• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

FX8350 Overclocking FSB vs Multiplier

DMF1986

DMF1986

DMF1986

Associate
Joined
17 Nov 2013
Posts
407
Location
United Kingdom
Lets see if we can get some information on the best overclocking method for the FX8350 be it either using the FBS or the multiplier.

I currently have my CPU at 4.7Ghz @ 1.47v using the multiplier only FSB just left at the stock 200.

I have been told that using the FSB and leaving the multi @ 20 will help reduce my voltage therefore heat generation and also you get better performance.

Doing this effects the RAM, NB & HT
If using the FSB what's the best voltages you use (CPU - CPU/NB)

I assume I would also need to increase my CPU/NB voltage which is stock at 1.2v. I have heard 1.3v is safe for this but not completely sure.

I am hoping to hit 5Ghz completely stable its under water with a XSPC 480 & a 280 so I have sufficient cooling available
 
I have managed to get my FSB up to 260 by using this advice which is a couple of pages back on the official 83XX thread:

ALXAndy said:
From my research I found that raising the FSB came at a much lower temperature than hitting the multi. So basically for my 4.9ghz bench stable (and I ran the torturous Asus Realbench 2.0) I could only achieve it with a balls high FSB and my 20.5 multi. If I raise my multi any more my CPU becomes unstable and temps will soon see the PC locking up.

I was given some advice the other day from a mate of mine who is a bit of an expert with overclocking and he basically told me (this was for my Xeon but I would imagine it's similar in practice to AMD) that I should lower my mutli to the lowest setting (this is a Xeon 1366) and then start cranking the bus at 5mhz intervals. Once I fail to post or get into Windows take the highest stable setting and then start working on the multiplier.

Which tbh seems like sound advice. He also told me that bclk or FSB was not about how good the motherboard was but more about the CPU itself.
Click to expand...

Raising the FSB does change the other parameters, though the Asus mono is nice as if left on auto, it will find the setting which most closely resembles what is stock frequencies. However, if you want to overclock your ram (like I have to around 2133Mhz) it gives you a range of options from which you can choose the one closest to your rated ram speed etc.

The same is true of your CPU/NB and HT frequencies. I have got mine to 2340 without having to change the voltage - and I vaguely remember reading somewhere on that thread that there are only limited gains to be had passing 2400Mhz but I am not sure if I am just telling myself that so I don't have to try overclocking something else which is harder to measure :P

Edit: also, forgot to say that on the Crosshair mobo, the lowest multiplier is 4x. From this, once you have found the highest bus freq, you can then go on to increase the multiplier. For this step I followed the guide on the Asus ROG site, where you put in the multiplier you want (so for me 18.5x) and a voltage of 1.5V and keep reducing the voltage until you have the best mix of temps vs stability

Edit 2: Forgot to add the link! http://rog.asus.com/216552013/cross...g-fx-8350-to-4-8ghz-on-crosshair-v-formula-z/
 
Last edited:
The best and highest overclocks i have achieved on the fx series have always been on both fsb and multiplier.

I have owned four fx 8 cores and have clocked over a dozen with teh same procedure.

First i disable all the normal power saver stuff. I ramp my cpu/nb up to 1.275/1.3 off the bat, since i am under water, loss of stability through heat is not an issue. This should give some more stability to the memory controller, which is vital when fsb clocking.

To remove any more memory errors while finding CPU limits, clock the RAM multiplier down to its lowest speed (this can be re-clocked later after the CPU sweet spot is found). Then i start raising the fsb and voltage as i need it, clocking down HT and NB near to stock (this can be clocked afterwards like RAM).

Eventually you will hit a wall on fsb. Increase the multiplier by 0.5 and if you cant get stability, drop the fsb a little. Sometimes you will only be able to raise the multiplier by 0.5 or 1 after dropping the fsb by a few notches. Sometimes the multiplier will be more stable lower than stock with a much higher fsb. Experiment with different combinations and it wont be long till you gain a feel for the best combination of fsb and multiplier to hit that magic clock.

After the CPU limits are found, i normally lower the clock a fraction and start working on the ht, leaving nb frequency at stock. After that i go on to RAM but concentrate on tightening timings rather than max frequency (just what i prefer to do, performance wise is probably the same).

If you get 0124 black screens, then try lowering/increasing cpu/nb voltage between 1.15 and 1.3. This will solve it if the problem was a voltage issue with an unstable memory controller.

GL and keep us updated!
 
Avenged7Fold said:
The best and highest overclocks i have achieved on the fx series have always been on both fsb and multiplier.

I have owned four fx 8 cores and have clocked over a dozen with teh same procedure.

First i disable all the normal power saver stuff. I ramp my cpu/nb up to 1.275/1.3 off the bat, since i am under water, loss of stability through heat is not an issue. This should give some more stability to the memory controller, which is vital when fsb clocking.

