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

FEATURED Info & Tips >> Using the AMD FX Bulldozer/Piledriver...

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
Be very careful of that Johan character t1nm4n he is a sniper :snipe:

Shrimpy might debate you on that one. :p

Meh it's beat to death. Don't matter to me. I know what matters first and foremost for overclocking. Cooling first.

As sarcastically as possible...

I run my AMD chips naked. Oh big deal one might say. Big deal dude run 4-6-8 vrms.

De-lid.

IHS plate removed. gains mhz.

Cooling first 8 vrm second.

And it doesn't matter who's rodeo you wanna watch.

You can see it in the Intel threads...

Simply put, cool it now, vrm it later.

And apologies for the thread choppage Rgone. I know cooling doesn't apply much to how many vrms a guy uses within his price range.
 
Johan and tinman you are both right about how the efficiency of a component or circuit work just as RGONE in his explanation is as well.

From looking at the author chart it looks like he is comparing the 8350 to the new 9xXx chips at stock clocks. Now I am going on memory so feel free to correct me if wrong. What really surprised me was the difference between the 8350 at something like 120 - 140 watts compared to the 9590 which was well over 300 watts. This comparison all one should bring sense to what shrimpy has been trying to say as well as how many issues we have seen lately with said chips even on good boards. Maybe better said then my first statement on this thread this am.

You are exactly right on Bassnut. Since the default speed of an FX-8350 is 4.0GHz @ about 1.3875V and the default speed of an FX-9590 is 4.7GHz at close to 1.5V, it is not hard for the majority of us to extrapolate just how much power our own FX-8350s can consume when overclocked, with overvolting to maintain the overclock.

Truth is most of us have known and believed all along that our clocked up FX-8350s were just as power hungry and as big a heat monger as the FX-9590s. Because both the FX-9590 and the FX-8350 are built on the same architecture, it would stand to reason they are all birds of a feather when clocked the same. Probably within a margin of 1 to 2% since some silicon is just better than others and the reason we all search for the golden silicon.

Good call Bassnut.

RGone...

- - - Updated - - -

And apologies for the thread choppage Rgone. I know cooling doesn't apply much to how many vrms a guy uses within his price range.

Hale man you know me...earlier I posted about how much of the VRM was used in just taking care of the heat so, yes heat matters. Matters probably as much as the caibre of power supply and capability of the mobo. I am running my CHV-Z on the stock AMD FX-8350 cooler right now and there is NO WAY to overclock the sumbeetch as far as it has gone regularly stable. So COOL the cpu or go home.

RGone...
 
Last edited:
Right on R_Gone! For the record no sarcasm and a serious note, I've only claimed an ever run this one 4 vrm board. Never once said yea ALL 4 vrm boards are good for it. Just the one I bought and had a good time with.

the default speed of an FX-9590 is 4.7GHz at close to 1.5V

I would stand to correct this, but it's neither false nor true.

At load stock turbo on... Full Prime95 load or other stress testing and gaming, The 9590 will usually sit between 1.416v and 1.4560v and occasionally grab all the way up to 1.5250v Cpu VID.

Also sometimes requests 1.5000v, sometimes 1.5125v.

But the processor and how it's designed is quite smart knowing exactly WHEN to get and pull it's own voltage as needed. I find it interesting to watch it jump way up from 1.416v up to 1.5250v never a stitch more.

If I run an overclock on auto cpu voltage - lets say 4900mhz with power savings (green) on, its pretty darn stable aside from the low volt hang. Reason behind this occurrence is that the power demand at 4.9ghz is a little more, not a lot really than 4.7ghz But enough to impact low volt hangs. Is perfectly fine at 1.5250v and running a chiller get the cpu to 5ghz. HellCat, that ain't no problem. Cpu loves the lower temps.

Could I replicate this on FX-8350? Why certainly. Cruising some old pictures, both 8320 (de-lidded) and 8350 did about 5ghz around 1.5000v give or take depending on temps on simple tap to drain liquid cooling.

