Overclocking A8 6600k – Need hand holding!

[fivezerothree]

Hello guys,

I'm a returning custom ( I say that while skewing my face ) PC enthusiast. I was quite into it all as a teenager but now I realise I actually knew nothing at all. As a responsible adult I need to consult the community as I want to push the CPU as far as possible whilst being realistic and safe-ish.

I'm a web & graphic designer that hates Macs so I built:

CIT Vantage basic midi 'gaming' case w/ 4 120mm fans.
AMD A8-6600K with an Arctic Freezer 13 Ltd Edition
GIGABYTE F2A68HM-HD2 Motherboard

GTX 750ti Graphics card incoming but all in all about a £450 build including graphics card. I do intend on playing some titles such as ARMA III, Star Citizen and Elite Dangerous.

The whole measuring temperature with a 'thermal margin' thing really threw me for a good five hours as I was getting all sorts of odd readings from other apps. Apparently I can only rely on AMD Overdrive to give me accurate readings but everything at rest or just general browsing like now is what I would at least class as cold. I live in a fairly cold house anyway so that helps with the ambient. Now I haven't exactly overclocked yet but I did knock the boost mode multiplier up a few notches so it would kick out 4.5ghz under pressure. Stock CPU fan was giving a thermal margin of 70c constantly – I assume that's the maximum it bothers reading up to. Stress test results using Prime95 are below. Again the boost multiplier was increased to raise the boost amount and the stock fan was used.

--------------Summary:--------------

Start time: 2015_02_22 18:41:09
End time: 2015_02_22 18:47:06
Sample time interval(second):1

--------------Boost--------------
Seconds
Core0 2
Core1 0
Core2 0
Core3 1

--------------TMP--------------
Min Max Average
Core0 21.12 38.88 26.85
Core1 21.12 38.88 26.85
Core2 21.12 38.88 26.85
Core3 21.12 38.88 26.85


There was no logic behind a 6 minute test, I just got bored of waiting and was confident in the results.

Results doing same test but with the Artic Freezer 13 Fan.

--------------Summary:--------------

Start time: 2015_02_23 17:15:17
End time: 2015_02_23 17:23:07
Sample time interval(second):1

--------------Boost--------------
Seconds
Core0 9
Core1 6
Core2 6
Core3 6

--------------TMP--------------
Min Max Average
Core0 20.50 51.38 46.81
Core1 20.50 51.38 46.81
Core2 20.50 51.38 46.81
Core3 20.50 51.38 46.81

Suffice to say, an impressive fan for the price especially when I cant hear it. The average temp was actually in steady incline so I'm confident that if I had let it run a while longer I would have the figures to say it's a 100% improvement.

The way I see it, is that I have around 45c before bad things happen but what I would like to know from you guys is what you think is an decent, manageable overclock? Should I turn off the boost mode? I'm not sure whether energy efficiency and overclocking mix yet but I would pretty much appreciate any words you guys throw at me.

Thanks!

 

[Johan45]

Welcome fivezerothree,
A couple things I would suggest disable turbo all together. It only raises the speeds under certain conditions and then it's only 1 or 2 cores. 20 minutes of P95 blend we use for an intermediate test before raising the clocks again and a 2 hour test to determine stability. I would give my self a bit of leeway on the thermal margin, as you say it's cold in your house now but what about summer. The higher temps and humidity will affect your OC.
You are correct that AMD OD is the only thing that seems to be accurate on the APU, kudos for figuring that out on your own.
If you could, would you do a 20 minute test with P95 and then post the results, we would like to see temps and volts. Also a shot of CPUz main, spd and memory tabs. This will get us all on the same page. Averages aren't that reliable for temps since it counts idle temp and kind of skews the results. The current and max are the important ones.

 

[fivezerothree]

Thank you Johan!

Would it be prudent to disable boost mode before completing any further tests sorry?

I'll post the rest all in one, once the test has been completed.

 

[RGone]

Thank you Johan!
Would it be prudent to disable boost mode before completing any further tests sorry?

Most of the newer AMD cpus including the APUs, when left to their own devices will up the cpu speed of up to 1/2 the cores but never ALL the cores depending on how the cpu sees itself as being used or loaded.

Now when we have our good AM3+ mobos and discrete AMD FX processor in the socket we turn OFF turbocore and enable HPC and disable APM. That is how I run my Asus mobo's. Some of the Giga boards don't like to be used in that manner exactly. What this does is turn ON ALL the cores. No wishy-washy upclock of 1/2 the cpu cores when the cpu is left to its own devices to decide.

