Firstly apologies for the following wall of text ...
I've recently upgraded my PSU from a Thermaltake 500 TWV to a Corsair TX650 and my GPU from an Asus 8600GTS (Passive Cooler) to 2 x HIS IceQ 4 4850s.
A few reasons for going HIS ...
1. 71 pages of problems with 4800 series cards by Sapphire, Asus and GigaByte. A few replaced thier Saps, Asus and Gigas with HIS and all reported thier problems had dissapeared. More have battled on flashing, modding files, etc.
2. Supposedly they run cooler. Reference idles around 70 and reaches around 90 under load.
3. Any hot air generated is dumped outside the case and not into it.
Anyway I wanted to see whether or not these cards lived up to thier name. So I downloaded GPU-Z and logged the temps of both cards to text file for graphing later. Well simply they are cooler ... by a country mile. Max on GPU 1 67 and on GPU 2 66 under S.T.A.L.K.E.R. Clear Sky.
Just out of interest I also checked my Northbridge and CPU temps as seeminlg the CPU is worked harder whilst running 2 GPUs. I noticed my Northbridge was at 40 degrees and CPU was at a respectable 36 degrees in EasyTune 5. This is the highest I've seen my Northbridge run so I wanted to know what it (and the CPU) was doing whilst I was gaming as I know temps can drop rapidly between exiting games and switching to the temp monitoring package.
So after a bit of hunting I found SpeedFan which also logs to a text file. On opening SpeedFan I got the correct Northbridge and CPU temperature (as confirmed when running EasyTune 5 in tandem). However SpeedFan also shows Core 1 & 2 temps (processor is dual core E6700) and these were not what I'd thought they would be. I thought the CPU temperature reading would be (Core 1 temp + Core 2 temp) / 2. But nooooo it's got to be different hasn't it !!!! and higher !
Anyway as I haven't seen any problems I thought what is the harm in actually logging the Northbridge, CPU, Core 1, Core 2 and CPU fan speeds through several gaming and PC/3D Mark tests.
I've logged the MAX readings and the associated application that maxed the temp ... they are as follows
Northbridge (Fallout 3) 41 degrees.
Core 1 (PCMark05) 53 degrees.
Core 2 (PCMark05) 50 degrees.
CPU reading (PCMark05) 39 degrees.
Max fan speed (PC Mark) 1520 RPM.
Just a couple of questions ...
1. Is the CPU reading taken from a thermocouple (or other temp sensor) on the motherboard or is it an additional sensor (below IHS between the two cores?) that is separate from the core sensors?
2. Are these temps OK ? It's not giving me a problem but are they high, average, better than average ?
3. I hear different temp monitoring programs give different readings, I'm inclined to believe SpeedFan though because of the fact it displays exactly the same readings as EasyTune 5 for both the Northbridge temperature and the CPU temperature but what about the core temps ?
RealTemp doesn't log for very long so it didn't get all the data I needed but it did record the maximum temperature at 53 degrees. But then again the maximum temperature reading from the SpeedFan log file was 53 degrees as well.
Interestingly enough though when you look at the graphs they are very similar but the individual values aren't the same. Looks like one of them is averaging over 3 seconds and then storing the average and the other storing the actual reading that just so happens to fall on the 3 second storage time.
Anyways here's what I've seen so far (don't know which one to believe !!!)