So in theory, Running the FX-9590 at 5ghz with less than 1.5000v is possible given Cpu temps are way down. Not frozen, just not really over 40-45c on load.

And have run TEC. You guys might have read the thread at CP. It was a complete gas brother.

Here's a simple example. 4 vrms TEC only 2 cores, plenty of mhz - screen shots of higher than this actually, this is FX-8350 5300mhz 1.440v.

Please note the solid 1.440v. No fluctuation.
DSC08765.JPG
 

Attachments

  • FX-8350 2 cores X 5.3ghz 1.440v .png
    FX-8350 2 cores X 5.3ghz 1.440v .png
    84.9 KB · Views: 479
Here's the FX-4300 (de-lidded) is sent ya Rgonesta! Very similar ti the Big Fx-8xXx and 9xXx series chips. (you'd like this one, clocks like mad)
 

Attachments

  • FX-4300 5300mhz II.png
    FX-4300 5300mhz II.png
    73.7 KB · Views: 509
Search and Update...

...Remember that seemingly in computering there is the average of things and then the fliers that defy explanation.

Been out looking for some things that speak to some results have seen around and about lately. Like fluctuating FSB away from what is set in bios by 5MHz. Up front would think this is not a good thing since the CHV board was not supposed to do it or in other words was fixed when the original CHV came out.

From review in our own Forum.

One very important point to mention is that the Crosshair V does not suffer from bus clock fluctuation that plagued the Crosshair IV. After staring at CPUz for five minutes under load, I saw literally 0.0MHz bus speed deviation. HT freq deviated max 0.3MHz and CPU-NB freq deviated max 0.2MHz. I’d call that a vast improvement over the 890FX fluctuations of the past. Those fluctuations didn’t much affect most people other than to exist. However, those that pushed the Crosshair IV under extreme cooling at the bleeding edge of stability were hampered by their system crashing due to too much bus fluctuation.


I used DFI AM2+ motherboards while working for them and the first board I bought after DFI went non-DIY was an Asus CHV non-Z mobo the first release. So my first experience with Asus was a board that supposedly fixed the fluctuating HT Freq.

NO I do not see my CHV-Z board with a fluctuating HT Ref Freq. I set 205FSB...I get 205 FSB with that 0.07 add that Asus has had for years. I am pretty close to believing that boards that have a FSB that jumps around has a problem. I get that way more as I do not see my CHV-Z and non-Z that do not do it with my FX-8350s.

Alright moving on. Been working with a user in our forums and seeing things that on the surface seem not to add up exactly. SO I went back into "let's go see" mode. Came across some things that while we may have spoken of or about some of these before it might be time to say something again. REmember the fliers.

Our own Johan45:
To run 2400MHz ram you'll need the CPU_NB at 2400+. It's a lot easier to OC the CPU with a ram speed around the 2000 MHz mark or lower. AMD FX processors don't benefit a whole lot from high speed ram and still prefer lower speeds with tight timings. Having high speed ram is just harder on the IMC. Now Johan may have had to change his stance and go thru more ram and cpus after joining the benchmarking team where all previous bets are off. RGone...

We are not the only ones seeing a difference in FX-8cores that are now in use since early 2014. As seen around forums on the net. What about the FX-8350s pre FX9370 or FX-9590? The 8350's no longer seem to have superior or golden chips among their number. The great chips have all been binned into FX-9370s and FX-9590's. FX-8350s now produced seem seldom to reach 5.0 GHZ. REmember the fliers.

Now how many times have we begun to see this?
Give the 8320 a try, its got 8 threads and most will get you to 4.7?

Not so true any longer it begins to appear. The current 8320 production usually tops out at about 4.5 to 4.6 GHZ. Have an FX-9590. I can not get stability even with 1.60 volts. It has an unusually high VID of 1.488 volts at 4.7 GHZ stock frequency. It will run games but Prime '95 locks up after running a few seconds. I could try 1.65 volts but h*ll I would never run it 24 7 at such a high voltage so I am settling for 4.85 GHZ at 1.58 volts. So NoT even all FX9590's give a stability at reasonable voltage at 5.0 GHZ and above. Humh. Heard similar more and more as FX-9590s begin to come around OCF more.

