EPT
Simple Questions Simple Answers - Page 59
| Forum Index > Tech Support |
|
TheToast
United States4808 Posts
| ||
|
JingleHell
United States11308 Posts
On February 29 2012 01:41 TheToast wrote: So good to see JingleHell back Somehow I doubt that's a majority opinion, but thanks. | ||
|
Josh_rakoons
United Kingdom1158 Posts
| ||
skyR
Canada13817 Posts
| ||
|
Josh_rakoons
United Kingdom1158 Posts
On February 29 2012 02:32 skyR wrote: I assume you have a Gigabyte board... in which case, dynamic vcore is not called DPI, it's called DVID. It adds or subtracts from the voltage that the CPU is requesting for. Yup, you're right, it was DVID. Still got the same problem though. No vcore option. | ||
|
skyR
Canada13817 Posts
| ||
|
Josh_rakoons
United Kingdom1158 Posts
| ||
|
skyR
Canada13817 Posts
| ||
|
Josh_rakoons
United Kingdom1158 Posts
| ||
|
JingleHell
United States11308 Posts
| ||
|
Josh_rakoons
United Kingdom1158 Posts
On February 29 2012 02:44 JingleHell wrote: You should be changing it by REALLY small increments, to be more specific. And no. Reset the QPI to default and don't touch it again. Ofc.. This is something i AM going to be working through very thoroughly. | ||
|
Josh_rakoons
United Kingdom1158 Posts
| ||
|
JingleHell
United States11308 Posts
On February 29 2012 02:46 Josh_rakoons wrote: Then what is the requesting voltage for an i5 2500k? Varies by mobo and BIOS, and what the chip actually needs varies wildly. What you should be asking is "between increments x, y, and z, which should I raise it by first?". | ||
|
Josh_rakoons
United Kingdom1158 Posts
| ||
|
JingleHell
United States11308 Posts
Thus, what matters is whether it's stable at the clock you want, and if not, you need to raise the DVID number by a SMALL increment, somewhere in the 1-2mv range, and test for stability again. To give you an idea, to turn on HT on my i7 930@3.8, I go from 1.25625v Vcore to 1.2625v Vcore, and that's a HUGE jump that causes heat issues long term (partly due to HT being insanely hotter clock for clock.) Mind you, that's with a proper manual Vcore setting, and it's for a different generation CPU, and it serves ONLY as an example of the numbers involved and what counts as a big jump, NOT any indication of what you should be using. | ||
|
Josh_rakoons
United Kingdom1158 Posts
I know that i need to do stress testing at those clocks but i originally thought that if you didnt raise the voltage but you raised the clock, the system would refuse to boot. Therefore, i came to this conclusion. | ||
|
JingleHell
United States11308 Posts
On February 29 2012 03:01 Josh_rakoons wrote: Well, earlier when i accidentally raised the clock to 4.5 ghz and the QPI to 1.24v i forgot to change the clock back down and it booted fine. Does that mean its a relatively stable clock? No. It means you booted. That's like saying you woke up in the morning, so you don't have high cholesterol. For a stability test, you need a CPU stress test, run 6-8 hours. | ||
|
Josh_rakoons
United Kingdom1158 Posts
| ||
|
JingleHell
United States11308 Posts
On February 29 2012 03:07 Josh_rakoons wrote: Ok. just changed the clock back to 4.5ghz and will now run a prime95 mark for the next few hours. What settings should i be using on prime95? Torture test, blend, with an appropriate number of threads. As in, one per core, unless you have HT, at which point 2 per core. There's other stress tests you can use if you have end up running across stability issues later. Different tests do different things different ways, and some things won't hammer it appropriately for the load you end up using. P95 is a good baseline for mild to moderate OCs though. | ||
|
Wabbit
United States1028 Posts
On February 29 2012 03:04 JingleHell wrote: No. It means you booted. That's like saying you woke up in the morning, so you don't have high cholesterol. LOL @Josh_rakoons I would initially do 10 to 20 cycles of IntelBurnTest / LinX (same thing) with all available RAM. That'll find instability quicker than Prime95. Monitor your temps closely though, as those programs (linpack essentially) tend to heat up the CPU more than Prime. They're not kidding when they call it "burn test". If it goes above 75*C I'd stop the test. Idk if 6+ hours of Prime95 blend is still required to prove complete stability after passing linpack. | ||
| ||
