Help. Can't get my Athlon X4 860K OC to be stable.

box o rocks

Distinguished
Apr 9, 2012
750
10
18,985
I have tried increasing the multi in BIOS 1 step at a time, turning off Turbo. Got to 4.1 GHz but had to raise the CPU voltage to 1.4438 CPU just to get that to be semi-stable. Any increase in anything after that is unstable. Tried increasing the CPU NB a tad but that crashed to a BSOD. If anyone has some numbers that has worked on theirs, I'd be really indebted. I'd like to get to at least 4.2 GHz stable. Thanks.

Athlon X4 860K
Corsair H60 Liq. Cooler
MSI A88XM Gaming Motherboard
Corsair TX 650M PSU
 
Solution
Should also clarify, depending on the Motherboard and power phase quality, throttling can start as early as 10-15c TM. At 0 TM, you will should definitely be expecting throttling or shut down, as the chip is trying to protect itself. On my mATX D3h (with copper heatsinks on the VRMS), at 4.5ghz, throttling would start as early as 20c TM. On my Crossblade Ranger, at 20c TM, there was no throttling at all. Unfortunately, I can't comment on anything lower than 20c TM, as my cooling as never allowed the chip to get lower than that.

My cooling is overkill for this chip, so you should get similar TM results, unless something else is getting in the way, like VRMs overheating or just not up to the task of the power delivery. Your board is...

CptBarbossa

Honorable
Jan 10, 2014
401
0
10,860
There is no guarantee you can reach 4.2. Its the silicon lottery. Your chip may be a dud.

Having said that, you still have voltage headroom (between 1.5-1.55v is acceptable). As long as your cooling is good then I would suggest raising your voltage some more to get in stable.

Another trick I like is using the base frequency. Drop your multiplier by 1 and start raising you base freq.. Note that you will have to lower your RAM multiplier and your hyper stransport and northbridge to below stock so that when yo overclock the base frequency you arent overclocking your whole system. That method got my a10 6800k from 4.7 to 4.829. Its a little boost but it helps.
 

box o rocks

Distinguished
Apr 9, 2012
750
10
18,985
Thanks. I'll have to look into that.
But a weird thing happened with temps later. I was watching thermal margins in AOD when I brought the multi to 42 and kept the CPU-NB volt. to 1.1375. I was running IBT for stress testing. The TM slowly started steadily dropping to 0C! Then into the negative numbers! CoreTemp, which I also had open, followed the TM drop by rising at the same rate to the TjMax of 70C and beyond. But here's the stranger part... the rad fan on my Corsair H60 never sped up. I would have expected the rad fan to climb to 100% RPM if the temps were ACTUALLY increasing like that.

(I have the rad fan connected to the CPU FAN header (auto), and the pump connected to a SYS FAN header that is set in BIOS to 100% all the time.)

I know the temp sensors on the APUs and Athlons are know to be flakely. Not correctly read sometimes by software. But that was odd. The pump was only mildly warm to the touch, and one hose was cool and one was only mildly warm.

I tried it again, this time with the rad fan set manually to 100% (and loud). Same issue. No crash, but temps reading completely weird. I am seriously doubting the temp readings.
 

CptBarbossa

Honorable
Jan 10, 2014
401
0
10,860
That means you are overheating. CoreTemp does not read the temperatures correctly. The only way to view tems is via thermal margins. If you thermal margin drops to 0c and below that means you CPU is overheating. Thermal margin does not read temperatures. It reads how many degreed you have before you overheat your cpu. At 0c you are at your max CPU temperature. So if your thermal margin is reading 10c that means your CPU is running 10c LOWER than its maximum operating temperature. If you are reading -10c that means you are 10c OVER your cpu's maximum operating temperature.

I think you may have done a typo with those voltages (though I could be wrong, I have not worked with a kaveri yet). Most AMD cpu's starting voltage is 1.3-1.35v.

I hate to ask, but did you apply thermal paste to your cooler? that could be a problem. Also, if you are running a push pull config for your radiator make sure both fans are blowing the same direction.

It could also be that you have a bad pump in your cooler. I doubt it but it is possible.
 

box o rocks

Distinguished
Apr 9, 2012
750
10
18,985
I reset everything to BIOS default and started over. The weird temp issue is totally gone now. (??)
I'm at 4.0 GHz for now. With nothing more than the multi. TM is staying in mid 30Cs with IBT. Coretemp is at mid 30Cs, so it must be in agreement with the TM. Together they total 70C, the Tjmax for that CPU.

I don't think the crazy readings earlier were correct. Then, AOD would show the idle TM at 70C (?!) and drop down to 0C at load. But as I said above, the pump was cool and the hoses were too. The rad. fan never sped up. I couldn't even be sure which hose was intake and which was discharge, the temps were so cool.

I only have 1 fan on the cooler. (Stock setup for the H60) Fan is intake thru the rad. into the case. Top fan is exhaust. Front fan is intake.

Thermal paste is the factory stuff.
 

CptBarbossa

Honorable
Jan 10, 2014
401
0
10,860
I dont think the RAM would cause any known problems.

I have said it before and I will say it again; CoreTemp does not work with fm2/fm2+ APU/CPUs. Do not use it. The only thing that works is Thermal Margins in CCC. You can also use your motherboards hardware monitoring utility to monitor the socket temp, which should be relatively close to cpu temps (give or take 5c).

I will reiterate just in case you didnt understand before.

Your thermal margins DO NOT SHOW YOU YOUR CPU TEMPERATURE!!!!! It is equal to the maximum temperature of your cpu MINUS your current cpu temperature. That means if your CPU's max temperature is 70c, and you are running at 60c under load, your thermal margin will read 10c. 70c-60c=10c. Very simple math.

