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

Testing does FX-83xx seem to flat line after 4.3GHz

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
To summarize the info RGone, myself, and a few others have collected in the last few months. Zambezi(Bulldozer) suffers from clock scaling decline starting around 3.8Ghz and essentially becomes pointless beyond 4.3Ghz. Vishera(Piledriver) does not exhibit the same effect.

Attempting to go past 4.3-4.4Ghz on either will require a "creative cooling" solution.
 
Too many pictures to look at. lol

So there's little point in going over 4.2-4.3GHz?

I cut the pictures by 80%. Put the majority in Spoiler.

If you look at the graph for VIshera processor as below, I think you will see the percentage increase is up and down by each 100Mhz increase in cpu speed. If I had good motherboard; good case cooling and a good CPU Cooler; I would hit just as close to dead-on 4.2Ghz or 4.4Ghz or 4.6Ghz as I could.

[
attachment.php
 
Great work, the added spoilers make it much easier to move around the page!
 
One more thing that most do not know but I have compared extensively against anothers FX-4xxx series cpu at the same speeds and you can almost dead-on take one half those Cinebench CPU scores with the 8-core and "half" the number and that will be the score for the 4-core FX processor. That should mean that if you are looking to buy a 4-core you could easily see how the CPU performance will compare against the 8-core.

Maybe a few will gain some insight. It was the reason I took on the task to begin with.
RGone...

Now some more information has surfaced in talking with another forum member with more than one FX-6300 that are all able to be run at the same time because they are all in seperate motherboards. He has at least three FX-6300s that I am aware of.

Above I said cpu speed for cpu speed you could just about half my Cinebench scores and see what a 4 core processor would exhibit in that bench for performance.

Now after comparing the other member's FX-6300 Cinebench scores; we have found that you can take 75% of my Cinebench score and hit dead on what a six-core FX-6300 will score at that same speed. Pretty neat if you want to glimpse what an FX-6300 will approximate for cpu performance.

Thanks to Mr. Nameless. Hehehe.
RGone...
 
I have to give you a big :thup: on this post Rgone. That is a whole bunch of work you did and it definitely should be read by all overclockers, not just AMD people.
 
Very good bit of information.

Almost wondering if I should be going for 4.4ghz OC on my 8350.... Stability might be easier to find, less voltage, less heat and on your setup it's just before a drop!
 
Very good bit of information.

Almost wondering if I should be going for 4.4ghz OC on my 8350.... Stability might be easier to find, less voltage, less heat and on your setup it's just before a drop!

I went back to your thread and took a look at the last few pages and where "mandrake" asked you about THIS POST and how things were there stability wise.

I did see both the Cpu and Core temps bumping the limiter ever so lightly. Might be a good idea mostly temps wise to just hit 4.4Ghz and see if it is satisfying until you get big-water.
RGone...
 
I went back to your thread and took a look at the last few pages and where "mandrake" asked you about THIS POST and how things were there stability wise.

I did see both the Cpu and Core temps bumping the limiter ever so lightly. Might be a good idea mostly temps wise to just hit 4.4Ghz and see if it is satisfying until you get big-water.
RGone...

You've definitely done some leg work and I applaud your efforts! I'm hoping to just have a little free time to try and get a really good stable run here myself! I can't imagine the time you've invested in this project!
:comp:
 
You've definitely done some leg work and I applaud your efforts!

Just to let ya know this forum has the ability to recognize others for being awesome. When you see a post that you thought was really helpful or just full of greatness give them a thanks its on the bottom right corner of the post. :attn:


thanks.png
 
Nice work RGone. Seems to be pretty similar to what I found when testing this on my FX-8320. Have you discovered a "sweet spot" for the CPU_NB?
 
I personally find around 2400Mhz to be what I consider the "sweet spot" on CPU-NB. I dont have a bunch of records of my tests, but I have gone up to 3.2hz CPU-NB with my 8320 while running ~ DDR3 2200 speeds on my RAM, and the largest improvement was the jump up to 2400 from 2k(according to 3DMark Vantage, and 3DMark11). I definitely saw some gains in score from 2400 up through 3k but they were very small and required lots of CPU-NB voltage(on a gigabyte 990FXa-UD3) where 2400 only required a single bump in volts. 2600 brought another reasonable bump but it cost 2 more bumps in CPU-NB volts. Given alot of members here especially newer members are sporting motherboards without amazing voltage regulation, average PSUs, and entry level cooling going balls to the wall with CPU-NB doesn't really make sense. 2400 is the easiest to lock in at reliably and still get a good gain out of it.
 
