SOLVED Finally had to give it a go...FX-8120 to 4.8Ghz

[RGone]

4.1Ghz on 1.31Vcore overall.

I have Vcore in bios set to 1.3175 which is not quite 1.32V but software rounds up...so be it. I think the 4.0Ghz run was with the bios Vcore set to 1.3862V so not much increase yet in Vcore to get 100Mhz more speed. Made this 100Mhz jump on FSB/HTT Frequency alone. Memory speed jumped as well.

What I am looking for is that Mhz vs Vcore wall that must alway be jumped along the way up the hill. Does not seem to be up against that wall just yet. May be at 4.2Ghz and will see since that is my next stop on Max Mhz with 1.34Vcore or less.

I am using the CPU LLC on 'high' so the voltage to cpu is not bumped up as it would be using 'ultra'. As long as high will do the job, can get by using less Vcore when loaded and thus have less heat overall.

Graphics below.

 

[RGone]

Same CHV as in previous posting above...Now Vishera FX-8350

This is not a CHV-z mobo but the original CHV Formula mobo, bought from the Egg 3 days before they started to carry the Z board. Did not return for swap. Now I really see no need to have done so for sure; since the FX-8350 in my cpu socket is running on the CHV bios 1102 from 12/27/2011.

I am seeing some odd to me performance increases, not expected by only 200Mhz increase in cpu speed. There seems a likelihood that the Vishera cpus are somewhat more robust than the earlier Bulldozer cpus. I will have to do some digging thru my old benches and reduce the speed of this FX-8350 to 4.8Ghz which was the sane/stable limit of my original FX-8120 even with my very good cooling.

Seems everyone is always whining and wanting someone's exact bios settings and not finding their own which is going to a must anyway due to system differances, so there is a stitched together shot of all my Extreme Tweaker settings except for the Digi LLC setting which is a notch down from the upper-most setting.

I am well pleased so far. Took about 20 mins to get to 5.0Ghz on this FX-8350 on this 'good' mobo. It took 20 mins or so because one has to start a bench and see how it acts just to dial in enough Vcore etc. Needless to say having dealt with an FX-8120 and being able to use the same bios I have used for months was a plus.

Screen captures are 'text' annotated for clarity. Bios stitch is the last image uploaded.

RGone...ster.

EDIT:
CrapOla. The bios capture is too dang small to read at all. I will remove it and work-over so it is readable. Crap crap.
END EDIT.

 

[RGone]

Bios Capture for benches shown in post above.

Maybe bios shot is now readable. I had to use a new 'stitch' tool and there is a learning curve.

Dang it the bios stuff is not readable. Back to the freeken drawing board. El CrapOla. RGone...ster.

EDIT:
Screw Stitching together for easy 1/2/3 reading. I uploaded the four individual screen shots. The shots overlap slightly. RGone...ster.
END EDIT.

 

[bassnut]

Nice OC RGonster ......

Yes you are right these puppies do have a bit more bite to them as I was able to shave off about 25 - 30% off my video encodes when I switched to the FX 8150
With all the talk about benchin ..... I dont think I have any benchin programs but I am planning to do a test with mine when I get a chance using my video encoding software for comparisons from stock up to 4.9MHZ on some BLU-RA encodes. Should be interesting to see how the stepsaffect my encode times ...... lord knows I ahve plenty of source material kicking around. I just have to find the time which lately has been almost noexistant. .

 

[RGone]

I moved from 1156 socket i860 @ 4.0Ghz to my previous FX-8120 and smooth halfed the time to do my video encoding to MP4 for facebook and family vidoes. Now I can test some with the FX-8350 and see what happens.

 

[RGone]

CPU_PLL aKa CPU VDDA Testing...

In 'ssjwizards' thread about upping CPU_PLL and getting a big drop in temps and need for less Vcore on his analog VRM driven Gigabyte mobo; some of us are testing on other boards. My board is the CHV non-Z version by Asus.

After beginning to do this and 'now' seeing that HWMonitor says it reads the VDDA, I find that my bios setting is then read by HWM as 0.03V higher than set in bios. So I have got to really pay attention to what is what. When I set over 2.6 in bios my board would error or freeze in Lin-X. I was testing what I thought was less than 2.6V VDDA but then I saw the difference in HWM indicated VDDA so I dropped completely below 2.5V VDDA and will now do some testing at a bios setting of 2.55V VDDA since the setting of over 2.6V causes problems with my Digi-VRM driven mobo.

I see above post of my bios captures that I was setting VDDA to 2.5187 and getting an indicated 2.55V so I am sure there is a difference in set and HWM indicated. Which is off I do not know but will adjust upwards to compensate and stay below 2.6V.

So I do not see any real benefit to me yet on CHV of VDDA adjusting. I do know set below 2.5V VDDA in bios does not cause any issues like setting above 2.6V VDDA in bios. Hehehe. But no real benefits to higher VDDA as is seen by some with older style analog VRM on Gigabyte UD3 mobo, I think it is.

RGone...ster. _________

 

[csimon]

RGone, thanks for this wealth of information. Any chance of finding out what you're Digi+ VRM settings are at the 5Ghz OC?
Also, what is your 24/7 cpu, cpu/nb, and dram frequency?