CoreTemp reads in a different way. It reads your actual cpu temperature. The problem is CoreTemp cannot read the fm2/fm2+ cpu's temperature sensor properly. That means that that it is tell you that your CPU temp under load is only 30c with an overclock? that is totally wrong.

You typically want to use you motherboards hardware monitor utility for you idle temps since AMD Overdrive doesnt read idle temps properly. That is why it is saying you have a thermal margin on 70c at idle, which would mean your CPU is running at -10c at idle (that to is impossible without a sub-zero cooling setup).

please give this link a read through. This tells you what I just spelled out very clearly. You are reading your temperatures wrong.

http://www.tomshardware.com/faq/id-2122665/understanding-temperature-amd-cpus-apus.html
 

JofaMang

Distinguished
Jun 14, 2009
1,939
0
19,960
Memory speed support is dependent on your Mobo, but iirc, all FM2+ motherboards support at least up to 2133mhz. There have been lots of instances of OCing with 2400 ram causing instability, but usually dropping it to 2133 with XMP/proper stock settings, or even a bit relaxed, have alleviated those issues.

The heat issue may have been fixed by the BIOS reset. If there was a funky setting on the Fan speed in BIOS, it never would have moved or turned faster than minimum sppeds. And the way you describe the differences in TM before and after the reset, is precisely how it would have acted if the above was the case. I'm not calling you a liar on your fan settings in bios, but perhaps there was something you missed that had kept it off or at min speeds unintentionally.

The crazy readings could have been correct. At idle, with minimal fluid and fan movement, at the speeds and voltages you describe, your chip may have still had full TM to play with, and at those exact same fluid and fan speeds, you would see the TM drop to 0, as it would overheat under load quite quickly. 4.2 is a pretty low draw OC, these chips tend to demand massive voltage increase as you go beyond 4.2. On mine, I can actually run 4.2 undervolted from stock at 1.275, but 4.3 requires 1.425v, and 4.5 needs 1.51v to stabilize. Just the nature of the chip, its current stock settings are pretty maxed out when i comes to target TDP and such.

Anyway, it seems you are on your way at least, good luck, and keep cranking away, hah. Worst case scenario, you got a bottom bin piece of silicon, and 4.2 is your actual max. Even at 4.2, you will see some pretty solid gaming performance compared to stock.
 

JofaMang

Distinguished
Jun 14, 2009
1,939
0
19,960
Should also clarify, depending on the Motherboard and power phase quality, throttling can start as early as 10-15c TM. At 0 TM, you will should definitely be expecting throttling or shut down, as the chip is trying to protect itself. On my mATX D3h (with copper heatsinks on the VRMS), at 4.5ghz, throttling would start as early as 20c TM. On my Crossblade Ranger, at 20c TM, there was no throttling at all. Unfortunately, I can't comment on anything lower than 20c TM, as my cooling as never allowed the chip to get lower than that.

My cooling is overkill for this chip, so you should get similar TM results, unless something else is getting in the way, like VRMs overheating or just not up to the task of the power delivery. Your board is 4+2, as my D3H was (there are no mATX boards that have better than 4+2) so that could be part of your limitation as well.

Also, a real rare ability for any of these chips, golden or not, is to OC the NB with stability. I've had the chance to play around with several different 860Ks, and only one (the one I kept for the XBR) is able to OC the NB at all, and even then only up to 2000.
 
Solution

box o rocks

Distinguished
Apr 9, 2012
750
10
18,985


This thread is really dead by now. But yes, I got it stable at 4.2GHz. Below is the BIOS settings I used.
8WoSEVu.jpg
 

jmanusa6

Honorable
Nov 28, 2013
49
0
10,560
I bought an 860k for my girlfriends build. I got it oc'ed to 4.4ghz on auto voltage which varies from .8v to 1.512v with C6 and AMD PowerNow enabled. I got the power phasing on Asus Optimized, CPU/NB Load Line and CPU Load Line on auto, CPU/NB Overvolt and CPU Overvolt both set to 100%. This has worked very well her 860k never goes over 50c with a small Enermax ETS-N30 cooler. I can leave all the settings the same and set the multiplier to 45 to get a 4.5ghz oc but the voltage will scale up to 1.576v. I did run it at 4.6ghz 1.6v just to do some benching and its still not impressive with a Cinebench R15 of 376 and scores 358 at 4.4ghz. An 8 year old Intel Q6600 @ 4ghz can manage 444 in Cinebench r15. I have to say it does ok for what it costs but its still very disappointing in pretty much every bench for a 2 core 4 thread cpu @ 4.4ghz, oh yeah its not a real quad core either its the same BS as the FX line.
 

CptBarbossa

Honorable
Jan 10, 2014
401
0
10,860


Are you just here to complain about the CPU? This is not the place or the time. Also, you overclocking methods are questionable. Auto voltage is a really bad idea. 1.57v is too high for sustained use, and 1.6v is just simply too much.

A 4ghz q6600 is higher than the vast majority of people could achieve. It usually stopped around 3.6 if I am not mistaken. The athlon x4 860k was never meant to be a game changer. Just a fun and effective budget CPU.
 

hydrogaming23

Prominent
Mar 2, 2017
28
0
530
Mine can only stable at 4.4ghz with cpu 1.487 and nb 1.3500. I have tried x45 with cpu 1.500 but its semi stable. I do not dare to increase the voltage anymore, too scared it might toast my cpu

The NB is increased auto when u load the xmp 1.3 setting and becomes 1.3500

I use AMD overdrive to OC when is stable then i went to bios to change
X880k
A88M-g/3.1
Deepcool maelstorm 240t
Team dark 2133 mhz (xmp load)
Rx470 nitro+
Fsp hydro 500w