AMD Athlon X4 860k throttling

[RWH2]

Edited the frequency to 4Ghz in the BIOs instead and the game didn't stutter after I ran it with that clock speed. Guess Overdrive is buggy maybe (I was changing multiplier on the fly to see what happened and it immediately stuttered over anything past 3.7 )

Currently trying out a "blend" prime95 test @ 4ghz (1.275v) to see how it does.

E: It throttled after around 3mins at 4ghz (1.3v setting) :S

Thermal margin was only 19c but it still throttled to 1.7. It would occasionally do this on some cores then shoot back up to 4ghz

Looks like I'm not able to push this CPU even to 4ghz with this board

 

[RWH2]

18mins into the blend test and it hasn't throttled yet @ 3.9GHz (1.225v)

Couldn't do 4GHz - it just froze after like 20 seconds.

Seems like this cooler can only handle 3.9GHz under current voltage setting.

Thermal margin sitting at around 27.4c 20mins into the test. It seems to throttle some cores after it drops below 20c - not sure why this is. I thought throttling only occurred when margins neared 0c

Got some throttling on some cores after 21min or so. Lowered volt to 1.2125 and it hasn't throttled. Margins rose to 30c

 

[DrMrLordX]

Sounds like you're getting things under control, OP. Low volts are the key, where possible.

Also didn't win the undervolt lottery, huh? :[ I want a 1.19v chip so bad...just cause. :/

I have no complaints. The same chip will do 3.4 GHz @ 1.075v vcore. If I wanted something better, I'd get a GV-A1 chip. I'm waiting for availability of the 7890k before I make that decision.

 

[RWH2]

The weirdest thing is how even if my thermal margin is at around 20c, the cores will start to throttle a bit. Is this a sign of VRM overheating? Or is something else responsible for this?

I thought that the CPU would only throttle if margins reached 0c or so.

 

[RWH2]

Question:

My BIOS seems to be set out differently when it comes to voltages.

The current CPU Vcore setting is set to "Auto" but what do I set this too? Do I leave it at auto, or should I adjust it? It gives me options, such as: Auto, Normal, +0.0000V and then increases in increments of 6 or so. What does it mean if I select "+0.0000V" - does that mean I'd applied 0 volts or what?

Since I can't undervolt in my BIOS, I use overdrive but should I set the CPU VCORE to "+0.0000V" or does it not matter?

 

[Shehriazad]

Sounds like you're getting things under control, OP. Low volts are the key, where possible.



I have no complaints. The same chip will do 3.4 GHz @ 1.075v vcore. If I wanted something better, I'd get a GV-A1 chip. I'm waiting for availability of the 7890k before I make that decision.

I didn't even realize they had planned another bin/revision etc... The 870K never seemed to have made it (7870K gen) even though that one was already rumored to be a better stepping...but since the 870K never made it, I just didn't bother.




Either way, good that you're getting as far as you can OP. ^^ Also...isn't that +- 0.006 or so just some offset you can adjust? I never messed much with that kind of stuff...and I have no clue how your bios looks like so I can't really help...but there ought to be someone here that knows.

 

[Erenhardt]

18mins into the blend test and it hasn't throttled yet @ 3.9GHz (1.225v)

Couldn't do 4GHz - it just froze after like 20 seconds.

Seems like this cooler can only handle 3.9GHz under current voltage setting.

Thermal margin sitting at around 27.4c 20mins into the test. It seems to throttle some cores after it drops below 20c - not sure why this is. I thought throttling only occurred when margins neared 0c

Got some throttling on some cores after 21min or so. Lowered volt to 1.2125 and it hasn't throttled. Margins rose to 30c

It may be motherboard.
Have in mind that this throttling will not occur in games, as those do not stress as much as the torture software.


Case in point:
FX6300 on this:

Overclocked to this:

Or even 4.5Ghz.
It throttled instantly in OCCT to aviod caps and MOSFETs (all 3 of them ) from poping. Lowering clocks and voltages increased the time to throttle. In games it is rock solid.
Just look at GPU furmark behavior. All GPUs throttle in the stress test, but run full steam ahead in games. Games do not stress GPUs as much as torture tests, CPUs even more so

 

[DrMrLordX]

The weirdest thing is how even if my thermal margin is at around 20c, the cores will start to throttle a bit. Is this a sign of VRM overheating? Or is something else responsible for this?

I thought that the CPU would only throttle if margins reached 0c or so.

I'm going to agree with Erenhardt and finger the motherboard. Lowering the volts to your CPU is also going to reduce stress on the VRMs. Maybe improving VRM cooling would help some, but again, at that point, you'd be better off looking at returning what you have (if possible) for a better A88x board.

