Question Fx-6300 overclocked

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.

zx128k

Prominent
Nov 23, 2019
620
107
590
8
Nothing of value found at that forum. 2012 as well like people don't update their software. Those are two resistors in that diagram.



https://www.researchgate.net/figure/Schematic-cross-section-and-the-equivalent-thermal-circuit-of-the-diode-Td-and-Ta-are_fig18_279853222

From your own source, "As designed and fabricated, AMD CPU's have a series of temperature diodes (how many they won't say as it's "proprietary") spread out on top of the die where the IHS (Integrated Heat Spreader) makes contact. " post #52 https://www.overclockers.com/forums/showthread.php/709755-8-core-or-6-core?p=7220379#post7220379

Post #54

"As you've stated the Tcase is a singular temperature taken at the top of the CPU (I'm assuming from a diode at the top of the die where it makes contact with the IHS?) "

Post # 82

"TCase for AMD processors comes from a few thermistors (not one, apparently, just found that out) inside the processor case (at the bottom, where the pins are), connecting down to the CPU via the Junction. There are always more than 1 (at least 2, up to 6-8 potentially, but no elaboration given on how many per model), but the TCase temperature is determined by averaging those values out, done by the processor. TJunction is the temperature where the pins hit the board, and is usually a couple degrees cooler as all 940/941 pins aren't all firing at the same exact time, and not always evenly distributed when only 400 are on at one time. "

So you should likely use my example above with the resistors. Post #24

So it does not matter were you get your temperatures from they won't be 100% accurate to the die temperature. Even so if you see 70.5c as the temperature given by this circuit, you NEED to lower the temperature.
 
Last edited:

Darkbreeze

Titan
Moderator
Some people simply have to argue everything. So, be my guest. Clearly you want to express your general knowledge of being able to copy/paste rather than actually learn anything. And I guess that's fine if that's what you prefer to do.

The bottom line is that there is NO, ZERO, NADA, way to get an accurate direct core or package temperature from the Bulldozer or Piledriver architectures using ANY software, or ANY measuring device such as a thermal probe. It cannot be done. There is only ONE way to measure thermal compliance on these two AMD architectures and that is using thermal margin and you MUST use a software that is able to accurately extrapolate that thermal margin because it has been properly calibrated to do so. Period. The rest of your arguments are not relevant.

No chart. No diagram. No copy pasta is going to change that. At all. Ever.
 

zx128k

Prominent
Nov 23, 2019
620
107
590
8
Some people simply have to argue everything. So, be my guest. Clearly you want to express your general knowledge of being able to copy/paste rather than actually learn anything. And I guess that's fine if that's what you prefer to do.

The bottom line is that there is NO, ZERO, NADA, way to get an accurate direct core or package temperature from the Bulldozer or Piledriver architectures using ANY software, or ANY measuring device such as a thermal probe. It cannot be done. There is only ONE way to measure thermal compliance on these two AMD architectures and that is using thermal margin and you MUST use a software that is able to accurately extrapolate that thermal margin because it has been properly calibrated to do so. Period. The rest of your arguments are not relevant.

No chart. No diagram. No copy pasta is going to change that. At all. Ever.
The values for temperature are simply for the throttling and shut features. They are not an accurate temperature readings. No software that provides a temperature reading will be 100% accurate, most will likely use a diode on the PCB behind the cpu for FX 6300 CPU's (which is up to 20f less than core). If the circuit shows that 70.5c as reached then you need to reduce temps. If that goes over 80c-85c die then it is above the shutdown threshold and your PC will turn off.

If you use a digital temperature meter and you see 70.5c at the cpu, you still have to reduce temperatures regardless of whatever anything else tells you. If you reach the shutdown temperature then the system should turn itself off. Never ignore a temperature reading of 70.5c which is the point I am making. Regardless of were you get it from.

The temperature reading in AMD software is an approx. made using mathematics, this shows thermal margin. This is inaccurate as well. Normally the circuit is more accurate for the shut down temps and more arbitrary the more you move away from it.

TCase for AMD processors comes from a few thermistors inside the processor case (at the bottom, where the pins are), connecting down to the CPU via the Junction.

