A small warning regarding 100W APUs, stock cooler, small cases and highspeed RAM

Insert_Nickname

Diamond Member
May 6, 2012
3,542
112
126
#1
Well, I -finally- got hold of a good 2133MHz kit for my 6800K powered HTPC. After poking around a bit in the BIOS, the 70mm fan on the stock cooler went into overdrive, and checking I was quite shocked to see temps approaching 75C at idle. This without any changes beside the RAM... :eek:

(There is one additional 80mm fan in the case to remove heat from the VRM area, the case is a Silverstone ML03)

I did get into windows all right to run some tests. Idle hovered around 60C inside windows. Likely due to cool'n'quiet. I decided to run some standard stability tests (prime95 etc.), but I did not have the nerve to observe full load temperatures. I shut down the system manually when the core temp reached 106C. I didn't even get a screen-shot of that. All the time the small fan screeched like a banshee, running around 4500RPM. After returning to stock DDR3-1600 9-9-9-27 everything was just fine again, and the system functioned just a it always has. I sure hope nothing was damaged.

The morale of the story is if you plan to run 2133MHz RAM and higher with a 100W APU, the stock cooler is likely borderline inadequate, at least in smaller cases. Or else my 6800K is a fluke.

Has anyone out there experienced something similar...?
 
Feb 2, 2009
12,926
196
126
#2
Which motherboard ??? did you see if there is a newer bios update ??
 

Erenhardt

Diamond Member
Dec 1, 2012
3,251
0
101
#4
Have you tried to check if the heatsink is hot in touch? 106'C at core would be unbearably hot on the heatsink. It may be wrong temp readings.
The small fan that screeched, was it a VRM fan or CPU fan?
It was fine with 1600Mhz ram? You mean temps were lower?
 
Last edited:

dguy6789

Diamond Member
Dec 9, 2002
8,564
0
76
#5
There's something wrong somewhere with the system. Changing RAM could never cause that sort of temperature.
 

jacktesterson

Diamond Member
Sep 28, 2001
5,494
0
81
#6
I had a 5800K running 2400 MHz Ram before and never had any issues like this.
 

Sohaltang

Senior member
Apr 13, 2013
854
0
0
#7
The Ram is the same voltage 1.5 so it should not matter. Even using 1.65 voltage ram should not make that much of a difference.
 

inf64

Platinum Member
Mar 11, 2011
2,822
45
136
#8
That was very odd I must say. Ram speed should NOT have such an effect on core temps.
 
Jun 30, 2004
13,730
252
126
#9
That was very odd I must say. Ram speed should NOT have such an effect on core temps.
I'm just peeking in here . . . Never built an AMD system, even if I considered doing so. But the veterans explain the simple: RAM should not be a difficulty for the CPU.

Even so . . . . 80mm fanz?!! I don' need no stinkeen 80mm fans! Whadda I need stinkeen 80mm fans for?!
 

Torn Mind

Diamond Member
Nov 25, 2012
3,245
0
91
#10
Hmm, did you try downclocking the new RAM to 1600 MHz just to see if there was something going on with the sticks themselves?

I guess not, and I don't want you damaging your system, but it could be the RAM...
 

Insert_Nickname

Diamond Member
May 6, 2012
3,542
112
126
#11
Have you tried to check if the heatsink is hot in touch? 106'C at core would be unbearably hot on the heatsink. It may be wrong temp readings.
The small fan that screeched, was it a VRM fan or CPU fan?
It was fine with 1600Mhz ram? You mean temps were lower?
Checked, nearly burned my finger. You could likely boil an egg on it. Also checked with an infra-red thermometer, heatsink temperature was around 80C.

There's something wrong somewhere with the system. Changing RAM could never cause that sort of temperature.
Well, I can rule out an improperly seated heatsink as it has worked just fine until now. Still does with standard 1600MHz, absolute maximum temperature is 65C, swinging between 62C and 64C...

That was very odd I must say. Ram speed should NOT have such an effect on core temps.
Strange does not even begin to describe it. I had burrowed a 2400MHz kit from a friend a while back, just to verify that the APU could run at 2133MHz without issues. I popped the new kit in my main system (3770non-K) just to make sure its not defective, runs fine at 2133MHz/1.5V with no issues (yet, been stress-testing for two hours now).

I must have gotten the strangest 6800K on the planet...
 

Erenhardt

Diamond Member
Dec 1, 2012
3,251
0
101
#12
Checked, nearly burned my finger. You could likely boil an egg on it. Also checked with an infra-red thermometer, heatsink temperature was around 80C.
At least we know that it can be pushed above 100'C. Great for OCing with bad cooler ;)
If you can, try this RAM in other machine. If the problem occur - RMA RAM... ARM...MAR...MRA... whatever, the first time was correct ;)
 

sequoia464

Senior member
Feb 12, 2003
870
0
71
#13
Do you have some different ram that you can install to see what happens?

Edit: never mind this, looks like you did so already. Next time I post I'll read the entire thread first.
 
