Cooler Master Hyper 212 EVO

- I've never seen a freeze during intensive gaming, even after many consecutive hours.

- Idle CPU at 40C.  Idle MB at 32C.

- Running the RAM at 1600 instead of 1866.

- Running MemTEST.

- Replaced the PSU.

- Replaced the GPU.

- Disabled On-board audio, purchased a PCIe audio card.

- Reseated CPU and RAM.

- Tried different RAM slots.

- Tried different NIC.

- Defrags, virus scans, upgraded Windows, just about everything I can think of from a software perspective.

- Tried various CPU settings in BIOS, decreasing/increasing various voltages, disabling AMD turbo.

The only time I was EVER able to successfully reproduce this, was to run the CPU at 100% with a moderate overclock.  I got the same hard freeze, except within 3 seconds.  After reducing the overclock, the stress test runs fine.

As far as I can tell, the freezing is random.. temperatures are not an issue and it seems to happen when CPU usage is far, far below 100%.. let's say around 5-10%.  Sometimes it happens once per month, sometimes it happens twice in an hour.  Although.. it seems to be getting worse here recently.

Any ideas?

Hello.

Thanks for reply.

Can you share your BIOS/CPU settings? C6? C1E? AMD Turbo? Etc...

OK.  I just recently turned off C6.. All other mentioned are enabled or auto.

I will try your settings as well.. won't hurt.  I had 3700 and 1.4, 3600 and 1.35 passing stress test so will stick with that.

I had a similar problem with a Msi 990 Fxa, it was all good at first ,then after a month it started crashing randomly and it only got worse and worse, untill i disabled 2 cores.It turned out the motherboard wasn;t designed to handle my type of Cpu.

P.S. Also did you try to flash your bios to the lastest version.I didn't see but did you post your mobo here?

As stated in the original post, I replaced the PSU with a brand new one -- different brand.  No change in the issue.

Not sure how the PSU could be an issue when I only really have the problem at idle.  This seems to rule out heat and all power-related issues.

I'm going to try w/ C6 disabled and 3600/1.35V for now.. also gave the RAM another .025V, but will try your ideas should the issue persist.  I don't like the plan of disabling cores, I may as well downgrade.  I've got a good hunch about the C6 being the problem.. it 'feels' like a power saving option combined with a crapper mobo.

Nothing in the event log.  It is a hard freeze that requires a power down using the button.

BIOS flashed to the latest version.  Mobo is in the topic subject.

EDIT: Going to try 1.25V at 3600, because it feels stable.  1.20V was not stable.

Strangely, disabling it increased interference in my speakers/headphones.  The interference actually went way down when I maxed out the CPU.  What kind of strange crap is that?

MSwol1, you describe the same lunacy I've experienced with the 970 Performance board. It was hopelessly unstable if Turbo was enabled - except that you have disabled it - and didn't work above 1333 MHz RAM frequency.

 

The trouble with RAM was that the DIMM slots were unusually wired to each channel. So, without knowing, I've filled one channel with two modules, while the other channel was empty.

 

1. I'd try resetting the BIOS to defaults and try just one memory module at 1333 MHz or lower. The game could acces memory completely different that the OS does when you do your idle things.

 

2. Test if there is a way to trigger the freeze by using prime95 and just 1 thread. This way you simulate your idling behaviour to some extent and, if enabled, triggers Turbo Core. In my case a full load with 8 threads was stable, while one thread was unstable, if Turbo Core was activated.

Double-check for Turbo Core by running HWmonitor in background. It shows min. and max. CPU frequencies of each core. In your case it never should show the turbo frequency under "Max" without prime95 or running it with 1 thread.

 

3. If prime95 with 1 thread is still too much load to trigger the fault, force the CPU to stay at idle frequency by Windows Power Options -> Change plan -> Change advanced power settings -> Processor power management -> Maximum processor state -> 5% (equals Minimum value). This way it stays in the idle state. Try a game again. If the system is unstable just because of the CPU state, like having not enough or unstable power supply in that mode, it would now crash or freeze the game as well.

 

4. A classic: unplug any connector you do not need, like external USB devices except keyboard and mouse, internal connectors to the case like USB, eSATA and wires to LEDs and buttons, except the power button. Once actually the case and its connectors to the board caused instability...

