Q6600 on the way - advice

setter said:
WingZero, by free do you mean simply input the amount of ram, (or slightly below it) indicated in the system resource monitor.

Rsm.jpg
Click to expand...

Input sligtly lower than 'Free' memory:).
Choosing standard, maximum, high or custom shouldn't make much difference. However custom with 'free' memory or slightly lower makes it the more accurate as you are always making use of 'free' physical memory and don't have to make use of cached memory/virtual memory.

Edit: you have to look for GFlops as stated many times. coretemps are dependant on GFlops values. Higher GFlops values will give higher coretemps :).
 
Last edited:
j.col said:
whats the average gflops with a q6600? i suppose it depends on the fsb speed.
mine seem to average between 37 and 40,
Click to expand...

It depends on the speed of the cpu. GFlops is the rate/speed at which the cpu executes instructions and is related to cpu speed.

[email protected] 2.4 x 16 = 38.4GFlops

[email protected] 3.3 x 16 = 52.8GFlops

[email protected] 3.4 x 16 = 54.4GFlops

Those are the maximum values for a given speed. In Intel Burn Test the goal is to achieve atleast 70-80% of the maximum value.

So for [email protected] I usually get 43-45GFlops which is around 80%:).

http://forums.overclockers.co.uk/showthread.php?t=18206940

http://www.overclock.net/intel-cpus/645392-how-run-linpack-stress-test-linx.html

http://www.intel.com/support/processors/sb/cs-023143.htm#3
 
chipps101 said:
oooops just noticed i have had downloads running the whole time tonight :( my bad :( but i don't suppose it would have affected my results much hopefully.
Click to expand...

Background processes, programs can affect the performance as each program makes use of cpu resource. That's why it is recommended to close all the non essential programs when running IBT.
 
Ran a custom 5 pass test, 3345mb of ram listed as free, input 3330 in IBT, 8 threads as for some reason auto doesnt max the cores on my i7.

Results.
ibtfeb.jpg


Temps about 2-3c higher than if i run with maximum memory setting.
 
setter said:
Ran a custom 5 pass test, 3345mb of ram listed as free, input 3330 in IBT, 8 threads as for some reason auto doesnt max the cores on my i7.

Results.
ibtfeb.jpg


Temps about 2-3c higher than if i run with maximum memory setting.
Click to expand...

8 threads will load cpu more than auto. I need explaination on this myself as to why it is lol:D.

i7@4Ghz 4 x 16 = 64GFlops (max value)

50GFlops/64GFlops x 100 = 78.125%

You passed the 5 runs IBT test:) and temps are wonderful. Now run it for 50 passes and if you are successful then you have done it :).
 
it's something to do with ibt only seeing actual physical cores and not hyperthreaded ones so on auto its only running the same amount of threads as you have physical cores.

it does funny things with amd chips with unlocked cores on auto settings also.
 
Heres a run i done last january, though at that time i was running 21 x 193 bclk on the cpu, (ram at 1547 mhz). Maximum ram selected in IBT, and the GFLOPs are a lot lower. 16 threads selected due to a bug in that revision of IBT when 8 were selected

i75jan-1.jpg


@chipps, hyperthreaded cores does seem plausible, btw mate, the spot cool has done wonders for my gpu cooling. Cheers again.:)
 
Last edited:
setter said:
Temps about 2-3c higher than if i run with maximum memory setting.
Click to expand...

As stated before:

IBT maximum stress level = Available memory

Available memory = Free + (Cached - Modified)

During testing Free and Modified memories get used up fully which I found from my experimentation. Windows frees up some more physical memory from the cached memory (strictly speaking it is the standby memory)(which is also using physical memory to store instructions).
However some data still remains in the cached memory so cached memory doesn't get used up to the point where the above equation can be balanced. So I think this is where windows provides the required virtual memory to IBT.

As virtual memory is slower than physical memory, this means that cpu has to wait for data to be sent to it. So in the mean time temps get lower. This is what I think Perspolis meant when he said that maximum stress level makes use of both physical and virtual memory.:)
 
Last edited:
it served me well when i was playing with sli also :) so its nice that it is serving its purpose again rather than sitting around my spare room :)

who is that in the wallpaper btw ? :)
 
I don't know how you got 68c prime at 1.5 volts on a q6600 setter it's amazing really what fans did you have on there what rpm and was it all lapped ?
I am hitting 69c prime under the archon at 1.46 volts 402x9

Anyway I have found that even a difference of 2-3 gflops lower can = lower temps by 4-5 c on IBT.

Btw you guys worry too much people have ran q6600's for 2 years on 1.5 volts and I have let mine run on IBT for extended periods hitting 85-95c and it's fine these things are virtually unbreakable.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom