You need to set the ram voltage to what it is supposed to run at. Its not going to harm it, and you wont be running it out of spec.
Printable View
plus if memory can handle it that will also be fasterQuote:
At equal clock speed the higher fsb will give more performance
Thanks for the advice guys, I'll give it a try and let you know how I get on later tonight :D
Well, I've got home (finally, that was one hell of a long day!) and upped the RAM voltage to 2.10 from 2.0 as per the RAM specs, run memtest to 100% coverage and found no errors so clocked the CPU to 355x9 (3.2ish). So far it's running fine but I've only been up for about 20 minutes, running Orthos still returns an error though, could anyone tell me what this means :
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
Torture Test ran 1 minutes 27 seconds - 1 errors, 0 warnings.
Execution halted.
I'm a bit perplexed! Other than that little glitch, running on CPU stress only I'm getting temps of up to 39 degrees under load which I'm quite impresses with so have to recommend the Asus silent square pro for anyone looking for a new heatsink :D
Anyway, any advice on what that error might mean in terms of problems would be welcomed, I'm off to search google until my eyelids close of their own accord...
If orthos was set to test using small ffts - cpu only then it's most likely you need more vcore (cpu voltage)
what is your cpu voltage set to in bios ??
Orthos was running a blended test when it gave me the errors, no errors while running small ffts - cpu only though, that runs fine which is why I was thinking RAM although memtest seems to have ruled that out... I'm just about to up the voltage one notch to the CPU, will post the numbers in a minute...
It's most likely a memory error.
when u say you ran memtest which 1 was it Memtest86+ or Memtest HCI??
Reason i ask is that both have their strengths & weaknesses:-
Memtest86+
Strength:-
tests nearly all memory as it runs in it's own enviroment
wide variety of tests
weakness
doesn't test memory as it is used in windows only test 5 & 8come close to emulating that. This often means that you can pass this memtest but still hve issues in windows.
memtest hci
advantage
runs in window so complements 86+ weakness. To be most effective run 2 instances.
disadvantage
only tests windows free memory - u need the full version to run like 86+, though in faiurness it isn't that dear.
Only by running both for longish periods do you get comfort that memory is ok.
my own preference i think would be to run 86+ 1st test 5. if this errors then tweak your voltage +/- a notch & then rerun stest. If that runs clear then run test 8 for an hour or so.
then i'd use HCI memtest in windows:-
testing as:-
From Manual:-
if those are good then go into xp & run hci overnight
it works in % coverage
100% coverage (a quick test to make sure your RAM is functioning reasonably)
Test for 1 hour (detects the majority of memory errors)
Test overnight (this will catch virtually all errors in your memory)
luck:)
Ok, have upped the CPU voltage to 1.35 and I'm still getting the same error, I'm on 12 hour shifts at the moment so I'll start memtest running in windows before I go out tomorrow and see what happens when I get back...
Well it is difficult to get too excited about OC with just an arctic pro.......IMHO it's the weakest link for me and starts struggling over 2.95ghz to get CPU temp below 65c during heavy CPU test load, on a cool day.
Mind you with three different temp measuring utilities I'm not sure what to trust, the bios gives very low temps compared to core temp...............I suppose if I get crashes on a hot day playing something like COD2 then I will know what one is accurate.
You can't possibly compare temp in bios with coretemp. that's like comparing an idle temp with a load temp & asking why is it dfifferent.
when looking in bios there is virtually no load on cpu, cortetemp in windows on the other hand hmay have a load that varies between 0~100% in a nmatter of seconds.
coretemp is probably about the best for windows, using relatively little resource small desktop footprint, but whatever takes your fancy.
I prefer TAT to coretemp as it gives me the worst case scenario.. Coretemp reads a few degrees lower.
Must vary with build coretemp for me reads a degree or so higher,is another reason i prefer coretemp:mrgreen:Quote:
it gives me the worst case scenario..
Well I'm thinking that Gigabyte's easytune is reading the temps in the bios as you click on a button with the same name i.e PC health........erm maybe it doesn't.
Temp in Tat is very high when using their CPU stress part........but some guys reckon it's not a real world test (it's worse) and orthos should be used instead.
Anyway I'm using core temp only because the guys in the know here are using it, and not because of detailed research by myself.
Well, I ran memtest while at work and came home to find that my machine had rebooted itself and 'recovered from a serious error' Looks like it's time to return the RAM and see what happens with the next batch. Do you think it's best to contact the people I got it from or Geil directly? Or should I carry on to find out which stick it is before returning it?