• 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.
At 4.4g seems to be the sweet spot on my 6300.I run on air, so water might be much different.Knowing my core temp maxes out at 42C, on that clock speed is a plus.1.35 volts is another reason.Anything over that clock requires more voltage, produces more heat, for little return.What I mean is, will a few hundred MHz's make a dramatic difference ? In gaming terms a few fps will not make a difference.What will make a difference is the the longevity of the cpu at very high clocks. I'm guessing 4.6 and beyond just from a voltage standpoint.From everything I read about fx cpu's, is it's power management on load.Its double compared to a Intel cpu.That along with increased voltage can not be good for a 24/7 oc.

My feeling is that AMD sold the private consumer a bill of goods like politicians do. Many of us want to support the underdog. Well good until it just hurts. AMD released the Bulldozer/Zambezi FX and then a year later released the PileDriver/Vishera line of FX processors that is what they should have released that year earlier. So we took a years beating on the early FX and then on top of that AMD dried all the 'good' unlocked Thubans up from retail. Did it in less that a few months. Now if you thought you wanted new perfomance you had to get a HOT azz Bulldozer and push the Pee out of it to approach the gaming ability of the Intel of the time.

M$ software does not schedule the cores on the FX correctly. AMD has odd registers turned on and others off as evidenced by the SuperPI patch floating around for X87 software.

Then after you consider every bit of that the average consumer has no clue but what he reads and it took ages for the hype sites to begin to tell the story in full. IMO. If you have a six core cpu at 4.4Ghz it is not going to game even close to 10% better with 300Mhz added is my actual suspicion. The video benches don't show 300Mhz on a six core FX giving a full 10% increase in gaming video testing. But you can be sure that the HEAT factor will go up by over 10% for that extra 300Mhz. It will vary from system to system but the extra 300Mhz will not quite need an added 10% extra in Vcore. Likely on the order of 6 to 8%. But the heat will rise considerably for that other 300Mhz.

Now if you get away from gaming and utilize software that is core aware, then I find my FX-8350 does more work while video edtting at 4.8Ghz than it does at 4.3Ghz. On the order of about 16 to 18% more work done with my video editting in the same time frame .Since I have COOL, it is worth it to me to run 4.8Ghz for the hours per month that I do video editting. Well worth it in my mind. YMMV.

I will add that after having gone thru 3 FX processors an FX-8120, FX-6300 and now having only the FX-8350 and spent enough time with it, I run with Vcore offset; windows performance mode in a modified balanced mode and ALL the green on in my bios. This gives me 1600Mhz idle at <1.0Vcore and 4.5GHz at 1.4Vcore when I need it and it varies very very little on my CHV mobo. It took me some work at the keyboard getting CPU_LLC to work with Offset Vcore to give enough Vcore at idle and the immediate jump to enough Vcore for 4.5Ghz, but heck that is what they make an adjustable bios for.

So if one is running 4.4Ghz with 6 or 8 cores, they are likely right in the very sweet spot of performance for >85% of the users.
RGone...
 
Bought this cooler at least 9 months ago...

.
...and have been discussing or cussing with some of the guys about installing it just to see what it would do. Cheaper than the big dog Noctua and rated just above the Noctua big dog at FrostyTech. So I bought it then bought an additonal super clamping force hold-down kit extra a few weeks later.

At the time of the original purchase I looked for and found an equivalent Xigmatek fan for the pull fan since I never intended to run push only but to run push-pull since I have an 8 core FX-8350 that I do video editting at 4.8Ghz with my jam-up water loop. I expected to need every bit of air-cooling that I could get to even get close to 4.8Ghz on air with a couple of hours of P95 Blend thrown in for stabiility testing. Good, no make that VERY good cooling is needed to run 4.8Ghz stable for extended periods of use. Well a d*mn good board as well and my CHV has been just dandy for my crazy usage.

This cooler is 6.5 inches from the board to its' top. I can mount the fans as close to the top of the fins as possible and can still get my G Skill Ares ram in under the fan out over the first memory slot. Neat-O. The reason I bought low-profile ram. Even more so since I have 4 sticks of memory in the mobo. Even if they are 1 8gig kit of DDR3-1600 overclocked to 1890ish and one 8gig kit of DDR3-1866 just over stock. 16gigs works better for my video editting and using software to reduce the size of Wn 7 intalls. One day I may buy a 16 gig kit of DDR3-1866 but for now I do not need it to get by since I tweaked the ram to run with the mis-matched kits.