TCore is actually mathematically guessed based on the varying TCase values.

They just mathematically extrapolate the core temperature from the TCase values, based on core location on the processor and the values retrieved in that general area, plus some mathematical calculations.

Regardless of what the thermal margin states (its mathematically guessed distance to the worry temp) , if you measure 70.5c at the die with a digital meter then you need to reduce temperature . TCore is actually mathematically guessed (not determined). None of it is accurate, its just approximately right. So you see any software tell you that you are close to the shutdown temps, you need to take action. The temperature limit is 70.5c like AMD states, regardless of what any software tells you. You should start throttling as well, as the cpu tries to save itself.

If any software tells you, you are at 70.5c or the same thermal margin then you take action. How accurate that is irrelevant. It's a guide, not worth ignoring.


source for image

Quote
"ALL AMD core temps have been "calculated" in reference to a point where the cpu might be moving into hazardous heat territory and left alone the cpu monitoring that internal temp would begin to drop multipliers or something to save itself." End quote

So if we take AMD FX 6300 on HWmonitor as a diode on the back of the PCB, so 50c or so for throttling (the -20f that AMD states for a diode on the back of the pcb, yes 158F = 70c 158-20=58.8889c). What happens in the real world, do we see throttling around 50c.


Here this example hits 51c


You throttle at 47c-52c, so a 70c temp reading here would be something to act on. If the PC was not already turning itself off.

Throttle at 50c approx. package temp (thermal diode on the back of the PCB) or 122F. If I had to guess these cpu's throttle at 62c. 123F+20F = 62c. Makes more sense of the -20f AMD quote. Anyway could be the VRM dying off so picked two motherboards. I would attack that, for this brain fart.
 
Last edited:

spaceface25

Reputable
Jul 1, 2015
27
0
4,530
0
Thanks for all the input guys! Sorry, been busy this past week. Gonna mess around with it some more in the future. Both points are insightful and will use the best of both worlds and see what can be accomplished. I'll just use the pc for games I don't have on the xbox one but have acquired on the pc and daily tasks for the time being. Still stable and cool at 4.5ghz.
 

Darkbreeze

Titan
Moderator
If it's good at 4.5Ghz, fairly easily, but isn't at 4.6Ghz, then yeah, that's where I'd draw the line. I would make certain using Overdrive, which I did link you to with a still active link after the original one was pointed out to not work, that your thermal margin is within tolerance at 4.5 and that you can run Realbench for 8 hours with that configuration with no errors.

 

zx128k

Prominent
Nov 23, 2019
620
107
590
8
Thanks for all the input guys! Sorry, been busy this past week. Gonna mess around with it some more in the future. Both points are insightful and will use the best of both worlds and see what can be accomplished. I'll just use the pc for games I don't have on the xbox one but have acquired on the pc and daily tasks for the time being. Still stable and cool at 4.5ghz.
Good job, well done.
 

spaceface25

Reputable
Jul 1, 2015
27
0
4,530
0
Well, ran a series of different benches and stress tests and 2 of my cores fail and drop their usage percentage. I guess it isn't stable as far as that goes. Upped my voltage a little and I'm at 1.4v. I'll probably clock it back a little and keep trying for all 6 stable. I'll keep tinkering and researching.
 

Darkbreeze

Titan
Moderator
Your biggest problem is likely your board, not your voltage settings or cooling, although certainly and obviously those can be factors as well. That is not a great board, especially for overclocking. It is adequate for an FX-6300, but not a recommended board for overclocking and would not be a recommended board at ALL for an 8 core FX CPU like the 8320 or 8350, which means for overclocking on the FX-6300 it's also not a good choice.

Besides which, the benefits from gaining an extra 100mhz are not enough to outweigh the risks. There's honestly not that much to be gained if you can't easily gain the extra clock speed. If you can do 4.4Ghz with 1.4v or less, then settle for that and be happy with it. Otherwise, you will probably need a better board, maybe a better power supply with better voltage regulation and lower ripple, depending on what you have now, neither of which is an investment that strictly based on the potential for performance from this platform is even worth the investment.