Oh the freeken ram thing again as mentioned already by Johan above.
I run 16gb of 2400 trident 4x4 dimms, I can get 2 sticks, 8gb stable at 2133 or a little higher, but for 100% stable it's back to 1866 for all sticks running totaling 16gb. No matter core or cpu-nb volts, just stressed the IMC too far. Using initial batch FX-8350.

Ram speed > 1866, 4 dimms, 8-9-9-24 Ram 1.65v & 1.275V CPU_nb Those were pretty good sticks!!!

Ram speed > 2133, 2 dimms, 8-9-9-27 Ram 1.7v & 1.3V CPU_nb

Ram speed > 2400, 2 dimms, 9-10-(9 or 10)-27 Ram 1.7V & 1.3V CPU_nb
The above may be a general "tale of the tape"about ram speed and amount of ram and elevated CPU_NB voltage as experienced by others.

In bringing the above into focus it seems trending that Pre-release of the FX-9xxx cpus the FX-8xxx cpus tended to overclock higher but with more voltage it seems and that now more often we see FX-8xxx cpus that will not run the bigger Cpu MHz. Dang roll of the dice when buying cpu silicon

As regards what I generally saw it appears one has to have a 'golden' IMC to run fast ram speeds with more than two sticks of ram. Still seems to be a consensus that AMD cpus respond more readily to tighter timings than faster ram. This is nothing that has not seemed apparent for years and more so if you look for 24/7 stability at elevated cpu MHz.

As time has gone on I try and ensure that this long sticky is still relevant and update as time goes on. I endeavored to cull down hours of reading into the real meat of the situation in posting this.

RGone...ster.
 
...

Our own Johan45:
To run 2400MHz ram you'll need the CPU_NB at 2400+. It's a lot easier to OC the CPU with a ram speed around the 2000 MHz mark or lower. AMD FX processors don't benefit a whole lot from high speed ram and still prefer lower speeds with tight timings. Having high speed ram is just harder on the IMC. Now Johan may have had to change his stance and go thru more ram and cpus after joining the benchmarking team where all previous bets are off. RGone...


As regards what I generally saw it appears one has to have a 'golden' IMC to run fast ram speeds with more than two sticks of ram. Still seems to be a consensus that AMD cpus respond more readily to tighter timings than faster ram. This is nothing that has not seemed apparent for years and more so if you look for 24/7 stability at elevated cpu MHz.


RGone...ster.

I still feel the same way, day to day you're not going to gain a whole lot from 1866 to 2400 just more headaches when looking for stability. When I test a new chip to find it's "qualities" if you will I use a set of G.Skill 1866 9-10-9 and leave the the NB at auto. Then I head to 5.0 and see how it goes. If I can make it without changing the CPU_NB settings I feel it's a "good" chip. Next is the CPU-nb speed 2400 at a 5.0 clock with 1866 I find is generally 1.25 to 1.28 for most of the testing I have done lately. Then you mention the "golden" IMC my first 9370 would run 3000-3100 NB @ 5.0 with 1.3v and never had any trouble with high speed ram BUT it did require more volts then. 90% of the time I bench at around 2200 CL8/9 with 2x2GB. Pushing too hard with the ram/nb lowers my max clock for the bench even cold and excessive CPU_NB volts won't fix it either.
 
Johan45 said:
Pushing too hard with the ram/nb lowers my max clock for the bench even cold and excessive CPU_NB volts won't fix it either.
Interesting observation "johan". One likely would do well to remember that. I see you still subscribe in practice with AMD does not really need or maybe care for extreme ram speed but still likes the tight ram timings.

Thanks man.

RGone...
 
Yes but there are exceptions and that comes down to the "golden" CPU. The one that actually works the way AMD wanted it to. I'm taking this pic from the bot just as an example. 3700 NB and 2600 ram. I can do those things with mine as well but I can't get it all to work together at a high rate of speed like this

image_id_1056019.png
 
Wow. Don't we all wish there were more of the "golden" cpu type. Heheh. I think I had a gold-plated cpu once but not the full on thru and thru golden type.

