FX-6300 first time OC

[trents]

Leave you CPU multiplier alone for now. Use only the FSB to overclock for the time being until your memory frequency shows approximately 800 mhz in the CPU-z "Memory" tab.

Keep your HT Link at about 2400 during this process which may mean lowering that component's speed at a certain point.

Allow your CPU/NB to rise to about 2400 mhz. If it gets much higher than that, knock it down also.

Add to the FSB in 10 mhz increments. After each increase, run a 20 minute Prime95 blend test to check tentative stability. Repeat this until you fail the stress test and then add .025 to the CPU vcore. Yes, your temps are very good right now. When you get the memory frequency up to the 1600 mhz range then leave the FSB alone and start overclocking with the CPU multiplier again. When you fail the Prime95 stress test add another .025 increment of vcore. Don't let vcore exceed about 1.5.

Watch CPU and Package temps. Keep CPU temps from exceeding 70c and Package temps from exceeding 60c. Don't add anymore CPU voltage when you get to those temp levels.

 

[GroovyMotion]

Ok I found a great setting, stable 20min in prime and was also usign ff while doing the test!
-vcore: +0.125
-LLc: Ultra High
-NB: 2250
-RAM: 750
FSB: 225
HT: 2475
Here are my graphs

 

[trents]

With that motherboard you should be able to get that CPU to 4.4-4.5 ghz with no problem if your cooler can handle it. But it will mean adding more vcore which will drive temps up.

Look at your Memory frequency. It's only at 1500 mhz (749.9x2). You are under utilizing your RAM which is rated for 1600 mhz. You need to start increasing the FSB as I pointed out earlier. The FX CPUs can handle 1866 memory easily and if you aren't even getting to 1600 on your RAM your are cheating yourself.

 

[GroovyMotion]

ok I got it stable @230 with vcore +.0150
here are my graphs...looks good!!
Should I bump my vcore some more since it's stable or just start increasing the multi now?

 

[trents]

Keep going. Lower your HT Link. It's getting too high.

Also, disable a setting in bios called something like APM (Advanced Power Management). It's part of the "green" cluster of bios controls along with Turbo, Cool N Quiet and C1E. They should all be disabled.

 

[Priller]

Keep going. Lower your HT Link. It's getting too high.

Also, disable a setting in bios called something like APM (Advanced Power Management). It's part of the "green" cluster of bios controls along with Turbo, Cool N Quiet and C1E. They should all be disabled.

His HT link speed is fine until around 2700-2800 or so.


OP: is your voltage staying steady at the 1.39X voltage under load?

 

[GroovyMotion]

Keep going. Lower your HT Link. It's getting too high.

Also, disable a setting in bios called something like APM (Advanced Power Management). It's part of the "green" cluster of bios controls along with Turbo, Cool N Quiet and C1E. They should all be disabled.

Yes APM is disabled, I can only lower the HT link by 200mhz blocks so now it's reportinh @2300
Thanks

 

[GroovyMotion]

His HT link speed is fine until around 2700-2800 or so.
View attachment 137009

OP: is your voltage staying steady at the 1.39X voltage under load?

it varies between 1.308 and 1.404

 

[Priller]

it varies between 1.308 and 1.404

So after a few seconds of prime95 running and during its run. The voltage will go go from 1.308 to 1.404?

That is a large swing in voltage.

 

[trents]

it varies between 1.308 and 1.404

Under load, when your vcore hits 1.404 does it tend to stay there or does it vacillate between those two numbers during full load?

 

[GroovyMotion]

So after a few seconds of prime95 running and during its run. The voltage will go go from 1.308 to 1.404?

That is a large swing in voltage.

Exactly! And my vcore is set @ +0.150v from stock
I think it was not making such swings when the voltage was stock but I was not able to go @ 225 FSB at default v
Would increase the vcore a tad more help or not?

 

[trents]

HT Link is fine at 2300. We can always try it higher later but better to keep it on the low side until you get other overclocking parameters nailed down.

 

[trents]

Groovy, what is the make and model of your PSU?

 

[Priller]

Exactly! And my vcore is set @ +0.150v from stock
I think it was not making such swings when the voltage was stock but I was not able to go @ 225 FSB at default v

Wow, alright.

I would stop trying to OC at this point.

You need to mess with your LLC until that voltage under load is steady.

If you set 1.4 and you get 1.392 that is fine. If you set 1.40 and you get a slight increase but it remains stable under load and keep that in mind when you set voltage.