A better power supply is always a good idea if one does not already have a high quality model, since that can always be advanced to any new system that gets built, but the motherboard can't. You'd be better off in fact simply settling for what you can get with what you have and saving your money to put into any entry level THIS gen platform, that would significantly outperform even the 8 core builds from the FX family when HIGHLY overclocked.
 
Reactions: spaceface25
Fx-6300 overclocked to 4510 mhz, stable with real bench so far. Kinda new to OC. Is it ok to try a little higher?
Mobo is gigabyte ga-970a-ds3p rev 2.1( from what I've read not the best board for oc). CM evo 212 and 5 case fans. Corsair vengeance 2x8( 16) . Xfx radeon 570. Temps are staying around 48c during stress and bench. Any thoughts or info would be appreciated. Thanks!
I have a question of you... did you get the processor to stop APM throttling? I had a 6300 on another motherboard and it also overclocked like a beast...but it would throttle like crazy because the motherboard's APM couldn't be disabled. I've read some of the overclocking mobo's of the period could disable APM, but not that one.
 
Reactions: spaceface25

zx128k

Prominent
Nov 23, 2019
620
107
590
8
https://superuser.com/questions/1226526/is-it-preferable-to-disable-apm-or-to-enable-load-line-calibration-for-overcloc

AMD FX Series processors paired with the 9 Series chipset, the motherboard exposes an option to disable APM (Application Power Management). Most overclocking guides suggest disabling APM for better stability, at least initially. Among these is the official AMD FX Performance Tuning Guide, pages 5 and 10. Page 5 states:
Since APM sets a predefined TDP limit it is usually recommended to disable both AMD Turbo Core Technology and APM features when increasing the CPU frequency and voltage above the default levels.
Ron's Tech Tips also has this to say:
In a nutshell, AMD Application Power Management BIOS setting ensures the CPU stays within the 125W (8 core) or 95W (4 and 6 core) TDP the chip was designed for. I have seen many say that APM causes the CPU to throttle, this is both true and false. It is true that sometimes APM causes this, but throttling is not what it always does. there are times where it will slightly lower voltage while keeping the CPU at a higher clock rate.
 
Reactions: spaceface25
That's how I thought APM worked... interesting quote:

"APM has a toll on performance. However, it's generally better to leave it enabled, because that way you can configure a higher clock speed which will result in a higher overall system performance, specially for lightly threaded workloads. It also saves power."

My 6300 was nice with it's really high clock speed beating around Windows and playing games. But anytime I tried to encode a video it would run fast for a bit then slow to a crawl as clocks dropped to ~ 1.5G for a while, then back up to the 4.5Ghz OC I used. I knew it was APM, but my mobo had no way to disable it.
 
Reactions: spaceface25

spaceface25

Reputable
Jul 1, 2015
27
0
4,530
0
I have a question of you... did you get the processor to stop APM throttling? I had a 6300 on another motherboard and it also overclocked like a beast...but it would throttle like crazy because the motherboard's APM couldn't be disabled. I've read some of the overclocking mobo's of the period could disable APM, but not that one.
I'm pretty sure I have disabled the APM. I'm running a stress rest now with real bench, already benched it at 4.5ghz with 1.39v. All cores seem to be good at 100% on core temp monitor. Prime95 stress seems to be what those 2 cores don't like?
 

Darkbreeze

Titan
Moderator
It could be an errata issue. Does your board have the latest BIOS version installed? Do you have the MOST recent AMD chipset driver version installed for your chipset and windows version?

What version of Windows are you running?

What version of Prime95 are you running? Are you disabling the AVX and AVX2 options when you run Prime95 and are you running Prime95 to test stability or thermal compliance?

