Overclock.net banner

1 - 20 of 64 Posts

·
Registered
Joined
·
202 Posts
Discussion Starter #1
For this discussion, a "Static Overclock" is defined as a state where the CPU Multiplier, Base Clock, and Vcore are constant. A "Dynamic Overclock" would be a state where the Mulit, Bclk, and Vcore change based on load.

Most of the overclocks I've seen have been static. The Bclk and multiplier are set at a fixed value and Vcore(and other volts) are set to the minimum needed to be stable under full load.

I want to try dynamic overclocking. I'm finding it to be quite tricky to get it 100% stable though.

Right now I've just set the Bclk to 170 and enabled Turbo Boost and CPU EIST Function and C3/C6/C7 State Support. This gets me 3570Mhz(170x21) running 4 workers in Prime95 and 4080Mhz(170x24) running 1 Worker.

Volts are all on auto. Load Line Calibration is off.

The Vcore seems really low. It's [email protected](Prime95 1 worker) and [email protected](Prime95 4 workers). When I run LinX at max memory settings (Problem size20460, Memory:3210MiB) it drops down to [email protected] Prime seems stable in short runs, while Linx fails after 6 or so runs.

So here's the tricky part. EIST/Turbo changes voltages. I tried changing the "Dynamic Vcore" setting in BIOS and it gave me the BSOD every time I changed it from AUTO.

Changing the "CPU Vcore" would seem to be an answer, but I'm worried about the dynamic changes pushing the volts over 1.4.

So, does anyone have advise or experiences with this?
 

·
Registered
Joined
·
409 Posts
Quote:

Originally Posted by CaptainSWJR View Post
For this discussion, a "Static Overclock" is defined as a state where the CPU Multiplier, Base Clock, and Vcore are constant. A "Dynamic Overclock" would be a state where the Mulit, Bclk, and Vcore change based on load.

Most of the overclocks I've seen have been static. The Bclk and multiplier are set at a fixed value and Vcore(and other volts) are set to the minimum needed to be stable under full load.

I want to try dynamic overclocking. I'm finding it to be quite tricky to get it 100% stable though.

Right now I've just set the Bclk to 170 and enabled Turbo Boost and CPU EIST Function and C3/C6/C7 State Support. This gets me 3570Mhz(170x21) running 4 workers in Prime95 and 4080Mhz(170x24) running 1 Worker.

Volts are all on auto. Load Line Calibration is off.

The Vcore seems really low. It's [email protected](Prime95 1 worker) and [email protected](Prime95 4 workers). When I run LinX at max memory settings (Problem size20460, Memory:3210MiB) it drops down to [email protected] Prime seems stable in short runs, while Linx fails after 6 or so runs.

So here's the tricky part. EIST/Turbo changes voltages. I tried changing the "Dynamic Vcore" setting in BIOS and it gave me the BSOD every time I changed it from AUTO.

Changing the "CPU Vcore" would seem to be an answer, but I'm worried about the dynamic changes pushing the volts over 1.4.

So, does anyone have advise or experiences with this?
Hey, bro
I oc'ed this i5 with tb on. For me, it only worked by determining a stable vcore for 4.0ghz (190x21) static. Then enabling tb/speedstep/c1e and c-state, multi at auto, BCLK 170x21 (to output 4.0ghz on one core) and setting vcore to an offset to output the same max voltage as was needed for the static oc. The amount set as offset + stock vcore should get you close to it.
IMC (vtt) needed an extra notch though.
From there on I could get to 4.2 on one core. This only gives me 3.6ghz on all four cores though.

To test, run prime95 on one worker.
 

·
Registered
Joined
·
202 Posts
Discussion Starter #3
Quote:

Originally Posted by genuine555 View Post
Hey, bro
I oc'ed this i5 with tb on. For me, it only worked by determining a stable vcore for 4.0ghz (190x21) static. Then enabling tb/speedstep/c1e and c-state, multi at auto, BCLK 170x21 (to output 4.0ghz on one core) and setting vcore to an offset to output the same max voltage as was needed for the static oc. The amount set as offset + stock vcore should get you close to it.
IMC (vtt) needed an extra notch though.
From there on I could get to 4.2 on one core. This only gives me 3.6ghz on all four cores though.

