SOLVED Overclocking FX-4300 Specifically

[Holymartyr]

Yes I do, I've seen anywhere between just a few Mv, to as much as .195 Mv more.

Default is 2.500v.

When you say 'people' do you mean just posts you've read, or have you found something else?

I'm trying to keep my ram at 1600 MHz~ish, because well that's the default my ram should be running at. Until I have a fan blowing right onto my ram to help cool it, I don't really want to overclock it. AKA Yes, I can't really adjust my FSB too much without throwing the clock of my ram all outa whack.

What do you suggest, from what you've read, for an increase to the CPU PLL? And if possible, would you please link the posts that you got this information from? (if possible - thanks)

 

[RGone]

Hey man help us out. I had to go back one page and to the first post to find what it is you are running in that case.
How to put up Sig at OCF.


New Shortcut method for putting a Signature with your system information following your every post so people can know what is in the case that they are trying to assist with. You can use something like what is shown in my signature as a good template of needed information Thank you.

 

[Holymartyr]

I was looking for that yesterday, I couldn't find it but i did find a post somewhere on the forums saying that someone couldn't make a sig until they had X number of posts or was a member for X hours. Something like that.

Creating a sig with all my info now.

Update: I just spent 30 minutes making a nice sig for all of you, well formatted and displayed, and when I hit save it did not save. Trying 1 more time..

Update 2: Just went back to edit it and it seems it's there. Maybe users can't see their own sig on their posts. Let me know if it's working.

 

[RGone]

You posting from a cell phone? Stuff like that does not let us see it and if it is your first post with sig...well we will not see it until you get to regular computer.

 

[Holymartyr]

I am not.

But now I see the sig in this post.

Guess I had to post a new post AFTER the sig had been updated.

 

[RGone]

Yes I do, I've seen anywhere between just a few Mv, to as much as .195 Mv more.

Default is 2.500v.

When you say 'people' do you mean just posts you've read, or have you found something else?
What do you suggest, from what you've read, for an increase to the CPU PLL? And if possible, would you please link the posts that you got this information from? (if possible - thanks)

Not everyone is suggesting or doing anything with CPU VDDA aka CPU_PLL. It is like any other tip/trick that is uncovered when overclocking.
1. It does not work for everyone.
2. It is not in the case of CPU_PLL a 'certain' voltage is the right one.
3. I doubt I would use over 2.6V on CPU_PLL, I have however seen others use more.

Here is link to a thread here in the forum with a user having an FX-8350 that would not clock for crap like a LOT of FX piledrivers seem to be having trouble with. You can see why I suggested CPU_PLL adjustment for HIM and how it worked out for him after hours of skilled trial error apparently. It is a good read since it is just another re-inforement of some PileDriver cpus just don't seem to clock-up high very well and quit before a lot of BullDozers did.


New to the forum but not my first rodeo.

 

[Holymartyr]

Well I've just tried overclocking it further again, but with no success.

I raised the CPU PLL to 2.55v, and raised my VCore to 1.50v.

I first started off with keeping my FSB at 240, and raising the multi of my CPU up one notch bringing it to 4680 (I think) but the CPU was less stable than it is now. So I tried pulling the CPU, NB, HT and Ram down to 4000ish~MHz, 2100ish~MHz, 2100ish~MHz, and 1300ish~MHz (respectively) so I could increase the FSB and see if that would make a difference. I got my FSB to a maximum of 284 before it wouldn't boot. Once I got it too boot, I raised the multi of my CPU to around 4550ish~MHz and tested for stability. At no setting could I get it any more stable. When I tried bumping the multi up one more past 4550ish~MHz, it would either BSOD or fail Prime95 even faster.

Temps were always within range. Using HWiNFO64, the temps reported are as follows WHILE CPU was being stressed:

AMD 10h+ CPU Thermal Sensor - CPU 0 = 46 Degrees Celsius
ITE IT8720F - Motherboard = 30
ITE IT8720F - CPU = 32
ITE IT8720F - Northbridge = 56 (fixing with a 40mm fan on the NB heat sink soon)

So I still have no clue on how to get it stable.

When I back off the clocks just a bit, no matter what the FSB or Multi is, My 1st, 3rd, and 4th core always do just fine running Prime95 - but my 2nd core fails still almost every time. It seems my 2nd core has some sort of defect, or something else wrong with it, and that is what is holding back my max overclock.

Any ideas?

 

[Anonaru]

Well I've just tried overclocking it further again, but with no success.

I raised the CPU PLL to 2.55v, and raised my VCore to 1.50v.

I first started off with keeping my FSB at 240, and raising the multi of my CPU up one notch bringing it to 4680 (I think) but the CPU was less stable than it is now. So I tried pulling the CPU, NB, HT and Ram down to 4000ish~MHz, 2100ish~MHz, 2100ish~MHz, and 1300ish~MHz (respectively) so I could increase the FSB and see if that would make a difference. I got my FSB to a maximum of 284 before it wouldn't boot. Once I got it too boot, I raised the multi of my CPU to around 4550ish~MHz and tested for stability. At no setting could I get it any more stable. When I tried bumping the multi up one more past 4550ish~MHz, it would either BSOD or fail Prime95 even faster.

