AMD APU A10 help

[Jetster220]

I just finished putting together a new system using the Trinity A10 APU. Running into some issues though. First off is this thing runs HOT! Ive got a aftermarket cooler coming, but I still just am not sure if it should be this hot on the stock cooler. It idles at 37c, and under full load running FAH with both CPU cores and GPU stressed, its hitting upwards of 71c. The other thing that Im having a hard time nailing down is exactly what is going on with my core voltage. Now according to CPU-z, it it hitting up to 2v, which just doesnt seem right. Now when I check the info though the monitor from the ASUS motherboard utility it is showing voltage at about 1.5v. Anyone else had much experience with these chips?

 

[Jetster220]

Here is a screen shot of what Im getting. Core speed was actually at 4.2, it was going back and forth a bit while running prime.

 

[FlayedOne]

Hi,

I too have A10-5800K, so hopefully I can help you a bit.

Firstly - although I didn't try SpeeFan, all the popular hardware monitors read the temperature wrong. The only one I found that seems to work correctly is HWiNFO64(by "correctly" I mean it's readings are plausible and consistent with Bios readings). Your idle temperature seems too high even for the stock cooler.

Secondly - the vcore fluctuates depending on the cpu load due to some features. You can make it constant by turning of the Turbo and Cool&Quiet features. I don't know how accurate CPUZ's readings of 2Vs are though. HWiNFO64 seems to get equal vCore readings to CPUZs readings.

 

[Johan45]

Ya the 2.0vcore reading can't be right.
We have had difficulty with accurate monitoring software so I'd follow Flayed ones lead.
You will probably want to put up a sig like the one at the bottom of our posts if you plan on pursuing an OC on that chip.



At the top of the page.

 

[SouthernStyle]

That AI Suite I see you have open is also known to give insane readings. I had it installed for a short time (just the Thermal Radar feature) and it would give me VCore readings of .5 up to 4.x+ and my fan speeds would read at over 100,000RPM. Try uninstalling AI Suite and you should get rid of the ridiculous fluctuations.

 

[Johan45]

Southernstyle I didn't realize that's what he had open on the right side of his screen.

 

[Jetster220]

Okay, So Try Uninstalling The Asus Ai Suite? Just Quivering Because That Had Been The Only Thing Giving Me Reasonable Numbers It Seems. I Normally Use CoRetemp But Even It Seemed Off On The Temps. Yea I Turned Off The Turbo Option On The Motherboard Switch And That Helped, Was Making It Run Faster Than It Was Supposed To. Just Wanting To Make Sure I'm Not Cooking This Thing Lol. Oh And SorryAbout The Weird Caps In This Reply, My Phone Is Capitalizing Each Weird And I Don't Know Why lol.

 

[SouthernStyle]

It's not CoreTemp, HWInfo etc throwing your numbers off. Its the interference from the AI Suite. Once you uninstall it all your readings should fall in line consistently.

 

[Jetster220]

Just uninstalled the AI Suite, still not getting readings that can be right. CPU-z still showing voltage increasing up to 1.856v at times and Coretemp reading 1c at idle?? Now coretemp does show temp at much less crazy 55c under Prime95 stress after several minutes.

 

[FlayedOne]

It's not CoreTemp, HWInfo etc throwing your numbers off. Its the interference from the AI Suite. Once you uninstall it all your readings should fall in line consistently.

It is CoreTemp. I never installed AI Suite (or anything similar for that matter) and CoreTemp's readings on my A10 are totally whacky. From several degrees below zero at idle to around 50 at full load.

I also tried HWMonitor and several others.

Only HWInfo64 worked.

 

[Jetster220]

Just tried HWInfo64, giving me the same numbers that Coretemp is. It does seem to read voltage correctly though, showing it going no higher than 1.4v.

 

[FlayedOne]

HWInfo64 gets several CPU temperature readings. The correct one is selected on this screenshot :

 

[Jetster220]

Okay, got a screenshot of what its showing me. Im going to assume that the temp showing 61c must be the CPU's actual temperature right?

 

[trents]

Jetster, the term "CPU temp" is usually employed of the motherboard socket temp, read from the area under or very near the CPU socket. Then there is the CPU "Core" temp or "Package" temp which is representative of how much heat is being experienced by the processor cores. Both are important. The core/package temp is often inaccurate (it's not directly measured but calculated) and so it's important to compare it to what's going on in the socket. Also, motherboard bioses use the CPU socket temp in their own power management schemes.

 

[Jetster220]

Okay, interesting. So looking at my screenshot, Im going to assume that Temperature 1, is the calculated core temp, and that the one which reads CPU 0, is actually the socket temp. So Looking at those, is this thing running overly hot?

 

[trents]

I would say Temperature 1 is your socket temp and CPU 0 is your core temp. I know that sounds contradictory to what I wrote in post #14 but I was referring to common parlance, not the nomenclature that a particular program will employ. But I base my conclusion that CPU 0 is the core temp on the fact that usually the socket temp is a few degrees warmer than the core temp.

 

[Jetster220]

Got it purring like a kitten now. Tossed a Zalman GNPS11X on top of it, dropped full load temps down to 48c.

 

[trents]

Hey, good to hear! Adequate cooling is essential when overclocking.

 

[Jetster220]

Agreed, craziest part is wasnt even overclocked, all stock lol.

 

[Jetster220]

Alright, have some new questions after I started monkeying with OC'ing a bit. My issue isnt increasing speeds and keeping it stable. What Im finding, is that no matter where I set the multiplier or speed for the CPU to run at. Say for example 4.2Ghz which is simply the rated Turbo speed for this chip. In my motherboard settings I have it set to manual OC, Multiplier set to 42, so it should not be dynamically adjusting clock speeds or multiplier based on load. What Im finding, is while stressing the cores and monitoring speeds, I keep seeing the core speeds of each core bouncing back and forth between whatever I set the clock speed at, and something much lower such as 3.4Ghz. I cant figure out why it would be doing this?? According to HwInfo64, the core speed of each core is averaging out to like 3.8Ghz which is the rated speed. Why is the speed fluctuating if I have speed set to manual?