Okay I took some pics of the cooler before installing and then I have a couple of pics of some OCCT P95 Blend mode runs while I was testing for stability. Stability good enough not to dork my videos while editting. I think I succeeded even if only just barely. If it were not for the CHV and ability to dial in just the Offset Vcore I needed, I would not have been as happy at all with the 4.8Ghz temps. I can live with them until I can re-examine the cpu to cooler contact patch appearance.

Just an add; don't fool oneself into thinking the fans can be slowed when pushing +1.5V thru the cpu for P95 Blend testing. Don't make me laugh. FX processors run HOT. Other than a check out of the cooler to top of the CPU, I have done a little tweaking the last 1.5 days and have it where it idles at 32c socket and 14c Core/Package at 1400Mhz and 0.980 Vcore and will jump to 4.8Ghz at 1.5Vcore and run fine as can be expected for that speed on air. Offset Vcore and Balanced mode in Win 7 are now my buddies.

I think that we can all easily see why VERY good water is best for fast running 6 and 8 core FX processors but it is good to know that for real I could drop to 4.6Ghz and run my video editting as long as I wished to do so.

It has been fun and challenging to have to RE-establish my voltages and even my CPU_NB speeds due to the greater heat of air cooling. The warmer temps of this air cooling vs my water loop caused me to have up my Vcore 0.02V for each of the speeds I ran OCCT P95 loop. It is not so easy either to run the 2600Mhz CPU_NB and HT speeds I was running with my water setup. I had to drop down to < 2400Mhz on CPU_NB and HT speed to get stable with air-cooling. Might be able to raise them again after much more testing and tweaking, but my gut says nope ain't gonna happen.

It as said was a fun exercise. Settled again for me the reason I went for the better CHV board out of the gate. Settled for sure that going with a jam-up water loop out of the box was my best choice as well. I mean I am not sure with this air if I could even bench at greater than 5.0Ghz as I have done with my water loop. I may try some plain bench runs just to see if over 5.0Ghz is benchable. Not sure. I hate the added Vcore. This is a rig I can use 24/7 at 4.6hz and not be concerned....but who knows how high it might go. Hehehe.
RGone...ster. :bang head:bump:

I tried to text the pics for clarity.

1.jpg

2.jpg

3.jpg

4.jpg

5.jpg

And now some stability testing since the job is not done until gettng stable is won.

1Air 4.6Ghz 2hr OCCT.jpg

2Air 4.8Ghz 2hr OCCT.jpg
 
Last edited:
Surprised by the Temps as well, i would guest they could have been a lot higher than that!!

Must of been a cool day in Jax? AJ.
 
U 2 guys are surprised temps that low and I am slightly out of joint they are that high.

I can stand the temps, but the truth is that I will likely just drop my speed to 4.6Ghz since it makes me more warm and fuzzy at that speed. I need to use the rig for some work and not really ready to pull the board just now and again. Hehehe.

I have a slightly modded 3705A block that I want to suss out, so throw it and another rad I have not tried in awhlle back into the mix and go there for a test and trial thingie. Not sure yet. We shall just have to see what pops into my head next.
RGone...ster.
 
I hope since I started this thread I may be able to Edit longer...

........................................................................
attachment.php
................................................................

Warning Will Robison, this is just some testing and speculation so far. It does seem to hold some value based on what we see as FX-8350 speeds by those we know well and trust and have a P-State read out given to us. If the data becomes greater, it might be a tool to have shown before trying to overclock an FX-8350 to use as a pointer to 'needed' Vcore and how fast the FX thing might fly on relatively normal cooling. Who knows? It is testing after-all.

Okay here goes. Everyone wanted a rank and I have done the best I can from some known data and from the P-States read-out posted. While I would not consider this to be a "scientific" exercise in the real sense of that word, it does seem to lend some information as to why some of us seem to run pretty good water cooled cpu speeds.

