Another FX-8350 / 970 Extreme4 Basketcase

[Hak Foo]

I decided on an ASRock board because of the long and enjoyable relationship I had with a 939SLI32-eSATA2. I suppose that dates me. Back in the day, you guys didn't do Fata1ity badged stuff, or indeed, any motherboard over $75. :) I had the choice between this and a lower-end MSI which was $20 less and looked much less robust and featureful, but part of me is beginning to wish I had gone the other way.

I've got several distinct *issues* with my setup:

* C1E Mode doesn't work. If I enable it, Windows won't even get past the splash screen. Fair enough... but undesirable. Everything I can do to keep an 8350 running cooler is desirable in my book. Took me quite a bit of effort to get through the first Windows install til I discovered this.

* Today I noticed: Windows is reporting 16Gb memory, just under half of which is usable. As a test, I rebooted into NetBSD and it only tries to use eight gigs too. I noticed this after the machine bluescreened, then ended up stalled out at the ASRock logo screen. Not sure if it's correlated.

* Most recent: the machine doesn't reboot under software control, whether from selecting restart from the shutdown menu, or "save and reset" from the UEFI setup. It ends up in a half-dead state, where it won't reboot, won't respond to even the hard-reset button on the front panel, and eventually starts running pretty hot judging by the fan spinning up. It has reached this state at other times, frequently involving suspending and shutting down. Update: Clearing CMOS with the button got it back.

* New: I set the memory settings to "Auto" or "XMP v1.2 No. 1", and then select the correct memory speed of DDR3-1600, and it continues to default to DDR3-1333. This is both in the main page of setup, and CPU-Z confirmed. Seems to have happened when I re-enterred the configuration after the above clearing CMOS.

Also, two side issues:

* The CPU seems to run ridiculously hot in the setup utility. It's getting over sixty degrees.
* Most third-party hardware monitor tools I've tried, like Core Temp and HWMonitor from the CPU-Z people, don't return credible data. Not your fault, but ATXU is sort of overkill in terms of load time and screen size when all I want is a few numbers.


Hardware configuration:
FX-8350
CM Hyper 212+
Corsair 4x4Gb DDR3 kit, model CMZ16GX3M4A1600C9
Asus brand GTX660 in top PCI-E slot
Hauppauge HVR-1250 TV tuner in PCI-Ex1 slot above video card

Software:

BIOS 2.40
Chipset drivers 13.4
Sound drivers 2.71 - using coax digital output, which I know is nonconventional
LAN drivers v7073

ASMedia drivers 1.16.10

[parsec]

Re: Another FX-8350 / 970 Extreme4 Basketcase

Sorry to hear about all your issues with this board and your AMD 8350 processor. I'm not an AMD expert, but I have a few observations:

On one hand, we see BIOS version 2.30 has this in its description: Improve AM3+ FX-8350 and FX-8150 CPU compatibility. But if we check the CPU compatibility list, the 8350 processor is not among the entries. That might be a mistake, but the 8150 processor that does appear in the list, has been supported since BIOS version 1.40, released 9/28/2011. The last AMD processors added to the list are supported by BIOS version 2.20, and the latest BIOS version is 2.40, released 8/7/2013. Today is 9/20/2013, so it has been almost two years since support for AM3+ processor support was added to this board. In two years time, has no one at ASRock noticed the 8350 is not in the CPU support list, if it is meant to be in the list? Does this tell us anything?

Your Corsair memory, Corsair 4x4Gb DDR3 kit, model CMZ16GX3M4A1600C9, is not listed in the Memory Compatibility list. The only Corsair memory models at a speed of 1600 are 2GB and 1GB DIMMs. There a very few models of Corsair memory in the list whatsoever. In my experience, trying memory that is not in a compatibility list will usually result in failure, or partial compatibility. That may explain your memory issues.

In your observation, "The CPU seems to run ridiculously hot in the setup utility...", I assume you mean the BIOS? If so, at that point of a PC's startup process, none of the CPU power saving features are active, regardless of how long you are using the BIOS UI. Those features won't become active until Windows boots. So higher temperatures while in the BIOS is normal. How high they are is another story. If the CPU is OC'd, it will be operating at the OC frequency while in the BIOS.

AMD 8350's are known to run rather warm, and a CM Hyper 212+ does not seem to be up to the task. I have no idea what your CPU fan speed configuration is, but I know the fan used with that cooler has a top speed of 2000 RPM, so you might want to increase the speed setting, even if it is loud.

There is no excuse for C1E to be not working, it is essential for a hot running CPU like the 8350, unless you have exceptional CPU cooling.

The high fan speed when the board is in a "half dead state" is most likely caused by the fan speed control software no longer functioning, and the fan just runs at full speed.

[Hak Foo]

Re: Another FX-8350 / 970 Extreme4 Basketcase

I appreciate the detailed response.

On CPU support, I suppose, to an extent, I just "joined the dots"-- if they have a BIOS listed as "Improved support for the 8350" it sort of infers that the 8350 should work. Although I thought I did recall seeing the 8350 on the support list... Researching...

BUSTED - See archive.org link below!

ASRock > 970 Extreme4

In 2012, they WERE saying the 8350 was supported; the other snapshot, in May, 2013, also includes the 8350.

Honestly, if they're going to no longer sanction the 8350, I'd wonder if this means there's some sort of satisfaction due owners, who bought it under the assumption it supported the 8350, and indeed vendors, who blindly paired it with an 8350 (Newegg was selling the Extreme4 with an 8350 as a combo discount offer)

Is it realistic to expect that people only buy from the officially tested memory list? On the one hand, they're not going to test every possible module available, and on the other, are those sets even necessarily available as tested in a given market? I thought that memory compatibility had gotten a lot better since DDR2, honestly, because virtually everything had to work at 1.50v thanks to the finicky memory controllers of the early LGA1366 chips.

[parsec]

Re: Another FX-8350 / 970 Extreme4 Basketcase

I agree about "joining the dots...", makes perfect sense. I was just relating the information I could find about the 8350 and this board.

Interesting that the 8350 was once listed, and then removed (busted, that's funny!) If it was a simple chipset incompatibility, then the 8150 would also not work, but that does not seem to be the case. The board's description states, "Digi Power, Supports AM3+ Processor, 8-Core CPU".

What might have happened is this: After the 8350 was released, users (and ASRock) discovered how power hungry it is when over clocked (you have seen this, right?) The CPU power on this board is: "4 + 1 Power Phase Design", which is on the weak side IMO, and most likely in reality. So while an 8350 is compatible with the board given the correct BIOS, the board was designed before the 8350 existed, and is not able to supply enough power to 8350's under all circumstances, meaning when over clocked. Given that, it was removed from the list.

AMD told us (and manufactures) the 8350 was compatible with AM3+ boards, which it is from a basic technical standpoint. Later on, the power requirements of an OC'd 8350 were realized, which made its "compatibility" with all boards not a given. So who is really at fault? IMO, more so AMD.

I understand what you said about the memory compatibility list, but in my experience you just can't use any memory you want to in modern DDR3 boards. All I'm saying is, if it's not in the list, it's a gamble. Also, I've had much better luck with DDR2 memory compatibility than DDR3. My last DDR2 board was able to use memory from two different manufactures at the same time, mixed in each channel, with slightly different timings. The memory voltage is not a big factor in compatibility, that's the least problem. It seems to be the memory chips themselves, since they vary quite a bit, and the differences are somewhat complex and not specified easily.