To remove any more memory errors while finding CPU limits, clock the RAM multiplier down to its lowest speed (this can be re-clocked later after the CPU sweet spot is found). Then i start raising the fsb and voltage as i need it, clocking down HT and NB near to stock (this can be clocked afterwards like RAM).

Eventually you will hit a wall on fsb. Increase the multiplier by 0.5 and if you cant get stability, drop the fsb a little. Sometimes you will only be able to raise the multiplier by 0.5 or 1 after dropping the fsb by a few notches. Sometimes the multiplier will be more stable lower than stock with a much higher fsb. Experiment with different combinations and it wont be long till you gain a feel for the best combination of fsb and multiplier to hit that magic clock.

After the CPU limits are found, i normally lower the clock a fraction and start working on the ht, leaving nb frequency at stock. After that i go on to RAM but concentrate on tightening timings rather than max frequency (just what i prefer to do, performance wise is probably the same).

If you get 0124 black screens, then try lowering/increasing cpu/nb voltage between 1.15 and 1.3. This will solve it if the problem was a voltage issue with an unstable memory controller.

GL and keep us updated!
Click to expand...


Thanks I will give this a try tonight

What stress tests are you using and how long between stages

Thanks
 
My max Gflops with IBT was at 302Mhz bus x 16 with ram at 2012Mhz. That was on an ASUS M5A99x Evo board

I normally run 250Mhz x 19 as a day to day OC which gives me 2000Mhz ram and 2500Mhz HT and NB clocks. +0.1V offset on CPU

I would not run lots of stress testing, maybe a 5 minute prime or a few IBT runs but use the PC normally, game or encode or whatever and see if it falls over, if not you are probably good
 
Sorry guys still haven't had a chance to try my overclock using your suggestions.

I will get round to it asap
 
Just a quick update for you all. I have not given up on the CPU overclock;)

But I have just received my 16GB of Team Group Vulcan Red 1600Mhz so I am busy overclocking this

It's XMP is 9,9,9,[email protected] 1600Mhz

I currently have it clocked at 2133Mhz 11,11,11,[email protected] and still in progress:D

With a set of 8GB dual I had it running at 9,10,10,[email protected] 2133Mhz but with the extra set of 8GB it's taking loser timings and a bit more voltage to achieve good overclocks but you can't complain with the current progress I am receiving:)
 
What are you running clock wise day to day then?

4.8 seems to be the sweet spot. I had to RMA my PSU and since then have reset to defaults and updated to latest BIOS, reminds me to get back at overclocking it.
 
I know the Max voltage people recommend on these chips is 1.55v but isn't the FX9590 just a specially binned FX8350 therefor can they not accept 1.65v also
 
DMF1986 said:
I know the Max voltage people recommend on these chips is 1.55v but isn't the FX9590 just a specially binned FX8350 therefor can they not accept 1.65v also
Click to expand...

You shouldn't be putting 1.65v in either the FX83 or the FX95 (Not that anyone should buy the FX95 like ever)

Back when I had a Crosshair IV Formula and a golden 1055T, best results were "FSB" (FSB doesn't exist any more and hasn't for half a decade, but for arguments sake I'll use the term) and multi, abusing the fact I could do 350 "FSB" meant lower multi which resulted in better clocks (Stable that is)
 
Martini1991 said:
You shouldn't be putting 1.65v in either the FX83 or the FX95 (Not that anyone should buy the FX95 like ever)

Back when I had a Crosshair IV Formula and a golden 1055T, best results were "FSB" (FSB doesn't exist any more and hasn't for half a decade, but for arguments sake I'll use the term) and multi, abusing the fact I could do 350 "FSB" meant lower multi which resulted in better clocks (Stable that is)
Click to expand...

Overclockers website clearly states normal operating voltage is up to 1.65v
 
DMF1986 said:
And there is the intel fanboys
Click to expand...

You've got to be kidding seriously?!

DMF1986 said:
Overclockers website clearly states normal operating voltage is up to 1.65v
Click to expand...

So it does, but that needs challenging, 1.65V is nigh on suicide, like even 1.55V (Even though I've ran such voltages on 45nm) I wouldn't really like constantly at 32nm, but 1.65v is simply not safe 24/7.
 
Martini1991 said:
You've got to be kidding seriously?!

It's my mate who made the comment get over yourself

So it does, but that needs challenging, 1.65V is nigh on suicide, like even 1.55V (Even though I've ran such voltages on 45nm) I wouldn't really like constantly at 32nm, but 1.65v is simply not safe 24/7.
Click to expand...

I can only go off the specs given
 
And they're not right, 1.65v is way too high Vcore, all I can think is it's a copy/paste error, 1.65v being maximum RAM voltage it can take (Official)
 
You must log in or register to reply here.
Back
Top Bottom