RGone...
 
It's the same with every pile chips. There are a few really good ones but a whole shett load of ok ones.
 
Do what?

I was advised that a number of links in this thread were no longer valid. I have gone thru this #151 Post and tested every link and RE-arranged as necessary.
3:15AM Nov 5, 2015. RGone...


...Took an early nap and so could not sleep tonight. Up doing some searching. Why search? I searched because it seems that AMD ZEN is not going to release until Q3 of 2016 and that is some year off still. Even if they were to move up the release date of the ZEN for desktop...probably have time for some fun still with a later production FX 8 core processor. Well maybe if I choose the best FX 8 core and my dice roll for silicon is golden. AND my water is cold enough. Hehehe.

The monies for the top four of the FX 8 core processors seems within $10.00 with each step up the processor scale right now at NewEgg, so not sure that monies should figure much into the purchase criteria at all. Or price seems not to hold much sway right now.

Worked up a list of all the Vishera processors as average ranked at HWBot. Interesting study for sure. I missed the really good deal on the FX-9370 with the included AiO cooler and a game for $209.00 Usd. It was for July 15, 2015 only. Now I know I have early FX-8350s from pretty close to their release date in Oct 2012. In fact one of them was ordered 3 days after the release of the Vishera FX processor series. Pretty good processor. The other FX-8350 I have was bought used and with a LOT of voltage will bench in the 5.2GHz plus neighborhood. But a lot of voltage is required. More than my other FX-8350.

Now beginning about Feb 2014 and the release of the FX-9590 and FX-9370 processors by AMD, the FX-8xxx series of processors have seemed to act somewhat different to what we could call the Pre-FX-9xxx series of processors. The, Post-FX-9xxx release, FX-8xxx processors seem to clock into the 4.7GHz-ish range or a little more, on a little less cpu voltage than the earlier FX-8xxx processors but often refuse to budge beyond 5.0GHz no matter the cpu voltage which was not the usual scenario with early FX-8xxx processors. Insert Edit here > it is also said my some that the later FX 8 core processors will handle somewhat faster ram speed but that seems little use with most AMD processors unless running with 'super-cold'. Tighter timings still seems to be the better/best situation for AMD today.

Additionally there "seem" a number of FX-8xxx processor that will not even do 4.8GHz with stability. Most prevalent in some later FX-8320 processors but even some FX-8350s just don't seem to have the same overclock potential as was the situation before the release of the FX-9xxx series by AMD.

The rankings below are "averages" for the processors according to HWBot. The overall submissions for the CHV Z and non-Z total over 5,000 submissions or more than nearly all the other motherboards combined. Almost. Add in the Sabertooth 990FX's and well...

FX-9590
http://hwbot.org/hardware/processor/fx_9590/ >> 856 Submissions
Air > 5093MHZ
Water > 5214MHZ
Cascade > 6111MHZ
LN2 > 6927MHZ

FX-9370
http://hwbot.org/hardware/processor/fx_9370/ >> 410 Submissions
Air > 5286MHZ
Water > 4503MHZ
Cascade > 4999MHZ
LN2 > 6983MHZ

FX-8370
http://hwbot.org/hardware/processor/fx_8370/ >> 275 Submissions
Air > 4969MHZ
Water > 4730MHZ
Cascade > 5266MHZ
LN2 > 7475MHZ

FX-8370E
http://hwbot.org/hardware/processors#key=fx_8370e >> 86 Submissions
Air > 5163MHz
Water > 4700MHz
Cascade > N/A
LN2 > 7692MHz

FX-8350
http://hwbot.org/hardware/processor/fx_8350 >> 3,771 Submissions
Air > 4714MHZ
Water > 4911MHZ
Cascade > 5886MHZ
LN2 > 7364MHZ