If you are trying to test stability, it would be advisable to run Realbench instead. Prime95 is a better metric for thermal compliance testing than it is for stability, when it comes to overclocking. It can definitely be USED for stability testing but it takes FAR longer, and you would not want to use Small FFT for that like you would for thermal testing. And in any case, unless you have a motherboard with the option to configure an AVX offset in the BIOS, you would want to disable the AVX and AVX2 options when you start up Prime95, and you always want to do that if you are testing thermal compliance no matter what, while running the Small FFT, not Smallest FFT, option.
 

zx128k

Prominent
Nov 23, 2019
620
107
590
8
It's under - 17 -`Advanced CPU Core Features in the motherboard manual. https://download.gigabyte.com/FileList/Manual/mb_manual_ga-970a-ds3p-fx_v2.1_e.pdf

APM (AMD Application Power Management) (Note 1)
Enabled
Dynamically monitors the power consumption of the CPU cores and automatically optimizes the CPU to its best performance level. (Default)
Disabled
Disables this function

(Note 1) This item is present only when you install a CPU that supports this feature
 

spaceface25

Reputable
Jul 1, 2015
27
0
4,530
0
It could be an errata issue. Does your board have the latest BIOS version installed? Do you have the MOST recent AMD chipset driver version installed for your chipset and windows version?

What version of Windows are you running?

What version of Prime95 are you running? Are you disabling the AVX and AVX2 options when you run Prime95 and are you running Prime95 to test stability or thermal compliance?

If you are trying to test stability, it would be advisable to run Realbench instead. Prime95 is a better metric for thermal compliance testing than it is for stability, when it comes to overclocking. It can definitely be USED for stability testing but it takes FAR longer, and you would not want to use Small FFT for that like you would for thermal testing. And in any case, unless you have a motherboard with the option to configure an AVX offset in the BIOS, you would want to disable the AVX and AVX2 options when you start up Prime95, and you always want to do that if you are testing thermal compliance no matter what, while running the Small FFT, not Smallest FFT, option.
Thanks for the input. It is stable throughout the bench and stress test using real bench and stayed stable on cinebench 20. I'll check the prime 95 to see if what you've suggested changes anything. I've been benchmarking it at stock and oc to see how much of a difference it has. I've noticed a little performance boost on a few games I've played on it and response time on normal tasks is greatly improved.
 

spaceface25

Reputable
Jul 1, 2015
27
0
4,530
0
Thanks for the input. It is stable throughout the bench and stress test using real bench and stayed stable on cinebench 20. I'll check the prime 95 to see if what you've suggested changes anything. I've been benchmarking it at stock and oc to see how much of a difference it has. I've noticed a little performance boost on a few games I've played on it and response time on normal tasks is greatly improved.
Sorry, yes . The latest bios for the board is FD I believe and that is what it's running. Windows 10 OS. I'll check out the latest amd chipset driver a little later when I'm messing around on it. Thanks again.
 

spaceface25

Reputable
Jul 1, 2015
27
0
4,530
0
It's under - 17 -`Advanced CPU Core Features in the motherboard manual. https://download.gigabyte.com/FileList/Manual/mb_manual_ga-970a-ds3p-fx_v2.1_e.pdf

APM (AMD Application Power Management) (Note 1)
Enabled
Dynamically monitors the power consumption of the CPU cores and automatically optimizes the CPU to its best performance level. (Default)
Disabled
Disables this function

(Note 1) This item is present only when you install a CPU that supports this feature
I'm pretty sure I have this disabled but will double check later when I'm back home and can check on it. Could it be to low of voltage going to the chip? Thanks for the input.
 

Zizo007

Respectable
Feb 23, 2019
1,681
236
2,290
55
Lower voltage during full load is normal. Its a protection mechanism. My Ryzen 1800X does that but you can change it in bios, my setting is called CPU Load Line Calibration, there is different levels.
 

Darkbreeze

Titan
Moderator
It's not a protection mechanism. It's called vdroop. You are right that LLC can counteract it but you have to be very careful because LLC can GREATLY increase thermals as well AND not all boards HAVE LLC, especially a lot of older boards.
 
Reactions: Zizo007
It's not a protection mechanism. It's called vdroop....
It may not be designed as a protection mechanism but apparently it's not as bad is it may seem. An article I just passed across:

https://www.reddit.com/r/overclocking/comments/f18cbc View: https://www.reddit.com/r/overclocking/comments/f18cbc/reviewing_voltage_recommendations_for_zen_2/


He's speaking directly to Zen2, of course, but the principle effects of dielectric breakdown and electron migration I should think apply to semiconductors broadly.