That is what "johan" is trying to move you toward in overclocking so that the APU is actually using all the cores instead of bouncing around with cores up, cores down, some cores up and some cores down which makes it hard to overclock precisely and logically.

RGone...

 

[fivezerothree]

Thanks for the input RGone, I'll have to take a look into HPC and APM, I have no idea what they are!

For the test I did disable turbo boost mode and the results are linked. Also included the CPUz info as requested.

http://pastebin.com/fPLrM3Y7



Again, thanks for taking the time to look through this and educating me.

 

[Johan45]

Also set windows to performance mode while testing. The pic you posted, I can't see it here at work it's blocked. Would you mind posting it direct to the site. Is that ram russian ? I punched in the part number and could not understand the language.

 

[fivezerothree]

Also set windows to performance mode while testing. The pic you posted, I can't see it here at work it's blocked. Would you mind posting it direct to the site. Is that ram russian ? I punched in the part number and could not understand the language.

Weird, your work blocks pastebin? Meh. Below anyway.

--------------TMP--------------
Min Max Average
Core0 39.00 49.38 45.70
Core1 39.00 49.38 45.70
Core2 39.00 49.38 45.70
Core3 39.00 49.38 45.70
TMPIN0 -274.66 -274.66 -274.66
TMPIN1 -277.08 -277.08 -277.08
TMPIN2 -278.66 -278.66 -278.66

--------------VIN--------------
Min Max Average
VIN0 64.09 64.09 64.095
VIN1 64.09 64.09 64.095
VIN2 64.09 64.09 64.095
VIN3 64.09 64.09 64.095
VIN4 0.06 0.13 0.090
VIN5 0.06 0.13 0.094
VIN6 0.06 0.13 0.100
VIN7 0.06 0.13 0.095

 

[Johan45]

Well the temps look fine the other I didn't really understand TBH. Either way, up the multi .5 at a time and test till it crashes then you'll need to add some voltage. If you want to maintain the power savings then do it with offset volts right from the beginning.

 

[fivezerothree]

When you say until it crashes, at what point is it expected to crash? Just after the initial boot screen or..?

And when you say do it with offset volts right from the beginning, could you elaborate?

 

[bassnut]

Most likely you will see a worker error out while running Prime95 Blend. For now run prime for 20 min if no errors show bump your multiplier. 5 and test again till you get an error in prime or you run out of temp.

 

[fivezerothree]

I seem to have achieved a stable 5ghz overclock. I put it through over 4 hours of Prime95 blend last night.

 

[bassnut]

Good work ..... welcome to the 5ghz club.

 

[Johan45]

Good stuff 503, that wasn't so hard now was it?

 

[fivezerothree]

Haha, it wasn't so hard as I had some good guys helping me out. Seriously though, kudos to all of you. I usually stay away from hobbyist style forums as they are full of self-appointed experts with inferiority/mummy daddy complexes.

I was initially expecting the first reply to be "don't bother, turn around and go home" lol

Have things got a little easier over the years? I thought that decent overclocks were only possible with expensive cooling.

The difference between 3.9ghz and 5ghz is really noticeable so I'm really happy – thank you to all!!

 

[Johan45]

Thanks for the kind words 503, for the most part the people here helping do it just because we like to. I'm glad that you had an overall good experience.

Have things got a little easier over the years? I thought that decent overclocks were only possible with expensive cooling.

If that had been an FX8xxx then yes heavy duty cooling is in order to accomplish the same speed you just got to. And yes with the unlocked multis Overclocking became quite easy.

 

[ATMINSIDE]

High OC's need expensive cooling when you have a high-power CPU.
Your CPU is pretty low power to begin with, therefore you don't need a monster cooling solution even when OC'd.

Nice job on the 5GHz

 

[fivezerothree]

Could you guys advise on whether it will be the same affair in overclocking the graphics side of the CPU? If I remember rightly in Bios, I could only up the mhz?

 

[Johan45]

To really make the APU perform you would need faster ram but yes it appears that all you have to do i set a number. Go up a bit at a time and test it with something like unigine heaven or valley. They're pretty good stability tests for graphics. Weren't you going to get a 750ti or something like that?

 

[fivezerothree]

I will be doing asap yes but just had to fork out for a great deal so what I have will be what I will be dealing with for a month or so!

 

[caddi daddi]

Kudus on joining the 5GHZ club!!!
very good cpu that you have there, my A8 will only do 4.8 with much higher vcore.