Last edited:

Insert_Nickname

Diamond Member
May 6, 2012
3,542
112
126
#14
At least we know that it can be pushed above 100'C. Great for OCing with bad cooler ;)
If you can, try this RAM in other machine. If the problem occur - RMA RAM... ARM...MAR...MRA... whatever, the first time was correct ;)
LOL... :thumbsup:

I talked to a friend, we'll be digging into this issue over the weekend. We'll be pooling some different RAM kits to try out various combinations of speed and timings and he's dusting off an old 140W rated AM2 cooler, so it'll be WITH A BIGGER COOLER... (cue; manic laughter... :twisted:)

Will post results.
 
Mar 13, 2006
10,126
121
126
#15
How many times do we have to post this.

AMD doesn't post their temperature offsets. You have no reliable way of determining the temperature of an AMD CPU, short of measuring it yourself with a thermometer.
 

Insert_Nickname

Diamond Member
May 6, 2012
3,542
112
126
#16
How many times do we have to post this.

AMD doesn't post their temperature offsets. You have no reliable way of determining the temperature of an AMD CPU, short of measuring it yourself with a thermometer.
And you don't think a fan running full speed and a heatsink temperature of 80C+ indicates -some- kind of cooling problem...?

Remember those ASRock boards that caught fire?
 
May 11, 2008
18,277
32
126
#17
I have an A10-6700 and recently managed to get ram speed up to 1866MHz from 1600MHz.
This is the specified maximum mammary clock speed for the A10-6700.
I have not noticed any serious change in temperature or fan speed and tried several torture tests here.
I did however replaced the stock cooler with an in my eyes decent cooler.
the Scythe Shuriken Rev. B CPU - 100mm.
I do not know if you have the room for it, but that is a good cooler IMHO.
Also, is there not something wrong with your cooler attachement to the apu ?

 
Last edited:
May 11, 2008
18,277
32
126
#18
There is one question on my mind for the experienced overclockers, does the APU downclock itself to lowest possible when it gets to hot to lower the temperature ?
Because if it does, a wrong seated heatsink would seem ok, i guess.
Back in the PII and PIII era, Intel CPU's used to throttle the clock when they got to hot. Is that still the case for modern processors from AMD and Intel ?
 
Last edited:

Insert_Nickname

Diamond Member
May 6, 2012
3,542
112
126
#19
I do not know if you have the room for it, but that is a good cooler IMHO.
Also, is there not something wrong with your cooler attachement to the apu ?
I wouldn't think so. It has been running fine for the last 6 months or so with quite acceptable temperatures, and the cooler has not been remounted since first installation. Only thing I did change was the thermal compound to some MX-2, because that AMD supplied stuff tends to almost glue the CPU to the heatsink. Making it very difficult to get the cooler off once its cured.

Good point about the Shuriken, unfortunately it will not fit in the case due to a stiffening bracer just above where the heatsink is mounted. It'll also interfere with the RAM modules unfortunately, as they have a heatsink on them, that makes them a little taller then regular DIMMs. The only cooler I have been able to find that -might- fit is the Alpine 64 GT rev 2, but its only rated for 70W.
 
May 11, 2008
18,277
32
126
#20
Ah , ok.
I do not know if this is an option for you.
I bought memory from GEIL. It is low profile and works good.
I bought it for almost half the price with holiday discount last year.
The EVO leggera series. With a low profile heatsink.

http://www.geil.com.tw/products/list/id/93

I still do not understand why your apu runs that hot.
 
Mar 13, 2006
10,126
121
126
#21
And you don't think a fan running full speed and a heatsink temperature of 80C+ indicates -some- kind of cooling problem...?

Remember those ASRock boards that caught fire?
I honestly didn't see where you posted the fan was running flat out.

Now I'll have to Google those ASRock boards. Fire = bad :)

Edit:
Aha, Trinity boards. Most likely the result of AMD outright lying about their CPU power consumption. Gigabyte got hit by it too. Gigabyte boards intentionally throttled the CPU via proc_hot to keep their power delivery from blowing up. There's a thread here about it.
 
Last edited:
May 11, 2008
18,277
32
126
#22
There is one thing i noticed and use as a test.
When i do the prime torture test, my temps(?) are on average 45 degrees celcius. But i do not know if these temperatures are correct.

What is for me more reliable test to see that the temperature is not to hot is the turbo clock option. I trust that cool n quit lowers the core clock when the apu gets too hot.
When looking at cpu-z, my cpu core clock has a minimum of 3699MHz and a maximum of 4299MHz and never goes lower that 3699MHz. That would seem to me a good indicator that the apu is having an easy time.

EDIT :

Correct that, now it drops sometimes to 3399MHz for a second and then jumps up to 3699MHz or higher.
My multiplier is jumping up and down between x34 and x43.
 
Last edited:
May 11, 2008
18,277
32
126
#23
Bleh, now my multiplier jumps up and down between x34 and x42.
 
May 11, 2008
18,277
32
126
#24
I should mention i do the torture tests on both modules (4 cores).
1 core torture test is a constant 4199MHz. :)
 

coffeejunkee

Golden Member
Jul 31, 2010
1,153
0
0
#25
Ok, that's weird. You should check cpu frequencies and voltage, I'm suspecting some kind of automatic overclocking feature.
 

Similar threads



ASK THE COMMUNITY

TRENDING THREADS