To test, run prime95 on one worker.
That's what I'm talking about. Thanks for the reply.
 

·
Registered
Joined
·
202 Posts
Discussion Starter #5
Quote:

Originally Posted by genuine555 View Post
You're welcome

Hope you'll hit 4.5ghz or something.
Heh, not likely with a 212+ as my cooling.

My motherboard has a feature called C.I.A.2 (CPU Intelligent Accelerator 2) that ups the base clock under load. I'm thinking about trying it out after I figure out voltages.

If i can get that working, I'd have ridiculously low speed at idle.
 

·
Registered
Joined
·
409 Posts
Quote:

Originally Posted by CaptainSWJR View Post
So, I've been tinkering a bit and this is my initial result.




Notice the 25x Multi in the second picture? I think it's just an miss-read. It just bounces there sometimes when I first run Prime95. It settles down to 24x after a couple of seconds.

So, what do you think?
Double post here, but nevermind


You got about the same results as me (3.6 and 4.2ghz)

Nice


Now let's see some
of stability...
 

·
Registered
Joined
·
202 Posts
Discussion Starter #10
I started Prime95 this morning before work. 11 hours in it's still cruising along at 3.78Ghz. Temps reached a max of 67-62-62-62. Right now it's 61-55-58-56 which is what it runs with Prime95 going. The max temps are from the LinX runs.

Going to let it finish 12 hours, take a screenshot, save the BIOS settings, then do some gaming


Edit: Couldn't wait the full 12 hours, close enough!
 

·
Registered
Joined
·
573 Posts
Hi,
Long time lurker, first time poster. It seems dynamic overclockers for this chip are rare, so could I trouble you with a cpl of questions?

1. What dynamic and vtt voltages do you have set in your bios for that overclock?

2. Is the cpu core voltage listed in easytune supposed to show 1.51250v? Tuner tab -> Advanced -> Voltage.
 

·
Registered
Joined
·
596 Posts
I've been out of the overclockers game for a few hardware generations, and recently looking to get back in... this is the first time I've heard of dynamic overclocking. What is the advantage to this technique? What makes this better than static overclocking?

And where can I go to read more about this idea?
 

·
Registered
Joined
·
452 Posts
dynamic oc ftw. ive been using this technique for a while on my i5. my only problem is that youtube videos and similar things will freeze it up every so often... itll run prime 95 24/7 without a hitch, but every now and then a video will freeze it up? its been confusing me for a while.
 

·
Registered
Joined
·
602 Posts
I have a semi dynamic overclock that I have been perfecting for a week or two now. I dont use speedstep though, only voltage offsets along with energy saving items such as C1E or whatever enabled. I've got a stable OC on my i5 750 @ 3.81 right now with core voltage varying between 1.01 and 1.244. This is my 24/7 OC. The core voltage cant go down any further without becoming unstable but I'm lowering things like IMC, PLL, DRAM Voltage, ect. I think I'm using a voltage offset of .10625 or something. I'll post my specs after I'm completely finished with it.
 

·
Registered
Joined
·
452 Posts
hey poppinj, i'm currently running what seems like your same oc but with speedstep enabled. i had to bump my vcore to 1.28 though.

does the voltage offset make it more stable than just straight manual voltage? ive been playing with both options, but it doesnt seem to make a difference with my oc.
 

·
Registered
Joined
·
573 Posts
I'm currently running 170x21 with all power saving features enabled and a vcore of + 0.05v and vtt of 1.18v. A vcore offset of +0.375 passed every stress test, but resulted in a handful of game crashes and one bsod. It was stable at 0.4375v, but I always add one notch more than the lowest stable voltage for peace of mind.

I've noticed that you can set VTT quite low on this chip and pass stress tests for hours on end, yet the system is not 100% stable. Applications will randomly crash until you push the VTT to much higher levels. This may be the cause of your lockups, spectre023.
 