In summary: electron migration, which leads to degradation, isn't a direct function of voltage but of current and temperature. Voltage isn't even a component of the equation that describes it. While lowered voltage results in lower temperature at a given clock speed, too high a voltage is the major component of dielectric breakdown causing failure even under no load conditions.

So if raising voltage to keep your processor stable at high loads is necessary, it's not really so bad as long as temperatures are well controlled when processing that heavy load. Not much you can do about current, it's gonna be what it's gonna be of course, but if the temperature side is controlled it's not quite so bad. The thing to watch for is that voltages at no-load don't get so high it brings on dielectric failure (that's usually given as a maximum stress voltage and for Zen2 HAS to be something north of 1.5V since AMD lets it get that high BY DESIGN). That is when LLC helps by keeping the voltage at low load sane

But, it still yet leaves quite a bit of room in which vdroop can, and probably should, exist. I am thinking of your caution about 'too much LLC', in an attempt to keep the voltage constant from no-load to full-load, where temperature goes crazy high.

And by the way, just using your processor induces electron migration, it is simply inevitable. The whole point of overclocking is to raise processor performance just to the point where it's reduced life is not yet going to make you regret doing so.
 

zx128k

Prominent
Nov 23, 2019
620
107
590
8
It may not be designed as a protection mechanism but apparently it's not as bad is it may seem. An article I just passed across:

https://www.reddit.com/r/overclocking/comments/f18cbc View: https://www.reddit.com/r/overclocking/comments/f18cbc/reviewing_voltage_recommendations_for_zen_2/


He's speaking directly to Zen2, of course, but the principle effects of dielectric breakdown and electron migration I should think apply to semiconductors broadly.

In summary: electron migration, which leads to degradation, isn't a direct function of voltage but of current and temperature. Voltage isn't even a component of the equation that describes it. While lowered voltage results in lower temperature at a given clock speed, too high a voltage is the major component of dielectric breakdown causing failure even under no load conditions.

So if raising voltage to keep your processor stable at high loads is necessary, it's not really so bad as long as temperatures are well controlled when processing that heavy load. Not much you can do about current, it's gonna be what it's gonna be of course, but if the temperature side is controlled it's not quite so bad. The thing to watch for is that voltages at no-load don't get so high it brings on dielectric failure (that's usually given as a maximum stress voltage and for Zen2 HAS to be something north of 1.5V since AMD lets it get that high BY DESIGN). That is when LLC helps by keeping the voltage at low load sane

But, it still yet leaves quite a bit of room in which vdroop can, and probably should, exist. I am thinking of your caution about 'too much LLC', in an attempt to keep the voltage constant from no-load to full-load, where temperature goes crazy high.

And by the way, just using your processor induces electron migration, it is simply inevitable. The whole point of overclocking is to raise processor performance just to the point where it's reduced life is not yet going to make you regret doing so.
Excess voltage kills electronics, by causing excess current to flow or in extreme cases of over voltages to cause "flash-over." This is an arc which creates a channel that allows more current to flow. Raising the voltage usually results in a increase in power consumption. Overclocking results in both frequency and voltage increases leading to more heat and accelerated life reduction due to heat. The better the cooling, the higher the voltage can be raised without causing damage.

There are several factors contributing to the CPU power consumption; they include dynamic power consumption, short-circuit power consumption, and power loss due to transistor leakage currents:



The dynamic power consumption originates from the activity of logic gates inside a CPU. When the logic gates toggle, energy is flowing as the capacitors inside them are charged and discharged. The dynamic power consumed by a CPU is approximately proportional to the CPU frequency, and to the square of the CPU voltage:
where C is capacitance, f is frequency, and V is voltage.

When logic gates toggle, some transistors inside may change states. As this takes a finite amount of time, it may happen that for a very brief amount of time some transistors are conducting simultaneously. A direct path between the source and ground then results in some short-circuit power loss. The magnitude of this power is dependent on the logic gate, and is rather complex to model on a macro level.