See my description for rank #4 below. I noted some oddities.

Some FX-8350 P-States and general ranking based on normal cpu speeds for daily use.

#1. > Bassnut >> Says his normal cpu speed is 4.9Ghz on water. No doubt that 4.9Ghz can get some real work done and not be a hinderance in gaming.
P-State FID 0x1A - VID 0x0A - IDD 13 (21.00x - 1.425 V)
P-State FID 0x19 - VID 0x0B - IDD 13 (20.50x - 1.412 V)
P-State FID 0x18 - VID 0x11 - IDD 12 (20.00x - 1.337 V)
P-State FID 0x12 - VID 0x19 - IDD 10 (17.00x - 1.237 V)
P-State FID 0xC - VID 0x21 - IDD 8 (14.00x - 1.137 V)
P-State FID 0x5 - VID 0x2B - IDD 5 (10.50x - 1.012 V)
P-State FID 0x10C - VID 0x35 - IDD 4 (7.00x - 0.887 V)

#2. > mR. Clean >> Shows to run daily at 4.7Ghz. Nice Freeken CPU speed here for daily use. Nice.
P-State FID 0x1A - VID 0x0A - IDD 13 (21.00x - 1.425 V)
P-State FID 0x19 - VID 0x0C - IDD 13 (20.50x - 1.400 V)
P-State FID 0x18 - VID 0x12 - IDD 12 (20.00x - 1.325 V)
P-State FID 0x12 - VID 0x1A - IDD 10 (17.00x - 1.225 V)
P-State FID 0xC - VID 0x22 - IDD 8 (14.00x - 1.125 V)
P-State FID 0x5 - VID 0x2C - IDD 5 (10.50x - 1.000 V)
P-State FID 0x10C - VID 0x36 - IDD 4 (7.00x - 0.875 V)

Posted by poco242 April 27, 2014
Rgone requested this. My P states-
See this thread for his chronicled OC journey.
http://www.overclockers.com/forums/showthread.php?t=745793

P-State FID 0x1A - VID 0x0A - IDD 13 (21.00x - 1.425 V)
P-State FID 0x19 - VID 0x0C - IDD 13 (20.50x - 1.400 V)
P-State FID 0x18 - VID 0x12 - IDD 12 (20.00x - 1.325 V)
P-State FID 0x12 - VID 0x1A - IDD 10 (17.00x - 1.225 V)
P-State FID 0xC - VID 0x22 - IDD 8 (14.00x - 1.125 V)
P-State FID 0x5 - VID 0x2C - IDD 5 (10.50x - 1.000 V)
P-State FID 0x10C - VID 0x36 - IDD 4 (7.00x - 0.875 V)

#3. > Johan45 >> Shows to run daily at 4.6Ghz but is in third place due to his very fast water cooled Cpu Validation of 5.7Ghz. Congrats man.
P-State FID 0x1A - VID 0x0A - IDD 13 (21.00x - 1.425 V)
P-State FID 0x19 - VID 0x0B - IDD 13 (20.50x - 1.412 V)
P-State FID 0x18 - VID 0x0E - IDD 12 (20.00x - 1.375 V)
P-State FID 0x12 - VID 0x16 - IDD 10 (17.00x - 1.275 V)
P-State FID 0xC - VID 0x1F - IDD 8 (14.00x - 1.162 V)
P-State FID 0x5 - VID 0x29 - IDD 5 (10.50x - 1.037 V)
P-State FID 0x10C - VID 0x33 - IDD 4 (7.00x - 0.912 V)

