Is this Core Temp Reading a problem

[hennethannun]

So i was planning on Oc'ing my new 89W 4400+ this weekend, and was running some Prime95 stability tests at stock just to get some baseline information, and I was a bit alarmed to discover that core temp and everest (which both read the CPU diode) are reporting one core 1 at 58-60C, while core 2 is down around 50C. Probe II and speedfan report the cpu temp as 50C overall.

is this normal with dual core processors, or does it mean that my Core 1 is just a bad overclocker, and I will have trouble getting much more performance out of my system (I was hoping for at least 2.4-2.6ghz)

ambient temps in the mid to low 20s, amd stock 4 heatpipe cooler...

 

[phile]

Prime95 will only stress one core, unless you run two instances, with the affinities set for each core. Don't even bother. Download Orthos Stress Prime, a vriant of Prime95 with native dual core support.

http://sp2004.fre3.com/beta/beta2.htm

As for your temps, they are a tad high for stock, but if those are your temps at full load, nothing to warrant panic. Stress with Orthos and report back your temps with both cores under full load.

Can you describe how you are cooling your CPU? Stock HSF? Good airflow in your case? For o/c'ing you're definitely going to want to get a good aftermarket HSF and use Arctic Silver 5 for thermal paste. You also want to make sure the HSF is making the best possible contact with your CPU heatspreader.

-phil

 

[hennethannun]

already running two instances of prime95 (with affinities set properly), that's why i was so suprised to see an 8C difference between the two cores.

all the temps i liste d those temps were all load temps.

at stock i am idling in the mid 30s (with C'n'Q). I am using AMD's four heatpipe stock heatsink (ie the heatsink that comes with the FX processors). all the reviews and benchmarks i have seen for this processor suggest that it is a very good mid-range heatsink.

case airflow is good (4 intake, 3 exhaust 80mm fans). I was hoping that I could get 2.6 or so with the AMD heatsink (since FX-60s run at stock with the same cooler).

So i tried overclocking a bit, just to see what sort of temps i get. Everything looked good at 2.4 and 2.5 with short prime95 stability testing, so I went up to 2.6

At 2.6ghz I am getting Core 1 61-63C / Core 2 55-57C with Core temp. Probe II and Speedfan both give me a CPU temp of 54-55C overclocked. that seems pretty good to me, so I am prime95 testing (two instances again) for a fwe hours, and will see what happens...

 

[Mucker]

The new version of Speedfan, 4.29, reads CPU #2 the same as Core Temp #2. This is described at the Speedfan website. The Speedfan reading is taken from the DTS just like Core Temp. Alfredo (the developer) is working on a version to read both core temperatures. I believe he will e-mail the beta to you if requested.

At 2620 MHz @ 1.425v my cores run 58C-#1 and 60C-#2 max temp while running Orthos with Big Typhoon....using Core Temp...

 

[hennethannun]

thanks mucker,

I guess my concern was the differential in temps between the 2 cores (rather than the differential between core temp and speedfan/probe II etc, which i expected). I have never oc'd a dual core system before, and I was just wondering if it is typical for the two cores to perform so differently (in terms of temps) at the same speed. Does this mean that my core 1 is not as good an overclocker as my core 2? or does it mean that with each core running one instance of prime95, the remaining system functions are being performed by core 1 (hence more core activity and higher temps)? or is it an anomalous reading?

 

[phile]

Originally posted by: hennethannun
thanks mucker,

I guess my concern was the differential in temps between the 2 cores (rather than the differential between core temp and speedfan/probe II etc, which i expected). I have never oc'd a dual core system before, and I was just wondering if it is typical for the two cores to perform so differently (in terms of temps) at the same speed. Does this mean that my core 1 is not as good an overclocker as my core 2? or does it mean that with each core running one instance of prime95, the remaining system functions are being performed by core 1 (hence more core activity and higher temps)? or is it an anomalous reading?

Just for fun, run Orthos Stress Prime and see if the cores are more evenly stressed. I don't think it's at all normal for there to be such a discrepancy in loading, between the cores, with a multi-thread app. I certainly never saw that when using either dual instances of prime, or orthos. Very strange, indeed.

-phil

 

[hennethannun]

thanks guys

i will try some other stress testing programs, and see if I get similar results...

 

[BlueWeasel]

I've tested 2 X2 3800+ and 2 X2 4200+ chips on my ASrock 939Dual board, and Core Temp consistently reads one core as being 7-8C hotter than the other. The hotter core has varied between the individual chips.