedfan's report of my core temperatures was wrong, too high (I'm more inclined to believe software made by AMD)
Temp3 was bogus, I think it represents either power usage or CPU load.
In any case the thing survived a Prime95 torture test with no errors and maintained acceptable temperatures.
-
Then yer good to go...
-
I use speedfan, I have never figured out what T3 is. I think it's not connected on my system, or not connected to a meaningful sensor on the Winbond chip. Not connected could account for the speed of the fluctuations being seen. The way its wired on my m/b is to read the Winbond sensor. By comparing with the readings in speccy and SiSoft Sandra Lite, I worked out that T2 was a measure of the CPU temp (my CPU doesn't have a temp sensor built into it), and T1 was a measure of the 'system' temperature, which means 'motherboard' or reckoned to be the T of the northbridge.
The apparent discrepancy in CPU temperature referred to could be the difference between the case and the core temperature. For core temperature, see what CoreTemp says. I can believe that the CPU case is a few C cooler than the core, maybe in certain circumstances even warmer than the core.
Dumb_Question
21.May.2015