Temps were always within range. Using HWiNFO64, the temps reported are as follows WHILE CPU was being stressed:

AMD 10h+ CPU Thermal Sensor - CPU 0 = 46 Degrees Celsius
ITE IT8720F - Motherboard = 30
ITE IT8720F - CPU = 32
ITE IT8720F - Northbridge = 56 (fixing with a 40mm fan on the NB heat sink soon)

So I still have no clue on how to get it stable.

When I back off the clocks just a bit, no matter what the FSB or Multi is, My 1st, 3rd, and 4th core always do just fine running Prime95 - but my 2nd core fails still almost every time. It seems my 2nd core has some sort of defect, or something else wrong with it, and that is what is holding back my max overclock.

Any ideas?

I'm 97.5% sure you are having an issue with the UD3 mobo-- The 2nd worker (2nd core of the first module I believe) fails on both of my UD3's when the voltage isn't just right. This board has a few issues with vdroop, and this is the board where a large increase to CPU-PLL makes the most noticable increase to stability (And also the one easiest to replicate in multiple cases). Your voltage here is safe beneath 2.710 (I wouldn't go above 2.675 for any reason) as far as I've been able to uncover. This board is wonky with PD parts.. I'd try upping your LLC (If you aren't on ultra already. Don't use ultra with a 1.5vCore) then change PLL to 2.665 for a single run, just to see if it makes it any farther in your run.

 

[ssjwizard]

Ultra LLC doesnt even work on my UD3. I find If I set very high I get ~ .07v rise under load. If I set ultra it will only add ~ .01v which is even less than what regular mode does.

 

[Anonaru]

Ultra LLC doesnt even work on my UD3. I find If I set very high I get ~ .07v rise under load. If I set ultra it will only add ~ .01v which is even less than what regular mode does.

Haha, I've heard of this, but not on my boards.. If I set it to "Ultra" from very high, it adds ~.08-.09v, which isn't much more than vHigh, but is always worth a shot. I don't like my ultra mode anyways, it seems to supply "dirty" power, the value jitters between .06 and .10.

Perhaps it is a 4 pin CPU power cable vs 8 pin issue with LLC not working right on these boards?

 

[ssjwizard]

Actually I feel it necessary to point out that my Gigabyte 990FXa-UD3 is a Rev 1.0, so there could definitely be a difference in versions here.

 

[Holymartyr]

I have a Rev 1.1 board, my LLC is on High ATM, and I will try upping to 2.665 PLL and see if it makes a difference. Will update this post with results.

Update! (Wooo!)

Just tried the PPL increase, and stress tested for 10 minutes and its STABLE! (I know its not a lot, but the 2nd core would always fail within 2-3 minutes. 10 minutes is certainly a record.)

 

[Anonaru]

Actually I feel it necessary to point out that my Gigabyte 990FXa-UD3 is a Rev 1.0, so there could definitely be a difference in versions here.

Aye, the 1.0 only got LLC control through BIOS updates I believe-- The 1.1s and newer employ LLC properly, but still suffer vdroop

Very High works very well, Ultra pushes a bit too in my opinion

 

[Holymartyr]

Oh, should I be using Very High? (Currently using High)

And why is Ultra too much? I have a basic understanding of LLC, but why does Ultra "push it"?

P.S. Check my last post before this one (CPU is stable now - more info in that post)

 

[Anonaru]

Oh, should I be using Very High? (Currently using High)

And why is Ultra too much? I have a basic understanding of LLC, but why does Ultra "push it"?

Well, all LLC options are "good." They basically allow you to over-volt your CPU only when it needs the extra power, allowing you to run a lower voltage 24/7. In theory, the higher your "LLC" the better. Obviously only in theory

My problem is with ultra is the extreme increase it gives, while not dangerous in its own, adds another variable into the works-- If it doesn't hold this voltage just right, or if it is too much / not enough, then you might get problems. I've only used ultra once in actual OC, and it was to prevent me from going too dangerously high on vCore while doing a higher OC in the shop.

Edit: Glad to see you got a bit of stability-- Will need to run it for a while just to be sure, but I've definitly found that tiny bit of extra "juice" helps with these new PD parts. The good thing is, if you pass a few hours of P95, you can start inching off the PLLv and most likely get it down significantly (Mine ended up at 2.565 and 2.585 at their current clocks.)

 

[Holymartyr]

Ok so where should I take my overclock from here?

I'm not sure if I should raise the FSB or the Multi. I had issues when I began overclocking the FX-4300 that when I only increased the CPU Multi and left the FSB at 200, AND increased the NB to 2200 MHz from 2000MHz, Windows 7's Aero theme would stop working. When I lowered it was fine. If I increased voltage for CPU-NB AND Northbridge Voltage, it wouldn't change.