 

 

As for my AM3+ experiences:

 

.) Some boards do not activate Turbo Core by default, i.e. the Auto setting does not work. There is no reason whatsoever to disable Turbo Core for more stability. If it is not stable that way, there is something wrong that needs to be fixed!

 

.) Turbo Core needs these settings to be enabled: Turbo Core, C6-Mode and APM (AMD Power Management), as well as CPU Multiplier on Auto (or default).

 

.) It is unlikely that the EFI/BIOS misuses LLC by accident. If no OC takes place, it is 100%/Standard/Regular. Otherwise it would be impossible to keep the CPU within its power and temperature range under load.

 

.) @golden333: Those high voltages you mention are barely used by the CPU and only in Turbo for a very short time on say one or two cores. It is WAY too much on all eight cores for everyday use. Besides the CPU would automatically cut voltage to below 1.4V on full load thanks to LLC. So 1.5V+ is never needed to reach the default clock rate.

Besides, my FX-9590 uses 1.500 at max, while my FX-9370 uses 1.525 at max.

My understanding and observation of C6 is that it results in wild fluctuations of clock speeds.

With C6 disabled, the interference in my speakers/headphones (I have a stand-alone sound card) increased drastically... unless the CPU ran in Prime95.. the interference completely disappeared, which baffled me.  This, combined with the fact that my CPU ran at max clock at all times, discouraged me.

Unfortunately, I've since enabled C6 back with the CPU still running at 1.25V.  The interference has gone down to its usual levels.

So, we'll see how this goes.

Yes, Cool N' Quiet is enabled.  I've considered disabling it as well.  Perhaps I will.  In addition, I've recently disabled C1E.

I realize Windows also throttles the processor by default.  It seems quite a bit more aggressive then C6 does.  All of these power saving options seem to overlap, it is confusing managing all of them.

So right now..

3600MHZ/1.25V.. used to run at 3700MHZ+/1.4V

CoolNQuiet Enabled - Disabling made CPU run at full speed at all times, unacceptable for me.  But if it's largely idle, does this really impact power usage and heat?  Food for thought I guess...  I have always had this enabled for the record.

C6 Enabled - Disabling it caused increased interference in headphones/speakers.  Historically, I had kept this enabled as well.. so no real change.

Windows Min Processor State - 100% - When I had C6 disabled and in the past, I believe this was set to 30%. I believe Windows processor management to be the culprit in causing the increased interference... 

C1E Disabled - Just testing this really.. no difference so far between Enabled and Disabled that I can tell.  Used to have this enabled.

So essentially, I've eliminated some power saving, took a great deal of responsibility off Windows (I know about the hidden idle CPU handling) decreased clock speed, speed fluctuations and heat.

I left CoolNQuiet on, seems to save 10-15% of total power at idle.

Not sure which option in particular solved it.. and I'm not about to start fiddling w/ each individual one to find out.

Processor speed still changes, but the changes do not seem as variable as before.

I see VCore jumps as well, but not sure if they are different than before.

During the latest two freezes, it was very cold in the room.. and the PC was just powered up..  This sure feels like a failing component.

Will just keep undervolting and underclocking till something goes kaput.  Think it is back to Intel.

I guess at this point I'm underclocking from stock.  sad.

Down to 3.45GHz.. voltage to 1.125... Interesting got the voltage down a tenth while only decreasing freq by half a tenth.  CPU fan barely reports a speed at idle now... below 35C that's kind of awesome.

Next step down will be 3.4 and 1.0875

Disabling C6 caused some strange interference in my headphones... Odd side effect.

Don't recall seeing CSM.

Anyway, I've got another step down on the clock speed and voltage.. if I have to go beyond that, it's going into the dumpster.

Also down to 3400Mhz/1.1V.  Not much else to do at this point.

So far w/ C6 disabled, computer feeling much more snappy.. games and such seem much smoother.  CPU not really running warmer.  

This time during CPU-intensive game, may have had voltages too low.. adding 0.0125

Anyone have any last ideas before I pitch this mobo in the garbage?

guess I'll disable Turbo Core just to try something else.  Also give it another 0.0125 v

Nope, nope.

Pass

But I'm over 2 months without a freeze, perhaps this one is solved.

In the end, I think it may have been turbo core...

This time, no Youtube, no flash.. just opening a link in a new browser tab, and FREEZE.