·
Registered
Joined
·
602 Posts
Quote:


Originally Posted by spectre023
View Post

hey poppinj, i'm currently running what seems like your same oc but with speedstep enabled. i had to bump my vcore to 1.28 though.

does the voltage offset make it more stable than just straight manual voltage? ive been playing with both options, but it doesnt seem to make a difference with my oc.

I really dont think it makes it more or less stable so long as the voltage at full load is the same. So if the voltage is dynamic I have to ensure that it goes up to the full 1.24 while under load. This can be a pain to get right though since changing other voltages (IMC,PLL, ect) can also change my max core voltage.
 

·
Registered
Joined
·
452 Posts
Quote:


Originally Posted by whybother
View Post

I'm currently running 170x21 with all power saving features enabled and a vcore of + 0.05v and vtt of 1.18v. A vcore offset of +0.375 passed every stress test, but resulted in a handful of game crashes and one bsod. It was stable at 0.4375v, but I always add one notch more than the lowest stable voltage for peace of mind.

I've noticed that you can set VTT quite low on this chip and pass stress tests for hours on end, yet the system is not 100% stable. Applications will randomly crash until you push the VTT to much higher levels. This may be the cause of your lockups, spectre023.

that hit it right on the head man.. dropped my vcore to 1.24v and upped my vtt to 1.21 and i cant get it to freeze up no matter how hard i try to duplicate the conditions that made it crash before.

Quote:


Originally Posted by Poppinj
View Post

I really dont think it makes it more or less stable so long as the voltage at full load is the same. So if the voltage is dynamic I have to ensure that it goes up to the full 1.24 while under load. This can be a pain to get right though since changing other voltages (IMC,PLL, ect) can also change my max core voltage.

yeah. i tried setting a manual vcore of 1.28 while i adjusted all my other voltages, and once i got it seemingly stable i adjusted my offset voltage to allow for it to get 1.24v and no more. that coupled with the advice from whybother seemed to do the trick.
i cant get it to crash no matter what i do.. even with multiple instances of firefox, memtest, and prime running in the background and then starting and running crysis wont destabilize it.. although crysis barely moves under that load hahaha. now i'll just leave prime running and hope it doesnt crash under that alone for some reason.
 

·
Registered
Joined
·
202 Posts
Discussion Starter #19
Quote:


Originally Posted by whybother
View Post

Hi,
Long time lurker, first time poster. It seems dynamic overclockers for this chip are rare, so could I trouble you with a cpl of questions?

1. What dynamic and vtt voltages do you have set in your bios for that overclock?

2. Is the cpu core voltage listed in easytune supposed to show 1.51250v? Tuner tab -> Advanced -> Voltage.


I'm running with LLC on, Vcore on "Normal" and dynamic vcore offset +0.025V. Bclk is set to 180, memory divider is set to 2:6 I believe. My RAM is running at 1440 (1600 rated ram). I'm not using the windows based utilities, I'm adjusting in the BIOS.
I bumped one step on IMC(i think, may have been PCH or VTT) and was stable at 180 Bclk.

1.5125V is HIGH. I would get that down under 1.35V ASAP.

The reason I went with this was that I was bothered by the thought of pumping full OC'd voltage to my CPU at all times. I leave my computer on all the time, and most of that time is spent at idle. This way it's not drawing much at all. I probably should shut down at night and start it up when I get home from work, but my old system was so slow to boot I just left it on, and now I'm in the habit. Plus I like just tapping my mouse to wake it up and go.
 

·
Registered
Joined
·
573 Posts
Quote:

Originally Posted by CaptainSWJR View Post
1.5125V is HIGH. I would get that down under 1.35V ASAP.
This value is only shown in the easytune utility, and I suspect it's simply a software issue. It shows a value of 1.5125v regardless of the offset voltage set in bios. All other software shows my vcore never going above 1.3v under load, dropping to <0.9v idle.

Cheers for the info. I'm going to give thos settings a try, although I don't think my chip likes such a low vcore voltage.
 
1 - 20 of 64 Posts
Top