FX-8320
http://hwbot.org/hardware/processor/fx_8320/ >> 2,850 Submissions
Air > 4655MHZ
Water > 4846MHZ
Cascade > 5768MHZ
LN2 > 7237MHZ

FX-8320E
http://hwbot.org/hardware/processor/fx_8320e/ >> 233 Submissions
Air > 4776MHZ
Water > 4959MHZ
Cascade > 6095MHZ
LN2 > N/A

FX-8310
http://hwbot.org/hardware/processor/fx_8310/ >> NO Submissions
Air > N/A
Water > N/A
Cascade > N/A
LN2 > N/A

FX-8300
http://hwbot.org/hardware/processor/fx_8300/ >> 25 Submissions
Air > 4915MHZ
Water > 5130MHZ
Cascade > N/A
LN2 > 7085MHZ

So now am trying to tabulate what I see with the user experiences here in this forum as well as other forums to try and roll my dice on an FX 8 core processor. I want to try and do my best guesstimate of which processor to choose and then hope I do not roll snake eyes but instead roll golden silicon. Well...

RGone...skisterzzzzzzzzzzz:salute:...:bang head...:bump:
 
Last edited:
Good luck with your search and may your purchase be GOLDEN ....... don't bump your head too hard man .....:cheers:
 
I'm assuming the FX-9370 didn't have many submissions on water or cascade as the air has higher average clocks?

I'll also assume that she'll likely see a monster water loop for the majority of her life. If so, by you're averages theory, looks like you'd be going FX-9590. Factor in the fact that you already have a wealth of knowledge with the FX-8XXX and an insatiable curiosity I could totally see this going the FX-9XXX route.

At any rate I'm in to see what comes of this.

Best of luck with your roll Bobertzzzzzzzz!
 
RGone, you think too much into this.

You've had outstanding clocks any many FX processors. A little chilling proves to go a long way.

To go shopping for a "golden chip" without chilling hit some magical 6ghz on air..... well good luck as the boiz here exclaim.

OR

Perhaps I'm missing the point of your search? LN2 and WR you seek??
 
I guess honestly, I must be looking for 'golden' silicon since I would like to see 4.8/4.9 GHz on air inside of the computer case. So that mark is probably just a 'golden' hope as it were. Since such cpu is not going to be the same type that sets WR or works the best on super cold. I have no real desire for WR or LN2 stuffz. Have two cpus in the NegEgg basket waiting to see how my finger twitches. Hehehe.

RGone...skitsterzzzzzzzzzz
 
How about an idea of experience.

My highest most stable overclocks are on brand new processors. When used and witness to plenty of heat, this degrades significantly.

I'd do my search a little further out than NewEgg personally. Try Fry's and Tiger_D as well. If you can walk into these stores and actually see the processors top plate and get the info before purchase, that my friend would be the way to go shopping.
 
I guess honestly, I must be looking for 'golden' silicon since I would like to see 4.8/4.9 GHz on air inside of the computer case. So that mark is probably just a 'golden' hope as it were. Since such cpu is not going to be the same type that sets WR or works the best on super cold. I have no real desire for WR or LN2 stuffz. Have two cpus in the NegEgg basket waiting to see how my finger twitches. Hehehe.

RGone...skitsterzzzzzzzzzz
If looking for an Air OC it seems you might have luck with an 83xx that ends in 70 Rbobster!.
 
I guess honestly, I must be looking for 'golden' silicon since I would like to see 4.8/4.9 GHz on air inside of the computer case.

That sounds like THE golden chip you are looking for.
From what I have gathered I believe it is impossible to aircool FX8/9 chips effectively past certain power level. There is not enough heat transfer trough the IHS plate. The delta of the IHS and the silicon gets too high.
My experience running mild OC at around 4.4 to 4.6 is easy cooling wise but after than my temps start to skyrocket while my loop temps effectively stay the same. This is the point where aircooler tends to fail since the CPU contact of the cooler is warmer than on watercooling.
I might be horribly mistaken with my "theory" but I have found no other explanation yet. Perhaps my chip has bad solder between the silicon and IHS.
 
Back