Well then why dont you help us all out run a couple of benchmarks that are memory sensitive before the change and after, you can record the differences and help us all lead new members in the right direction.

3DMark Vantage and 3DMark11 CPU tests are pretty memory heavy, SuperPi is also very RAM dependent. Something like SiSoft Sandra memory suite will tell the tale quite well I htink.
 
Lest we forget...

...I doubt that the mainstream that pass through this forum section, that being AMD CPUs, will really investigate this thread. Most of what 'seems' to pass thru are just looking for numbers to punch in to go fast. I have been here nearly day in and day out now for 2 years after becoming un-employed in the consumer motherboard industry. So I have gathered a sense of the ebb and flow thru this forum section. Make that I sense what seems to be the prevailing ebb and flow thru this forum section.

Before you take what I sense to be right or wrong or needs to be categorized, let me say that is not really necessary. I am only saying what seems to be; not correct, good or bad but only what seems to be.

For most of what you see in threads like this one here, to make any sense, you almost have to have a grasp of how your own rig operates already. Coming into threads like this "cold", like most everything we look at, may or may not be visually stimulating. Cold hard numbrers and lots of them just don't do much for many of us.

Take this thread for instance by ssjwizard,
FX Bulldozer VS Piledriver, 8120 VS 8320, Gaming comparrison
', and dig thru it. He spent a lot of time over the course of data gathering and the data was so dramatic that most of us knew that struggling for more than 4.3Ghz on Bulldozer cores, was most likely a useless exercise. Another bit of information that I gathered was that a lowered price displayed to move the old tech Bulldozers was not a good reason to buy other than a Vishera based FX-series processor.

When ssjwizard began that thread, I decided n0t to post something similar but not exactly what he did but to only graph/chart, what I had been doing in the background for a few weeks with my FX-8120. Guys he had it nailed. No need to RE-invent the wheel.

I did not search for the list, but if you come across what ssjwizard is using his FX-8320 for and I can only remember just a few of the things that run CON-currently on his rig. I think they were server, at least 2 VM's, F_at_H, and well you get the idea I think. His rigs have something like work to do and not just plow thru a game. He has spent countless hours testing, tweaking and testing again before tweaking some more.

There are others in this forum section that do the same thing. They never say a thing about their rigs or they are not particularly vocal about what they do. They don't game much and they have tested into the guts of their rigs.

I know fairly personally one user that has an application so large that it was taking a four-core Deneb 5 days to provide him with an answer. Five days at WFO. He taught himself overclocking in less than 3 months and went to 8-core AMD to cut that time to get answer in that giant application to just a shade over 2 days. I just received a report that his FX-6300 was able to do the same giant application in just about 2.5 days or a shade less. Do what? 6-cores of Vishera right at the doorstep of 8-core Bulldozer? There seems a measurable difference between Bulldozer and Vishera for 'work'.

So when a long, results oriented thread is started, there are multiple forum members out there that have contributed input to most of these threads in one way or another. Others have helped develop the stategy. They may not even know they had input to the thread but something they said along the way may have helped in the overall processing of the thread, be it if only by support and atta-boys.

I will not name names but I am fairly certain that at least 6 nameless forum members have in one way or the other helped generate this thread. Those users do a tremendous amount of legwork in the background before they even voice a "help" to the one coming in wanting to know how to overclock X. I respect those "users". I glean something from them all and I never expect to see eye to eye with them They after all are individuals. What we seek to do is use our stuff fast and furious within logical reason. Hehehe. Thanks men.

Long-winded ole faht.
RGone..._________________:bang head :bump:
 
:salute: Good job there Rgone. I see you just can't leave that puppy alone.
 
3DMark Vantage and 3DMark11 CPU tests are pretty memory heavy
Thanks for clarifying this Wizard, I was comparing Rgones Vantage tests "CPU" against mine. I kind of figured that it was his combo of higher clocked memory and higher NB Frequency that caused his numbers to be higher clock for clock. Considering we're using the same motherboard.
 
Back