LinX with more than 6Gb of RAM

Associate
Joined
2 Oct 2009
Posts
829
Location
Belfast, UK
Just upgraded from 6GB to 12 (3x4) and I am trying to find out if I can maintain my overclock at 4.2 Ghz with 18GB (six sticks).

What type of problem sizes and memory amounts would people use for this type of testing, would 36000 (10200 mb) seem like a viable amount to test or should I engage more RAM. (Testing takes forever at 36000).

I plan to do Prime95 as-well but LinX is much faster for finding initial instabilities.
 
Hmmm hit another snag. With 12Gb of ram using 10000 memory in LinX my NB peaks at 52c. With 18Gb in running at the same settings the NB reads 85c+.

What the devil?
Does it sound like some bizzare temperature sensor problem or can anyone think of a reason why running with six DDR3 sticks would push the NB temp up so much?

Note that the idle reading is always the same ~50c
 
All my memory is GSkill.

3X4GB GSkill Ripjaws CL9
3x2GB GSkill Ripjaws (original kit) CL8


I should clarify when I say NB I mean the IOH according to the bios. Apparently it is connected to the CPU and GPUs.


More ram will strain your NB going to have to cool that me thinks

Try a spot fan on it

when testing use as much memory as you can.

I have a Zaward 80mm fan attached directly to the NB. The fan spins up to 2400rpm and the IOH reading still hits 90c :S.

The fan always helped temps before when I was only running 6Gb, dropped 10+c off the temps at low rpms.
 
It's really bizarre. 30c + bump in temps just by adding 3 more sticks of ram.
Still doing some research and testing into this, may have to buy a NB heatsink but then again if mine peaks at 52c under normal circumstances I am not sure how much an aftermarket sink will help.
 
Firstly set all of the voltages manually in case of any bios bugs with 6 sticks, secondly don't set LinX to use 10000MB when you only have 12GB installed it might be that it's having to page to virtual memory and so it's spending all of its time idle waiting on the hard drive.
 
Last edited:
I have the DDR3 ram set at 1.6volts each manually. What would be a good amount of memory to test with?

I currently have 18GB installed. (only using the 3x4Gb sticks (12gb) works fine with normal ioh temps by the way.

Note my QPI/VTT is at 1.32 and my NB/IOH is at 1.13volts (Needed these settings for stable 4.2Ghz)
 
It could be the QPI volts doing it, the CPU and X58 chipset are both connected via the QPI bus, the X58 is little more than a PCI-E hub so memory shouldn't really have any affect... but I know you usually need more VTT to overclock memory and use XMP profiles so if that's linked to the QPI voltage it could be the cause?
 
Last edited:
That's what I thought, but even dropping the QPI down to 1.28 makes little difference to the temps.

Also I can bump the QPI up to say 1.35 with 12gb installed and the IOH only hits ~54c
 
Also I can bump the QPI up to say 1.35 with 12gb installed and the IOH only hits ~54c

Try using less memory than 10,000MB, 80C might be the normal temperature (bad cooling?) and 50c lower than normal due to paging.

What sort of heatsink does the chipset have on your board? are you sure it's the NB temp you're monitoring and not something else? VRM's?
 
Last edited:
Copper heatsink with no heat-pipes. It is not great hence why I added a fan unto it in the first place.

Biostar x58A Motherboard.

Edit: It is still hitting 85c running 6GB of RAM. What a mystery lol
 
Went straight to 83c at stock settings with 18Gb. A bit cooler, but not all that much.
With 12GB it went to 52c at stock settings. Odd
 
Back
Top Bottom