• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

My 1st serious overclock... with an arguably questionable stability (newest Prime95 displays issues)

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

EarthDog

Gulper Nozzle Co-Owner
Joined
Dec 15, 2008
Location
Buckeyes!
The concern comes when people don't know you have another thread and have to ask the same questions over (wastes your time, and the volunteers helping). This is all about tweaking your '1st serious overclock' so far. :)

It's six of one, half dozen of the other (offset/manual) to me. Just a preference. I used to run fixed all the time... never worried about chip life........ and I wasn't using a chip that could be replaced for $5. :)
 
OP
MiracleMatter

MiracleMatter

Registered
Joined
Aug 23, 2022
The concern comes when people don't know you have another thread and have to ask the same questions over (wastes your time, and the volunteers helping). This is all about tweaking your '1st serious overclock' so far. :)

It's six of one, half dozen of the other (offset/manual) to me. Just a preference. I used to run fixed all the time... never worried about chip life........ and I wasn't using a chip that could be replaced for $5. :)
Ok stop with this 5 $ thing, they're at least 50 $ used and beat-up if you include shipping...

I might try the offset thing eventually but since I've only confirmed stability with manual voltage and I can just restart the computer and switch between bios profiles in a few button presses... I'm not sure I want to bother testing stability all over again with offset type voltage... at least for now. But I consider the voltage drop question answered. So thanks for that !
 

EarthDog

Gulper Nozzle Co-Owner
Joined
Dec 15, 2008
Location
Buckeyes!
Ok stop with this 5 $ thing, they're at least 50 $ used and beat-up if you include shipping...
There are several that sold and shipped for ~$15. Incredibly cheap (how much is your time worth?) was the underlying point there. Hell, when you get access to our classifieds, you can post something in the WTB section and you may get someone to one to you for shipping only. Old... abudnant...cheap. :)

Totally get the stability testing again thing. The good news is that you already know what voltage is stable on fixed. All you need to do is set your offset to match that value under load and it 'should' be stable. Also, the process can be simplified, but some like to play with several apps and such. If minutia is in the cards, it's in the cards, lol.
 

Blaylock

"That Backfired" Senior Member
Joined
Jun 5, 2013
Location
Go Blue!
@MiracleMatter , is there any chance you can set up your signature? I've had to go back to the first page a few times to remember what you're working with. (Most recently trying to figure out what the hell a $5 chip is. LOL) At any rate, setting up your signature will help loads in the future. TIA
 

EarthDog

Gulper Nozzle Co-Owner
Joined
Dec 15, 2008
Location
Buckeyes!
(Most recently trying to figure out what the hell a $5 chip is. LOL)
Heh, that was Neb's link to ebay. The first FX-6300 is $5 currently (auction still active), so I said, $5 dollars. My link is recently completed auctions where one can easily get them for around $15 (closer to $5 than $50+ :bday: ).
 
OP
MiracleMatter

MiracleMatter

Registered
Joined
Aug 23, 2022
Heh, that was Neb's link to ebay. The first FX-6300 is $5 currently (auction still active), so I said, $5 dollars. My link is recently completed auctions where one can easily get them for around $15 (closer to $5 than $50+ :bday: ).
If you can provide me with the means to teleport this shipment to my appartment for free AND make sure I win the bid before it gets near 40-50 $, then I'll concede your point. Realistically, the effective price is around 50 CAD... and then I'll have to have more thermal paste too because the only one left is well squeezed between my current CPU and the heatsink.

@MiracleMatter , is there any chance you can set up your signature? I've had to go back to the first page a few times to remember what you're working with. (Most recently trying to figure out what the hell a $5 chip is. LOL) At any rate, setting up your signature will help loads in the future. TIA
What should I put in my signature ? Just my current setup ?
 

EarthDog

Gulper Nozzle Co-Owner
Joined
Dec 15, 2008
Location
Buckeyes!
If you can provide me with the means to teleport this shipment to my appartment for free AND make sure I win the bid before it gets near 40-50 $
I listed completed auctions that went for $15 with shipping. BUT...... you just mentioned that you're in Canuckistan, so, that changes things, lol (don't hesitate to list a location in your profile.. :p). Still you can find them from $30CA shipped (buy it now just went for 29.12 shipped yesterday) but many are close to that $50 mark, indeed. Context. lol.
What should I put in my signature ? Just my current setup ?
Yes.
 
OP
MiracleMatter

MiracleMatter

Registered
Joined
Aug 23, 2022
Hi everyone, thank you for all your inputs so far.

I am back here essentially because I watched this video and fell into the whole LLC (Load-Line Calibration) rabbit hole last week-end. I do see from the search results that the critical knowledge contained within this video is already the topic of a few threads of this forum from several years ago (and technically, I'm the n00b here) so I won't introduce it.

The apparently stable overclock I attained that is already mentioned in this thread has the CPU LLC option set to "Ultra high" (75%) with that BIOS voltage setting of 1.4 V (I'm calling it "setting" to distinguish it from the real core voltage). This was the apparent consensus of all the overclocking guides I consulted. Hence, I am now a little concerned that my overclock is probably not that stable for mainly two reasons:

For one, the purpose of running Prime95 and such is to confirm stability in the worst case situation. If the instantaneous core voltage is not at it's lowest during at least a large fraction of the stability stress test time, what is even the purpose of stress testing ? And second, when the LLC setting was at "High" (50%), I recall the Prime95 test failed after a long while with every other parameter identical to my current overclock profile. All the more reasons to think the stability of this profile is marginal at best and any downward Vcore spike can corrupt the result of possibly critical instructions.