 

[GroovyMotion]

Wow, alright.

I would stop trying to OC at this point.

You need to mess with your LLC until that voltage under load is steady.

If you set 1.4 and you get 1.392 that is fine. If you set 1.40 and you get a slight increase but it remains stable under load and keep that in mind when you set voltage.

Ok will try. thx

 

[Johan45]

Wow, alright.

I would stop trying to OC at this point.

You need to mess with your LLC until that voltage under load is steady.

If you set 1.4 and you get 1.392 that is fine. If you set 1.40 and you get a slight increase but it remains stable under load and keep that in mind when you set voltage.

Yes I had told him that in post #6. Drop it from ultra to regular and see if that lowers the increase. You may have to raise your offset voltage to compensate. The .1v swing in itself can make things unstable.

 

[RGone]

What is most likely to occur when overclocking the 990FXA UD7, UD5 and UD3 boards by revision over the last 2 years is shown pretty accurately in the thread linked below.

http://www.overclock.net/t/1023100/official-gigabyte-ga-990fxa-series-owners-thread-club
[Official] Gigabyte GA-990FXA-Series Owners Thread/Club

The thread linked above begins before the actual release of the GA-990FXA motherboards. The thread chronicles the Rev 1.0 boards. Chronicles the Rev 1.1 boards. Chronicles the Rev 3.0 boards and speaks in the last 20 or so pages about the Rev 4.0 of those type boards by Gigabyte.

I think you have to get to at least page 15 of 827 pages to see a user that actually has a released UD7, UD5 or UD3 990FXA motherboard.

It shows that the Rev 1.0 boards had NO CPU_LLC in the bios at all.

It shows that the Rev 1.1 boards did pretty fair. The Rev 1.1 was the revision that over a year ago, showed up in the AMD CPU Forum section here mainly. Tough to fine tune for overclocked stability but somewhat doable.

Then some of those boards that appeared in Rev 1.2, and began to be almost too much of a pain to overclock. Oh yes there is always the exception to the general rule of thumb.

I forget just how many of the last pages of 827 and of 8266 posts that speak to the issues with the Rev 3.0 of those Gigabyte boards. There are copies of the replies from Gigabyte Tech Support to the various posters in that long thread. Not awe inspiring for sure.

And finally toward the end of that monster chronicling thread there is discussion of the Rev 4.0 of the 990FXA series of motherboards.

The very first picture shown in that thread shows a UD7 and the Rev number right there in the upper left of that picture. That UD7 was Rev 1.0. You can look on any of the Gigabyte motherboards and find the Revision number. The revision of motherboard the user has makes a huge difference in how the board will act when trying to overclock an FX processor.

Revision 1.1 and the most current Revision 4.0 are the two most likely candidates for best overclocking without huge amounts of Vdroop under load and no amount of CPU_LLC adjustment helped the situation for the large number of users posting in that long thread if they did not have the better revisions of the boards.

If you really want to know about the Gigabyte boards for AMD FX processors you can find a thread like that over at OCN about even the 970 UD3 boards. All of the threads are now over 800 pages long and tell the timeline story of the various revisions and why I just never could suggest the Gigabyte AMD boards for the last 19 months I have been following the actual users and their problems.

You really need to know what Revision of motherboard is in your computer case since the revision; will with at least 95% accuracy, determine just how your own overclocking journey will likely proceed. Could even save someone having hair go toward gray, once they know which way the overclock is most likely to go.
RGone...

 

[GroovyMotion]

Groovy, what is the make and model of your PSU?

Trents, it's a Corsair CX750M

 

[GroovyMotion]

What is most likely to occur when overclocking the 990FXA UD7, UD5 and UD3 boards by revision over the last 2 years is shown pretty accurately in the thread linked below.

http://www.overclock.net/t/1023100/official-gigabyte-ga-990fxa-series-owners-thread-club

You really need to know what Revision of motherboard is in your computer case since the revision; will with at least 95% accuracy, determine just how your own overclocking journey will likely proceed. Could even save someone having hair go toward gray, once they know which way the overclock is most likely to go.
RGone...

RGone, it's a revision 3, brand new bought a few days ago

 

[GroovyMotion]

Yes I had told him that in post #6. Drop it from ultra to regular and see if that lowers the increase. You may have to raise your offset voltage to compensate. The .1v swing in itself can make things unstable.

I tried the default LLC setting, medium, high and ultra high...I will monitor the voltage much closer and report thx