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

[RGone]

C_D, it is a shame momma don't want that 18 fan aluminum rad thing in the house no more. Hehehe. I shorely don't want to be the one to push her while she recovering frum surgeries. Somebody might not recover. Hehehe.
Me is RGone...

 

[caddi daddi]

she is being a real trooper about it, I am amazed with her at times.
I am finding out that pile driver is not bulldozer.
bulldozer responds to brute force and pile driver must be tuned and tweaked.
That's not a bad thing as it will make me get a little better at this stuff.

 

[caddi daddi]

using Rgone 4.8 numbers gives me this.
this clock is 2 hours p95 stable on my rig.

 

[Mandrake4565]

Can't wait for the "trying to get 5.5" stable post like the ol 4.5 955 one.

 

[RGone]

Can't wait for the "trying to get 5.5" stable post like the ol 4.5 955 one.

Drake man don't be saying at stuff. He aleady said he might ship it to me to put under this MS water and that ain't necessary. I told him to take it back and get another one, but oddly he seems very pleased with the little dumping. Strange stuff since momma went in for surgery.
RGone...

 

[Mandrake4565]

Ha, don't you see the halo over my head gonester? I wouldn't try to instigate CD to do anything "5.5" "5.5" "5.5" "5.5"

 

[RGone]

Ha, don't you see the halo over my head gonester? I wouldn't try to instigate CD to do anything "5.5" "5.5" "5.5" "5.5"

Huh? Naw man I see this, I think. and those don't look like no halo. Hehehe. What is that number 5.5 .5.5 5.5? I don't what to forget when the sky starts falling.
RGone...ster.

 

[caddi daddi]

a little strange, and we all missed it.
I plugged in Rgones numbers and it worked like a charm, we don't see that very often.

 

[Mandrake4565]

5.5 is what cd will shoot for you can take that to the bank

 

[RGone]

a little strange, and we all missed it.
I plugged in Rgones numbers and it worked like a charm, we don't see that very often.

Missed what? What numbers of mine worked. Vcore?

I almost told you last night that you might be over-driving the cpu like I think I did on the Fatlady when I had to swap back to the CHV. You finally found a cpu that did not like to be over-volted? Dang.
RGone...

 

[caddi daddi]

aaaaaaaaaaaaaaaaa, would i do that???????

 

[caddi daddi]

Rgone, I plugged in all your 4.8 numbers including your ram timmings.

 

[RGone]

aaaaaaaaaaaaaaaaa, would i do that???????

Hear none. See none. Speak none...Hehehe.
RGone...

 

[Mandrake4565]

Rgone, I plugged in all your 4.8 numbers including your ram timmings.

Aren't those timings protected by the secret service?

 

[RGone]

Upload more than 3 pics at one time...

I need this somewhere that I can keep the link to so that this project can be more easily explained to others.
In order to UPLOAD any images, you must first be in ADVANCED MODE.


Clicking on the Paper Clip brings user to here.


Choose Three images to upload.


Do the UPload to forum.


This shows that at the Paper Clip location are TWO functions that can be done from that one two-sided button.


Another couple of images to edit.
RGone...

Okay I went and made some captures of what I did to get the first five images uploaded as above. Two more images to follow with text edits.

Pic of all files uploaded.


Pic of what it looks like to click on Down Arrow after all files are uploaded.

 

[storm-chaser]

thanks rgone, this is great material. I have a question (okay, maybe a few questions) that pertain to the original inspiration for this thread. Regarding the 8120/bulldozer core and its effective overclockability.

Sounds like you and others here have determined the 4.3ghz is a realistic oc limit imposed by diminishing gains and perhaps an optimal, safe core voltage. My question is, would you still see higher Cinebench scores with the processor at say, 4.4, 4.6 or 4.7 Ghz? Perhaps at the expense of excessive heat and/or an unhealthy vcore?

Have you been able to run Pi calculations with the 8120 ... or see how it trends above 4.3 with any other benchmarks?

and why is there such a dip and then jump in performance at ~5.3ghz on the 8350 chart? sorry if some of these have already been answered, I read the entire thread but couldn't find much other than a response on page 2 from ssjwizard.

 

[RGone]

Sounds like you and others here have determined the 4.3ghz is a realistic oc limit imposed by diminishing gains and perhaps an optimal, safe core voltage. My question is, would you still see higher Cinebench scores with the processor at say, 4.4, 4.6 or 4.7 Ghz? Perhaps at the expense of excessive heat and/or an unhealthy vcore? >=> Some increase as you continue upward in Mhz but not nearly the amount of increase as seen when moving up to 4.3Ghz. That is what makes it not really worth the real extra heat that caused many of us that do overclock 24/7 to have to resort to + $250 dollar water cooling to ward off the heat.

Have you been able to run Pi calculations with the 8120 ... or see how it trends above 4.3 with any other benchmarks? >=> No matter what benches anyone runs and posts; there always those wanting another. No I have not run others than some real work I did with the FX-8120 and now the FX-8120 is no longer mine. Got the 10 to 15% more powerful FX-8350 in its place.

and why is there such a dip and then jump in performance at ~5.3ghz on the 8350 chart? sorry if some of these have already been answered, I read the entire thread but couldn't find much other than a response on page 2 from ssjwizard. >=> There was an odd jump on both the FX-8120 (no longer have) and the FX-8350. A little less of an odd jump on the FX-8350 if I remember right. Have not looked at the graphs in quite some time.

I honestly have not debated that jump with anyone. It takes some really good water; even cold water, to run up there in the plus 5.0Ghz speed for any length of time. When a user hits his board with DICE or LN2 or Helium, they are not playing around at the just above 5.0Ghz speeds but are running for BIG numbers, so the oddness has not come up anywhere at all that I was able to find.

So given that it would take at least single stage phase cooling to try and run above 5.0Ghz all day everyday...I just have not pursued the oddity beyond 5.0Ghz. I use 4.0Ghz day in and day out and do all my Video Editting at 4.8Ghz which I can do easily with my water setup.
RGone...

 

[Woomack]

SS will be overloaded on these CPUs so even good water will be better for 24/7. Simply, stable results on SS will be about as good as on water as there are some other thermal steps to OC FX series.
FX8120@5GHz and full load will cause moisture near cpu socket so you will have to lower power draw to keep it below 0 or insulate all really good.
Better SS will work up to -55*C idle ( about -25*C idle on CPU ) and ~300W load what is not enough for fully loaded FX@5GHz+. Other thing is that good SS/Cascade will be about 2-3kW+ unit so expect that electricity bill will change

I could run FX8120 at 5.4-5.5GHz but only on 2 cores using 1.55V what under load was giving ~0*C. 5.1-5.2GHz was possible for longer multi core tests at 8 cores but I've never runned prime95 or something like that.
This was chip that was running stable @4.7GHz 1.325V on water.

 

[Mandrake4565]

Woomack,

This was chip that was running stable @4.7GHz 1.325V on water.

I don't know much about the 8120's but that seems like a really good Cpu V for that clock. I need 1.46 for 4.7 24/7 stable on my 8350.

 

[caddi daddi]

Aren't those timings protected by the secret service?

yes they are, good comrade Mr. Mandrake but I used your own carnivore(now called dhc1000) to glean the numbers from his disk drive.