How to test just the turbo clock speed of an Overclocked AMD FX CPU?

CPU: AMD FX 6120 @ 4.739 GHz @ 1.476V on all 6 core.
Motherboard: Asus Sabertooth 990FX r2.0

My question is how can you test just the Turbo speed when it is set?

What I want to try doing is turning on the turbo mode in the Bios and upping just that since I can be stable at 4.838 but I don't want to run 1.572V for a 24/7 overclock. But I am 100% stable now and this frequency will only kick in when the thread count is low I want to know how to stress test the turbo frequency?

Can I use Prime95 and set it to test just 1 core and do the test for all 6 cores Or will this not work ?
If this wont work is the a better way to test the turbo frequency?
 
Solution
I would generally follow the temperature to voltage relation. As temperature drops, the electrons vibrate (in random motion) less, so you can safely give the electrons more directional energy (voltage).
You may give 50 extra millivolts every 10°C below 71.1°C/61.1°C. My CPU is at risk because I am above 50°C. I am relying on a safety factor here.
To be safe, like those Phase Change dudes running 1.65V-1.7V, you need,

stock 1.40V >90°C maximum
1.50V >71.1°C max
1.55V >61.1°C max (FX 9590)
1.60V >50°C max
1.65V >40°C max
1.70V >30°C max

You see how a 30°C chip at 1.7V is actually like a "burning" stock chip at 90°C. The transistors are actually under pressure. Voltage and heat makes electrons move faster.

At a certain point chips heat...
There is a turbo core frequency multiplier allowing for the adjustment of the turbo frequency. (I'm aware the my overclock is past all stock and standard turbo frequency's. Thanks though) I am hoping to boost the single core performance and keep my 4.739 Multicore frequency. Now whether this will work the way I want will remain to be seen But due to being so close to the edge of stability now I want to stress test the turbo frequency to ensure I remain stable.
 

kgt1182

Reputable
Jun 8, 2016
420
0
5,160
That is an average FX chip you got there. I also got 201*23.5= 4.72GHz at 1.49375V.

If temperatures allow you, you can try hitting 4.90 GHz at 1.60625V. (like my FX 4350) 10 trials of IBT with AVX at Standard will be sufficient to tell if your system is stable.

If you want ultimate performance with what you have right now, disable 2 cores down to 4 cores. Most games use 4 cores anyways. And I believe that the thermal headroom from disabling cores can net you an overclock past 5.0 GHz, given that Bulldozer is designed for higher clocks than its Piledriver successor.

Good luck. :)
 
What I am trying to do is better performance at the same voltage. I've hit 5.0 @ 1.67V Stable (2 hours of Prime and 20 runs of IBT on maximum) but that is just playing around and not for everyday use. Many of my games actually use all 6 threads that my CPU can handle. While point and click games, older COD games, Crysis 1 and many others I own don't use the extra cores the newer games that I do own do use them and would take a major hit in performance.

As for Bulldozer vs. Piledriver clocks, it has been my experience that the Piledriver architecture seems to get better clock speeds than Bulldozer.
 

kgt1182

Reputable
Jun 8, 2016
420
0
5,160
@bgunner
Does your IBT have AVX? Cause AVX does double the floating point operations at the same clock. (and also reveals instabilities within 1 trial, either it passes or fails miserably) My FX 4350 gets ~47 GFLOPs 4 threads with AVX.
But 20 hours IBT would be good enough.

How do you handle 1.67V with 6 cores? I can barely keep 1.60625V on my 4 cores at 4.9 GHz below 71.1°C.

And I really dont know if my chip is bad at overclocking or the reviewers are all golden chips and mine is average.

EDIT: I am keeping my 1.61V as a 24/7 overclock. You should aim for 4.9, 5.0+ hits a voltage wall on most FX CPUs. Then try to find the lowest stable voltage.
 
The reason I can run that voltage at that frequency is because I have a AIO liquid cooler and a high airflow case that helps keeps it cool.

The Version I'm using for IBT is 2.54. I don't run IBT for 20 hours just set it to run 20 runs with maximum amount of ram available. With my 6120 I have found that Prime 95 finds errors better than IBT.

I refuse to run that high of voltage for a 24/7 OC due to how much it will degrade the CPU. Running that high a voltage will case Voltage Degradation eventually causing the chip to become unstable at that voltage much much quicker than it should requiring more and more voltage to stay stable. For me to gain less than 300 MHz more I need to add 0.2V and the performance is just not there to justify that much of a jump. If you are hitting 71.1°C for the CPU temp then it is throttling OR the program you are using is not reading the temps correctly. I recommend using AMD Overdrive and watching the Thermal Margin temp to keep the CPU from throttling. Overclocking is a balance between Core Frequency, Voltage and Temperature not just how high you can get the cores unless your aiming for records breaking clocks. The FX and A series CPU's start to throttle at 62°C so over this the core frequency starts to lower causing a performance drop.
 

