Assembled this new setup over the week end:
Corsair Carbide 500R
Asus P8Z77 LE Plus
F3-14900CL9Q-16GBZL Ripjawsz 16GB 4X4GB DDR3-1866 CL9-10-9-28
Antec Kuhler H2O 920
Crucial M4 C400 256 GB
Seagate Baracuda 2TB
XFX Radeon HD 7970 GHz Edition
So far, I am quite happy with all the components.
Always on the safe side, I begin O/C with the safe setup of the MotherBoard BIOS. After a few reboots, ASUS gave me a FSB of 103 MHz and a mult. coeff of 41, which gives me around 4.2 GHZ under full load. System is stable at that speed (tested with Prime95 over several hours).
Then I began experimenting with manual O/C: Put the VCore on "Offset +", took FSB back to 100MHz but increased the mult. to 45 to achieve 4.5 GHZ which I read is quite common with this setup. Once again, system seems stable (ran 1 hour of Prime95 stress tests). But then I took some readings and got a few questions... During this stress test, ASUS Probe 2 gave a CPU temp of 45-50 degrees, Antec software gave me a liquid temperature of no more than 40 degrees (hence the fans never really increased in speed past 700 RPM), but RealTempGT gave me CPU cores temperatures of 95 degrees (Core Max temp seems to be 105 degrees on CPU specs). Seeing that, I decided to stop the tests.
Hence, my questions : how can one achieve 4.5 GHz with such Core temps so close from the TJ (Max temperatures) ? Is it normal that the liquid stays so cool when Core temps are so high, or is it a bad transmission from the CPU to the cooler ? And what could explain the reading being so different between ASUS Probe 2 and RealTempGT ?
Thanks for any insight you might provide, as those results surprised me quite a bit.
Liquid temperatures are not the same as the reported temps from CPU monitoring.
Liquid temps are derived as your Delta-t(emperature) as they relate to ambient air temps, coolant flow rate, radiator surface area and fin density, fan speed and the ability of the block to transfer heat from the CPU into the liquid. Also to consider are the specific heat and thermal conductivity of water and components that comprise the loop.
For example, for a 10C delta, your liquid temps are 10C above ambient room temperature due to the above variables. If your room temp is 80F (26.5C) then your coolant temp would be no more than 36.5C or 98F....lukewarm to body temperature. (this is just an example) Your processor might register 45-55C at load, but this doesn't relate 1:1 to water temp unless your delta is seriously high and you are dumping more heat into the loop than can be dissipated.
Core Temp is good for checking CPU temps. Try that program. Also, make sure all temps are in Celsius. I wouldn't let ur CPU go above 80C in OC'ing. You should be able to get some high clocks and keep it that low with ur cooler.
Thanks for your answers guys, but is that normal that even with a TCore of 95 celsius, the liquid temps stays as low as 40 celsius ? I know about temperature intertia, but the delta seems a bit big to me...
I also noticed that under full load, with this "Offset +" VCore setting, voltages reaches 1.4V for 4.5 GHz, could that explain the Tcore being so high ? Is it best to fix the Vcore manually or to use this "Offset +" setting ?
In Core Temp you should be maxing at 70-80 range on the high end, running prime95 for 5-10 min. 1.4v is also a bit on the high end but it's not crazy high, I wouldn't recommend to run it at that all the time.
It sounds like the heat is not getting off of the CPU as it should be which usually means the CPU block is not seated properly. Try redoing the thermal paste with AS5 and burn it in over a week and see how that fairs. With the 920 mount you should be able to screw the screws until they are all the way in if you installed the back plate correctly.
Also go into the over clocking options and make sure you blck is set to 100, the ASUS P877 series likes to ramp this up before turbo which only creates problems, in my experience.