#4. > caddi daddi >> Does huge mountains of work on his FX-8350 rig where days and days are run with the CPU @ 100% load and
if I remember correctly he does so @ greater than 4.5Ghz most of the time. His cpu P-states read out indicates the SAME as my own FX-8350 which is odd to me since I have run his FX-8350 and my CHV was assigning a default Vcore of 1.392 to his FX-8350 and to my own FX-8350 the same board and bios a default Vcore of 1.375 was given. In addition his FX-8350 gave me fits in pushing it over 5.1Ghz compared to my own FX-8350. I knew there would be 'flyers' but not in this small of a sample.
P-State FID 0x1A - VID 0x0A - IDD 13 (21.00x - 1.425 V)
P-State FID 0x19 - VID 0x0B - IDD 13 (20.50x - 1.412 V)
P-State FID 0x18 - VID 0x0F - IDD 12 (20.00x - 1.362 V)
P-State FID 0x12 - VID 0x17 - IDD 10 (17.00x - 1.262 V)
P-State FID 0xC - VID 0x20 - IDD 8 (14.00x - 1.150 V)
P-State FID 0x5 - VID 0x2A - IDD 5 (10.50x - 1.025 V)
P-State FID 0x10C - VID 0x34 - IDD 4 (7.00x - 0.900 V)

#5. > RGone... >> Does all his video editting for hours at 4.8Ghz. Better P-States shown than for Johan45 but man that 5.7Ghz Cpu Validation is awesome by Johan45.
P-State FID 0x1A - VID 0x0A - IDD 13 (21.00x - 1.425 V)
P-State FID 0x19 - VID 0x0B - IDD 13 (20.50x - 1.412 V)
P-State FID 0x18 - VID 0x0F - IDD 12 (20.00x - 1.362 V)
P-State FID 0x12 - VID 0x17 - IDD 10 (17.00x - 1.262 V)
P-State FID 0xC - VID 0x20 - IDD 8 (14.00x - 1.150 V)
P-State FID 0x5 - VID 0x2A - IDD 5 (10.50x - 1.025 V)
P-State FID 0x10C - VID 0x34 - IDD 4 (7.00x - 0.900 V)

#6. > tungureanu >> Currently holding down the #6 spot with 4.3Ghz daily due to the strength of his water cooling system most likely since he has some awesome appearing P-States listed.
P-State FID 0x1A - VID 0x0A - IDD 13 (21.00x - 1.425 V)
P-State FID 0x19 - VID 0x0B - IDD 13 (20.50x - 1.412 V)
P-State FID 0x18 - VID 0x11 - IDD 12 (20.00x - 1.337 V)
P-State FID 0x12 - VID 0x19 - IDD 10 (17.00x - 1.237 V)
P-State FID 0xC - VID 0x21 - IDD 8 (14.00x - 1.137 V)
P-State FID 0x5 - VID 0x2B - IDD 5 (10.50x - 1.012 V)
P-State FID 0x10C - VID 0x35 - IDD 4 (7.00x - 0.887 V)
 
Last edited:
Nice work RGone. It certainly does seem to form a pattern.
 
Nice work RGone. It certainly does seem to form a pattern.

It does seem to be a pattern for sure. Is it 100%? I doubt anything like this to be 100%, but if I were buying used, I certainly would use such information as a prognosticator over having no clue.
Bobert...
 
It's really the only thing we have at this point to even make a half arsed edumacated gues at a FX CPU's possible performance.
 
Too bad we cannot find the Vid/P-States before we purchase. Speaking of them, Johan where's the P-States on the 9370?
 
His cpu P-states read out indicates the SAME as my own FX-8350 which is odd to me since I have run his FX-8350 and my CHV was assigning a default Vcore of 1.392 to his FX-8350 and to my own FX-8350 the same board and bios a default Vcore of 1.375 was given. In addition his FX-8350 gave me fits in pushing it over 5.1Ghz compared to my own FX-8350. I knew there would be 'flyers' but not in this small of a sample.

I think the IMC plays a big role in this as well. Just a hunch but it really is what holds this chip together. So it may work fine at " normal" clocks but start to "stutter" once the speed increases.
 
I think the IMC plays a big role in this as well. Just a hunch but it really is what holds this chip together. So it may work fine at " normal" clocks but start to "stutter" once the speed increases.

Well this was an exercise in guesstimation anyway. We were all looking for an "edge" as it were. Frankly I do see a pattern. One I would certainly use if buying used and seller gave the P-State for the FX-8350 cpu.

Would my purchase be guaranteed? No but if I could buy an FX-8350 with a substantially better P-State read-out, then I would feel pretty confident that my purchase of a used FX-8350 would equal and more than likely better my current cpu. And this maybe as far as the exercise can function as a "predictor".
Bobert...
 
Back