kgt1182

Reputable
Jun 8, 2016
420
0
5,160
Actually I have found out how some chips hit high clocks at lower voltages. You should check the delidding side of Intel chips, they delid, lose 10°C, and as a result of lower temperatures they can undervolt further.
It also explains how 8 GHz is obtained using "only" 1.9V; the sheer coldness of LN2. The heat increases the resistance of the circuit, so more voltage is needed to push the same amount of current through the CPU. At a point adding voltage make you need more voltage and the only way is to go colder when voltage does no good, that is why high overclocks on Phase Change is used for very high end setups.
Have you tried lapping your CPU? I lapped my FX 4350 CPU and lost 7°C from 75 to 68°C Which allowed me to turn the voltage down from 1.60625V to 1.58750V, which lowered my temps another 4°C to 64°C after 20 IBT AVX standard trials which allowed me to lower the VCore again to 1.58125V at 62°C stable.

That version of IBT is AVX. No wonder you have been using so much voltage for 4.9/5.0 GHz, like mine. That is the price for true stability, isnt it?
 
The inferior TIM on Intel's CPU's is placed between the core tops and the lid. All CPU's have TIM ( whether it be soldier or thermal paste ) between the core units and the lids to allow for thermal transfer so using it is a good thing and not bad. The issue was that the type of TIM used on those Intel Chips was crap and not done as well as it could have been.

The removing of the lids does not always allow for the voltage to be dropped temperature drops does not say you can drop the voltage and stay stable. The Chip that hit 8 GHz was a cream of the the crop chip so you can not use that as an average reference of any sort other than if you have a really good chip and cool with liquid helium you could hit this clock long enough to get a validation in CPU-Z.
 

kgt1182

Reputable
Jun 8, 2016
420
0
5,160
Yup mah fault; I thought TIM was thermal paste...
So you mean that those FX chips that hit 8 GHz @1.9V on LN2 were like those golden chips, except the most golden of the golden that can hit like 5.0 GHz on 1.25V on air? Unbelievable! I didnt know that the silicon lottery could vary so much until I searched it.

BTW- Should I try for 5.0 GHz? I am currently at 4.9 GHz @1.58125V, temps 62°C.
 
TIM = Thermal Interface Material. This means any material that helps conduct heat.

If the CPU is at 62°C then the CPU will start to throttle if you up the frequency or voltage. I personally keep a 10°C temperature buffer to allow for ambient temperature fluctuations. My recommendation is not to go f0or the higher clocks with out better cooling.
 

kgt1182

Reputable
Jun 8, 2016
420
0
5,160
Well, you could try enabling Turbo and increasing the voltage slightly. Your CPU would need less voltage at higher clocks if you have fewer cores active.
An FX 9590 runs at 1.53750V, 4.7 GHz with 8 cores, 5.0 GHz with 4 cores.
Safe SOI chips voltages range around 1.55V.

Looks like I got to settle with 4.9 GHz then, not worth getting a custom loop and much higher voltage just for "5 GHz club membership".

TBH you could set the 1 Core per CU option and become 3 cores, because each compute module have 2 cores sharing the FPU and cache, disabling one core every module can increase IPC slightly by around 10-15% and lower temperatures, allowing for lower voltages or higher overclocks.
3 cores is a good compromise for dual core games and 1 core left over for background tasks and music streaming.
 
I have enabled turbo core but it will not activate unless APM (Application Performance Management) is enabled causing the CPU to lower the frequency of all cores to 3.1 GHz when under load. This was far from what I wanted.

Is there a way to activate Turbo core without APM activated?
 

kgt1182

Reputable
Jun 8, 2016
420
0
5,160
Since APM and turbo core is designed for stock usages, it would be very difficult. The reason why your CPU is throttled in clocks is because APM limits your TDP to 95W regardless of CPU voltage.
I suggest that you try overclocking a little further on your CPU to 4.8 GHz @1.5-1.525V and call it a maximum 24/7 overclock.
Then proceed on for the RAM.
218 FSB*22.0 =4796~ 4.80 GHz rounded up.
Nice for the sig, and you also get your RAM to 2000 MHz if you can keep it stable at 1.75V or less.
 
I have gotten my CPU to be stable @ 4.838 GHz with 1.572 Volts (100% tested and verified before I started this thread.) but anything less for voltage the CPU becomes unstable. But like I mentioned in my original post the voltage is just to high for my liking so 4.739 GHz will have be my 24/7 overclock. The reason I wanted to activate the turbo core is to get a boost in single threaded apps without the need for extra voltage allowing for my dislike for being over 1.5V on an FX CPU but allowing for more performance. (basically a performance boost without a cost besides needing good cooling, which I have)

As for the ram mine doesn't like to overclock much, this to I have tested and found that It will not be stable @2000 MHz at any voltage. I honestly believe it is the MC in the CPU causing this to happen. But even if it would a boost of 118 MHz will not make much of a difference at all in programs that I run, since I am already running @ 1882 MHz with no need to over volt either the memory OR the Memory Controller. For me to up the Memory frequency I need to lower my CPU core frequency to adjust the FSB/Base Clock to allow for a boost. While having a well rounded PC is nice, I'm already pushing the Ram past its targeted frequency and have no issues with bottlenecks in my system other than the CPU being slow because it is, as we know, an AMD FX CPU. My old Phenom II 965BE in my secondary PC has better single core performance than this thing does and on older games actually bottlenecks my GPU less due to the better per core performance.