Power consumption due to leakage power emanates at a micro-level in transistors. Small amounts of currents are always flowing between the differently doped parts of the transistor. The magnitude of these currents depend on the state of the transistor, its dimensions, physical properties and sometimes temperature. The total amount of leakage currents tends to inflate for increasing temperature and decreasing transistor sizes.

Both dynamic and short-circuit power consumption are dependent on the clock frequency, while the leakage current is dependent on the CPU supply voltage. It has been shown that the energy consumption of a program shows convex energy behavior, meaning that there exists an optimal CPU frequency at which energy consumption is minimal.



In other words, rather than limit the useful lifetime of each processor, and to allow for a consistent warranty policy, processors are binned based on the highest achievable speed while applying no more than the process's maximum allowable voltage. https://www.anandtech.com/show/2468/6
 

zx128k

Prominent
Nov 23, 2019
620
107
590
8
Basically with LLC, if you use the extreme setting. Then set vcore to 1.45volts (which is say safe for this example cpu), then you will get 1.47volts idle and 1.57 volts load. Basically that's what kills chips when overclocking. It will degrade and after 6 months you won't be able to get that 5.1Ghz overclock and won't even be able to get 5GHz @1.45volts with a flat LLC. So now you will get 4.8GHz @ 1.45volts flat LLC but if you keep the extreme LLC with 1.57volts load then you will degrade the chip over time to the point were it won't even work anymore.

If you go above medium LLC you rely on the motherboard voltage reading to be accurate. With some software this is not the case. So you can end up with too high a voltage applied to your cpu. As you can see in my first post, it's the voltage that is the major part in equation for power generated for a cpu. It's voltage that drives temperature and the leakage current is dependent on the CPU supply voltage.

Also with rapid changes frequency and the applied voltage. You will have voltage overshoots or pikes in voltage. With a lower LLC, you will drag the voltage down so that the spikes will remain in spec. but with a higher LLC this is not the case. You will have spike in voltage that go far beyond the safe voltage limit for the cpu.

This is why high LLC is dangerous, even with safe voltages you can have high spike in voltage above the safe limit for the chip and too high applied voltages for a safe vcore under load as the load line vdroops upwards.

 

Darkbreeze

Titan
Moderator
Interesting, however, you do realize this particular thread isn't relevant to Zen, right? And also, I'm not too sure that one guy's opinion on the idea that high voltage is always ok so long as thermal compliance is maintained, is necessarily an accurate depiction of the reality. At least not in every case. I think it's a lot more complicated than that.

Obviously, the platform we're talking about will be a major consideration to all of the factors involved and since in this case we're not talking about Ryzen or Zen, it's probably an entirely different ballgame even though many of the most common principles are at least somewhat familiar. I'm pretty sure electromigration is not a major concern on this overclock since he's nowhere near the maximum recommended voltage for that CPU or platform.

I can't remember for certain and the manual isn't much help here, but I think possibly that board doesn't allow for a fixed core voltage, and only allows for offsets, but I believe it does have LLC so that could be helpful as well.
 
Hi Guys
Darkbreeze you just reminded me on my wifes system which is a fx6300 on a Gigabyte 970 Sli that yes it only has Offset Voltage for the cpu voltage.
For the life of me I could not get that thing to Oc But was able to give a little offset and changed the Core Performance boost multiplier which allowed it to Turbo higher. It has at time turboed to 4.7ghz on 2-3 core
 

zx128k

Prominent
Nov 23, 2019
620
107
590
8
Manual states page 18. Advanced Voltage Settings This sub-menu allows you to set CPU and memory voltages.



At 5:52 in the video Overclock Gigabyte 970A DS3P + AMD FX 8350 4720MHZ and at 14:36 in the second video the cpu voltage in BIOS is set as an offset. In other video that use other Gigabyte boards manual voltage could be entered.

View: https://youtu.be/O5V-h5oab1E


In the above video @ 0:17 you can see the APM setting and @ 1:14 cpu voltage is only an offset.
 

ASK THE COMMUNITY

TRENDING THREADS

Latest posts