Overclock.net banner

1 - 7 of 7 Posts

·
Registered
Joined
·
385 Posts
Discussion Starter · #3 ·
Quote:
Originally Posted by PulkPull;11948157
Rounding errors can be from unstable cpu OC. If you cpu volts are not too high already and your temps can take it, bump up your cpu volts 2 to 3 notches.
The temps are OK I'll try push it a little more then, I've got it set to +0.110v and I think that my vcore is 1.176 so it should result in 1.280v to 1.300v with that offset.

But what if the vcore is too high, do I need to find another bootable bckl * multiplier then?
 

·
Premium Member
Joined
·
6,404 Posts
You could try adding a touch of extra voltage to NB and the memory. You could be getting errors from the voltages on various parts being too different. Memtest86 is pretty sequential about testing things, the more random access from prime95 and real use is actually more stressful to the memory controller.

Remember: Memtest checks memory, prime stresses things.

I do agree that if you're only getting minor rounding errors that your voltages are probably very, very close to where they need to be to get 100% stable. You should only add a little more to cpu, memory and NB, until you get there.
 

·
Registered
Joined
·
385 Posts
Discussion Starter · #5 ·
Quote:
Originally Posted by allikat;11948218
You could try adding a touch of extra voltage to NB and the memory. You could be getting errors from the voltages on various parts being too different. Memtest86 is pretty sequential about testing things, the more random access from prime95 and real use is actually more stressful to the memory controller.

Remember: Memtest checks memory, prime stresses things.

I do agree that if you're only getting minor rounding errors that your voltages are probably very, very close to where they need to be to get 100% stable. You should only add a little more to cpu, memory and NB, until you get there.
Thanks, how do I find out how my my DRAM Voltage and NB Voltage is? It's set to "Auto" in my BIOS now and when I try to raise it, it starts at 1.5v and it says that going to 1.65 will damage my CPU.

*Edit*

I actually found the voltages in the MSI Control Center now I don't lnow if these are safe to change:

CPU: 1.296v ( pending between that and 1.3v )
ICH Voltage: 1.500v
3.3v: 3.395v
QPI: 1.350v
DRAM Voltage: 1.642v
5V: 4.961v
CPU PLL: 1.800v
12v: 11.880v

Another thing, my RAM is set to 1665Mhz but stock its 1600Mhz, should I lower the uncore/multiplier for the RAM's to be around 1300 instead?
 

·
Registered
Joined
·
699 Posts

·
Registered
Joined
·
385 Posts
Discussion Starter · #7 ·
That looks like the same memory. I've got 2 * 6GB Kit though, bought them separately. I'll bump up the DRAM Voltage to 1.6 then.

[ Edit ]

Got the rounding error on one thread after 12 minutes with DRAM Voltage 1.594v ( 1.6v in BIOS ). Should I up the DRAM voltage a bit or the vcore?

[ Edit 2 ]

I've made the following changes now:
VCore: 1.304v
DRAM Voltage: 1.632v
Multiplier: 25 ( same as before )
BCKL: 163 ( previously: 167 )

Meaning that I am lowering my OC a little bit, but still at 4.1Ghz. I've had to up the offset a little bit more on the vcore so the cpu is still a bit hot but seems more stable than anything else that I've tried today. So far I've successfully made 1 hour of small FFT-tests with prime95 reaching a top temperature on core #0 at 79 celcius.

It's currently running Test 4, 340000 Lucas-Lehmer iterations of M329727 using FFT length 16K. If this works out of a couple of hours more / over the night, I might try lowering the voltages a bit more to see if I can get the CPU cooler.

The RAM is also set to 818MHz ( =1636Mhz DDR ) so I hope they will hold up fine aswell.
 
1 - 7 of 7 Posts
Top