So the upside of disabling LLC is obvious but is there really a downside ? Obviously, one would think achieving a perfectly stable overclock at a lower core voltage setting is one not to be overlooked. After all, CPUs have voltage tolerance limits. However, I seriously wonder if it is applicable at all in my situation even if I choose to further overclock knowing I am unwilling to compromise stability. I question it because as far as I recall, temperature is the main limiting factor for a safe overclock and even if it isn't, the maximum voltage a CPU should be subjected to when idle is higher than when under load (or when hotter, I don't remember if it's a matter of temperature or current).

Should I even be using LLC at all ? If yes, which percentage should I set it to ? (0 to 100 % by increments of 25%)
 

ShrimpBrime

~MadHatDeLidder~
Joined
Apr 19, 2012
LLC is for all core load v-droop. You adjust it accordingly to temps vs stability. If the cpu remains stable when it drops, you could essentially lower the vcore and raise the LLC.

These chips like only 2 things. And is all you need to know actually.

Colder temps and more v-core.

So if it's not stable, all you have to do is apply better cooling or more v-core which ever comes first.
 

EarthDog

Gulper Nozzle Co-Owner
Joined
Dec 15, 2008
Location
Buckeyes!
+1... LLC is LOAD thing (single/all threads doesn't matter) and is supposed to minimize vdroop. It minimizes the spread between idle state and load. I like my LLC level to match what I set in the bios. So if I want 1.4V load and it 'droops' to 1.35v, I'll raise LLC so load is 1.4v.

If you're stable under load at say, 1.4V, then you should be stable at 1.4V no matter how you get there. Adaptive, offset, fixed, llc1-4, whatever.

Downside.... you need to set a higher voltage in the bios to compensate for the vdroop under load. LLC is an Intel/cpu spec... its supposed to happen. But for higher overclocks, it tends to help with stability in that transition from idle to load. At stock, it doesn't matter... but when it's running over spec....

But that's the name of the game dude... use the least amount of voltage to be stable (stable defined as stable for YOU. If that means overkill multiple stress tests or just one, whatevs). Makes sense you start throwing errors sooner when you lower the voltage.;)
 
Last edited:
OP
MiracleMatter

MiracleMatter

Registered
Joined
Aug 23, 2022
Thank you for all these advises. I settled for an LLC setting of High (50%) and an increased of Vcore setting by 25 mV. I concluded with my researches (and based on what I already know about integrated circuits) that the harm in using too high a voltage gets worse under a heavier load. Eg.: 1.5 V at 1 % CPU usage =/= 1.5 V at 100 % CPU usage.
A higher Vdroop also means a wider stability margin between the stability testing conditions (LinpackXtreme) and the more realistic heavy usage I will need this overclock for so I think I prefer it this way; with not too much LLC even if it is very convenient to pass stability tests. After all, passing these tests is NOT the end goal, rock solid stability in real life usage is.


In the meanwhile, I've accumulated many more questions (all related to my CPU overclock so far) that I would ask all in this one thread rather than assign a different thread to each. They are mainly geared toward understanding what part of my system will be at fault if my system fails a test, freezes, BSODs, etc. Because, for example, changing the base clock affects everything and I don't want to end up raising the voltage of a component a whole lot only to discover much later that it wasn't the cause of my system failing. Hopefully you understand that asking questions directly is a lot more time efficient than spending an entire day or more stability-testing for each one given how rigorous my stability standard is.

1. Can a CPU-only overclock be pushed further by using the base clock + CPU ratio instead of just the CPU ratio other than just for fine-tuning the frequency beyond the level of precision accessible with just the CPU ratio ? For example, can it make any difference at all on stability to have 20x225 MHz (or 25x180 MHz) instead of 22.5x200 MHz even thought it leads to the exact same frequency ? (I've even seen an article implying a CPU runs less hot with a higher base clock and lower ratio for the same frequency)

2. Does the base clock affect anything other than the frequencies of CPU, DRAM, HT speed and North Bridge ? (you know, all the crap that already have an adjustable multiplier in the BIOS.) Do I have to worry about anything other than these four frequencies if I increase the base clock by even a crazy amount ? For example, can I hypothetically get it to 400 MHz if I halve all ratios and be sure it'll be stable or is there (I presume) something else that'll prevent that from happening and if it does, around what frequency is that likely to happen ? In other words, up to which base clock frequency can I be sure that any form of instability is caused by the overclocked component(s) with a ratio and not anything else ?
2b. Also, throw back to question 1. Could I encounter any problem by bringing down the base clock a whole lot and compensate by raising up the ratios ?

3. Can this base clock on 990FX motherboards affect or hurt graphics cards through the PCI-E frequency or does it remain independent ? (I've seen in a few places that it doesn't even bring any gain to the table to increase the PCI-E clock speed and that it can even damage components, is that true ?

4. Is there a way for me and my FX-6300 in my M5A99FX PRO R2 to overclock different pair of cores to different frequencies so that I don't have to shut down core pair(s) to reach a higher frequency ? I'm asking since I know that "turbo core" thingy can do that to my CPU when using its default BIOS settings and I noticed all throughout my past stability testing that the core pair #4-#5 is always the one producing errors. I don't expect to be able to do it through BIOS necessarily so maybe there is an overclocking software ? I've been avoiding those so far but if you recommend one... I would try it.

5. Speaking of BIOS, I never updated its firmware because as far as I recall, we're never supposed to update it unless we need to. Am I wrong about this one ? The BIOS version of my mobo is 2201 (M5A99FX PRO R2.0 - Support). Is there a way to know all of what the two newer versions do ? I know it says something in one line under each entry but is that really the only change made ?

That will be all for now. I will ask the rest later if my future web searches don't end up giving me the answers I need.