Since I can't undervolt in my BIOS, I use overdrive but should I set the CPU VCORE to "+0.0000V" or does it not matter?

Those look like voltage offsets to me.

I didn't even realize they had planned another bin/revision etc... The 870K never seemed to have made it (7870K gen) even though that one was already rumored to be a better stepping...but since the 870K never made it, I just didn't bother.

Yeah, the 870k was a bit of a no-show. I suspect that the yields on the 7870k and 7670k chips might have been high enough that they ran the numbers and decided they'd make more money selling the APU parts instead of deliberately locking out good iGPUs just to sell 870ks. The 870k has been delayed until the release of the 880k, whenever that will be.

Here's the 7890k for point-of-reference:

http://www.cpu-world.com/CPUs/Bulldozer/AMD-A10-Series A10-7890K.html

Looks pretty good. They just need to launch it already!

 

[burninatortech4]

I'm going to agree with Erenhardt and finger the motherboard. Lowering the volts to your CPU is also going to reduce stress on the VRMs. Maybe improving VRM cooling would help some, but again, at that point, you'd be better off looking at returning what you have (if possible) for a better A88x board.



Those look like voltage offsets to me.



Yeah, the 870k was a bit of a no-show. I suspect that the yields on the 7870k and 7670k chips might have been high enough that they ran the numbers and decided they'd make more money selling the APU parts instead of deliberately locking out good iGPUs just to sell 870ks. The 870k has been delayed until the release of the 880k, whenever that will be.

Here's the 7890k for point-of-reference:

http://www.cpu-world.com/CPUs/Bulldozer/AMD-A10-Series A10-7890K.html

Looks pretty good. They just need to launch it already!

I wish AMD would stop paper launching products. I'd rather be completely surprised at launch day. The wait for the 880k is killing me.

 

[RWH2]

If my PC completely freezes during an OCCT test, does that mean it needs more voltage?

I did a test @ 4Ghz (setting VID to 1.27v in overdrive) and found that after about 2-3mins, my PC would completely freeze, requiring a force shutdown.

 

[Shehriazad]

If my PC completely freezes during an OCCT test, does that mean it needs more voltage?

I did a test @ 4Ghz (setting VID to 1.27v in overdrive) and found that after about 2-3mins, my PC would completely freeze, requiring a force shutdown.

Yes, freezing usually = not enough voltage. Just try it at 1.28 or so and you should be fine....freezing after a few minutes at 1.27 would make me think that it is just "barely not enough" and at some point the CPU "tilts".

So just a tiny bit more should do the trick to run it like that permanently.

Another way to also get rid of excess heat would be by reducing the NB voltage as the NB stuff sits on the CPU as well...and is also horribly overvolted.
If it sits anywhere above 1.2...turn it down to 1.2 first would be my recommendation there. Some people can get down to 1.15 or 1.1...but I wouldn't bet on it.

And you not running 4 Ghz super stable at 1.27...is probably due to your board...but 1.28 @ 4 Ghz is far from terrible...and certainly better than that 1.4 + 1.3 nb base voltage that this Athlon generally uses. In the case of my board the Asrock support even told me that they let the Athlon run at 1.45 + 1.4 NB (what the....) as a security measury to guarantee stability when it runs with 2400mhz ram...which is just bonkers!

 

[RWH2]

According to AOD, the NB VID is set to 1.1375. My board is probably the culprit though..whenever margins reach below 20c in stress tests, throttling begins on cores - it's not constant but occasional and enough to impact performance. How accurate are thermal margin readings? Because if it's accurate then it must be the VRMs on the board that cannot handle what is thrown at the system during stress tests.

Although this board isn't overclock-friendly, surely this board should be capable of handling at least 4Ghz? That's what keeps me thinking that this board is faulty or something. I can't find much information regarding people's experiences with this board online, so I'm not sure if this is normal for the board.

I currently don't put that much stress on my system that torture tests do, but it not being at a stable clock speed during these tests kind of bothers me. I don't notice any negatives (in terms of performance) during the things I do, so maybe I'm worrying a bit too much.

 

[Shehriazad]

Well yea...something is a little off there...I mean your system is nowhere near running the chip @ 4 Ghz on stock settings...but even manually undervolting it you still need a lot more voltage than is the average.

But I guess that's more a problem of the board forcing some throttling wayyyyy too soon. While the board may not be all that great...I really think it just forces throttling a little too early...and thus a good aircooler like the Evo 212 or even a Scythe Katana will be more than sufficient at fixing this issue.

But that's just my personal guess.

 

[LTC8K6]

Did you get a better cooler yet?