But when I increased my FSB, and change the multi to get to 2200 MHz on the NB, Windows would boot just fine. Aero would work as it should.

So I feel I do need a higher FSB, and everyone seems to be saying that Piledriver loves a higher FSB. Just not sure what a good medium is, and/or how to find that. Obviously 200 doesn't work well, but pushing the FSB to my maximum 285~ish probably isn't good either..

Ideas?

 

[Anonaru]

Ok so where should I take my overclock from here?

I'm not sure if I should raise the FSB or the Multi. I had issues when I began overclocking the FX-4300 that when I only increased the CPU Multi and left the FSB at 200, AND increased the NB to 2200 MHz from 2000MHz, Windows 7's Aero theme would stop working. When I lowered it was fine. If I increased voltage for CPU-NB AND Northbridge Voltage, it wouldn't change.

But when I increased my FSB, and change the multi to get to 2200 MHz on the NB, Windows would boot just fine. Aero would work as it should.

So I feel I do need a higher FSB, and everyone seems to be saying that Piledriver loves a higher FSB. Just not sure what a good medium is, and/or how to find that. Obviously 200 doesn't work well, but pushing the FSB to my maximum 285~ish probably isn't good either..

Ideas?

I'm finding these get along very well with a FSB of about 220. I've gotten great results in the 216-236 range, but keep in mind that the PD processors don't like having a HT / NB above 2400, and are extremely difficult to get stable beyond 2500MHz. Multi-Only OC doesn't give you the performance that a combination of FSB / multi does, and the second option typically allows higher OC with lower vCore requirements

 

[Holymartyr]

Well this seems to be the most stable I can get it.

I tried upping the FSB in 5 MHz increments from 240 to 265, and changing the multi to keep my CPU around 4.6MHz, and NB/HT/Ram below 2400/2400/1600 respectively. Unstable no matter what I tried. Even upping the Vcore to 1.500V.

So then I tried the opposite, going in 5MHz increments down from 240 to 226. Once I hit 226, the Windows 7 Aero issue came back for me. Still I couldn't get it stable.

So kinda stuck and not sure what to do. It's still my 2nd core that's failing. Do I just have an outlier CPU that performs less than those who have gotten to 5.0 GHz? Or did they have an outlier CPU and mines 'normal'?

Or something I just thought of, is my ram speed limiting my CPU by any chance? Or my NB or HT by any chance?

Ideas? o.o

 

[Anonaru]

Well this seems to be the most stable I can get it.

I tried upping the FSB in 5 MHz increments from 240 to 265, and changing the multi to keep my CPU around 4.6MHz, and NB/HT/Ram below 2400/2400/1600 respectively. Unstable no matter what I tried. Even upping the Vcore to 1.500V.

So then I tried the opposite, going in 5MHz increments down from 240 to 226. Once I hit 226, the Windows 7 Aero issue came back for me. Still I couldn't get it stable.

So kinda stuck and not sure what to do. It's still my 2nd core that's failing. Do I just have an outlier CPU that performs less than those who have gotten to 5.0 GHz? Or did they have an outlier CPU and mines 'normal'?

Or something I just thought of, is my ram speed limiting my CPU by any chance? Or my NB or HT by any chance?

Ideas? o.o

I think your 2nd core might really just be a power-hungry grumper in this case. You could try kicking the PLL up a few bumps and setting your LLC to "Very High," but monitor voltages and temps carefully if you do this. I've had very good luck with PD parts, but there's always the possibility of a 'bad apple'

Still, I've not heard of anybody who couldn't get 4.6 on this board

Edit: You should set your RAM down to the "1333" or "1066" multiplier and set the timings to SPD values, or relatively loose. This will allow you to raise the FSB more without it causing non-CPU stability issues. Just in case you haven't taken this step!

 

[RGone]

@ Anonaru, have you heard of the PileDrivers being less clockable Mhz wise but actually doing 'more' work even though not reaching the bigger Mhz of the BullDozers?

I know you have a computer store and at that point price becomes a consideration and you have to 'use' what you sell to know what is going on. Well let me say that using makes you more able to serve your customer. Something the big boys only do from an engineering standpoint and likely little real hands-own. I am saying I don't like the cheap UD3, but I can glimpse why some use them. So your using; makes it easier to know what the UD3 will do.

The OPs windows areo goes nuts because CPU_NB is too slow for ram transfer effectiveness or a quirk in bios setting NOT visible to us. It picks up when he FSB overclocks but the same thing could be done by just setting the CPU_NB divider for more speed. Of course the CPU_NB voltage should be raised to complement the increased CPU_NB.

None of this stuff is rocket science nor is what works on one a sure thing on another system. Freeken lot of trial and error.

Happy Holidays "Anonaru". RGone...ster.