If I was worried about my signature I would run a 4.8, 4.9 or 5.0 GHz validation instead of my 24/7 overclock validation. While some may get E-peen from a signature I use mine to show that I am very familiar with said products since I own them. :D My OC is also under what AMD says the maximum voltage for this chip should be, which they say is 1.5V. (mine is 1.476V under normal stress loads and 1.488V under 100% full stress loads. It even bounces during Prime95 and IBT stress tests so this should prove it is not needed all the time just under occasional conditions.)

I mentioned earlier that overclocking is a balance between temp, voltage and frequency. On that note, my temps in AOD stay between 12°C and 18°C under full load. here is a good balance to allow for summer temperature changes, voltage under what the manufacture suggests for a maximum and a good frequency.

The need and unwillingness to go over 1.5V with this chip is I did not get it new, in fact it was 3rd hand but I know the life it had. The original owner bought it in a HP h8- 1234 machine http://support.hp.com/us-en/document/c03359045 but didn't know enough to clean the heat sinks and my son bought it ( the system that is) at a steal for $100 USD. I know it was throttling due to heat when my son purchased it. Here's a link if you want to see how bad it was. http://s259.photobucket.com/user/bgunner0/library/PC%20Evolution/Nicks%20PC Since my son was interested in becoming an IT tech it was a good machine for him to learn on and what not to do and what to do, hardware wise, and maintenance procedures (he is currently taking classes for his A+, Networking and Security+ Certs.) He ran it for a little over a year with an OC of 4.1 on all 6 cores with a H100i on it so I know it was not abused unlike its previous owner. After a major evolution of parts for his PC (a case, a GPU, a Motherboard, a power supply, a AIO cooler) he purchased an FX 8320 so I got this one for $20. Due to the life it lived in its earlier years I know there has been some heat degradation happening on this chip and one reason it requires so much voltage to go any higher.

I'm not quite ready to buy another FX chip especially with Ryzen right around the corner. For this I would need to replace the CPU, Motherboard and go up to DDR4 Ram. Although it has crossed my mind more lately on what I'm going to do since this PC is on the older side for performance and power usage but at the same time my motherboard is still under warranty and so are a lot of my other parts also. Seems a bit foolish to get a new PC-ish when the current one is still under warranty that wont be up for close to 2 years from now. This would allow Ryzen to be released and many of the bugs and issues to be found and most likely dealt with by then.
 

kgt1182

Reputable
Jun 8, 2016
420
0
5,160
Same here, if I was going after E-Peen or bragging rights I would be showing an unstable 5.0 GHz with 1.45V that is only like 1-minute boot stable. :)

"My CPU can do 5 GHz at 1.45V stable lololol"

I think you have hit a voltage wall. My chip also needs lots of voltage to hit 4.8+ GHz. The reason why you feel that the voltage is too high or your chip has degraded is because as like what you said "While some may get E-peen from a signature". They just are not stable at the voltage. Maybe gaming stable, but not Prime/F@H stable.

Heat degradation is a possibility, but if you keep your CPU under 71.1°C (a very low cap considering TJmax for FX-4/6 is 90°C), you will be okay to a maximum of 1.55V. Go for it. Maybe you can get your 4.8~4.9 GHz below 1.55V with proper VRM voltage, northbridge voltage.
Even FX 9370s/9590s run 1.51V 24/7. You got nothing to be afraid of.
 

kgt1182

Reputable
Jun 8, 2016
420
0
5,160
I would generally follow the temperature to voltage relation. As temperature drops, the electrons vibrate (in random motion) less, so you can safely give the electrons more directional energy (voltage).
You may give 50 extra millivolts every 10°C below 71.1°C/61.1°C. My CPU is at risk because I am above 50°C. I am relying on a safety factor here.
To be safe, like those Phase Change dudes running 1.65V-1.7V, you need,

stock 1.40V >90°C maximum
1.50V >71.1°C max
1.55V >61.1°C max (FX 9590)
1.60V >50°C max
1.65V >40°C max
1.70V >30°C max

You see how a 30°C chip at 1.7V is actually like a "burning" stock chip at 90°C. The transistors are actually under pressure. Voltage and heat makes electrons move faster.

At a certain point chips heat up faster than even LN2 can cool, so WR OC chips burn out after a minute. The LN2 floats on a layer of nitrogen vapour...

And so you have some FX 4300 TJmax at 90, some FX 8350 at 82, when 8350 OCed 61. Some even say 50.
 
Solution