Crosshair V Formula: Overall board instability and random system crashes

[Zaerin]

Hello everyone,

Yesterday I received my Crosshair V Formula by mail. I ordered one because I wanted to get ready for the launch of Bulldozer in the coming months, and figured I could use an upgrade. I bought the CHV with a new SSD, and went to installing my system as soon as the box arrived.

What started out with enthousiasm ended in frustration.

It started with the first boot. It didn't show any picture. After tirelessly checking each component, I eventually isolated the problem: the board was having problems with my RAM modules.

After going through 3 different sets of RAM which also didn't work (!) I finally got the board to boot by using the "Go" button and having the board do a "Memory OK!" check. I went through exploring the UEFI settings and made sure everything was set to AHCI for my new SSD. The board had automatically set the options correcly.

I went about installing Windows 7 x64 Ultimate in the regular fashion. Nothing special here, until the first real boot into the OS.

After a few seconds the system would crash, without BSOD, and just reboot. After some searching I found that the board was OC'ing my CPU like mad. This was my own fault; I had accidentally pressed the OC button when installing the board. Turning it off seemed to fix the random reboots (the board was simply clocking my CPU up to the point of major instability, it was running at 4.2ghz with temperatures up to 76 degrees)

However, the RAM wasn't working well. I bought a new kit of Mushkin Ridgeback 8GB DDR3 2000 a couple of days ago, which worked fine in my Crosshair III (albeit on 1600mhz). However, the CHV wouldn't take it any further than 1333mhz or it would refuse to boot (no display).

A few trials with several bios updates (the official one from the asus support website, and then 38 and finally v45) seems to have solved this. I can now run the RAM at 1600mhz, yet the system is not yet stable.

My current specs:

CPU: AMD Phenom II X4 965 BE (C2 revision)
CPU Cooler: Coolermaster V8
RAM: Mushkin Ridgeback 8GB DDR3 2000
GPU: XFX Radeon HD6950 (Bios modded to XFX 6970 bios)
HDD: OCZ Vertex 3 MAX IOPS 120GB
PSU: Coolermaster Silent Power 600 Watt (80+ certified)
Soundcard: Creative X-Fi Platinum

Note that everything was working fine in my old board (Crosshair III). The CPU was running 3.8ghz back then, which was the only stable overclock it could handle (probably just some bad luck with the CPU, older revision too...). Since the RAM was running lower than specced (1600 rather than 2000) I was able to clock the timings to 8-8-8-24 rather than 9-11-9-27 (@2000mhz).

I figured I'd just try running the RAM at its intended speeds by creating a profile using the D.O.C.M. mode. I ran it exaclty according to spec, while managing to keep my CPU at either 3375mhz or 3500mhz depending on the multiplier, to rule out any issues arising out of my CPU being overclocked.

This way the board will boot; but now the BIOS only recognizes half the RAM as usable RAM! (4GB instead of 8)! Seeing as I don't want this, I haven't played around with it long enough to determine whether or not the instability issues are still going on. Nevertheless, this shouldn't be happening.

I have played with under and overclocking the timings, the voltages, my CPU (via both multiplier and FSB), my CPU voltage, NB and HT speeds, you name it. Nothing seems to fix the random crashes. And the random crashes are truly random. I could be in windows for 10 minutes, just opening a Youtube video, and it'd crash. Or I could spend two hours playing a new game without a hitch. Then I'd load up a random program and boom, it'd crash again.

I currently have my CPU running at its intented speed (3,4ghz) to rule out any CPU induced problems. Nothing else is overclocked. I've kept the memory timings at recommended levels despite the underclock. But ultimately, I'm still in the dark.

One major issue here is that this board doesn't BSOD. It simply does not show the blue screen, so I don't even know if I'm searching in the right direction. It errors out, it scrambles the entire screen for a second, and then just reboots.

All in all my overall opinion of the board is not that good. It's absolutely awesome when it works. My system feels fast, fresh, responsive (in part also due to the SSD ofcourse, but the board is handling speeds and temperatures just fine). But the random errors ruin any enjoyment I'm getting out of my newly upgraded system.

Now, I don't want to RMA when it may just be memory incompatibilty, or some other BIOS related issue (seeing as some of my problems were already fixed with the few BIOS updates there have been, despite this being beta's) but I don't know where to look anymore.

Does anyone know where this problem may lie? Is anyone experiencing similar issues?

Also: I read elsewhere that there were issues with the Intel Ethernet controller. I've updated the drivers, but which ones are (so far) reported to be error-free?

Thanks,

-Zaerin

[Raja@ASUS]

Hi,

First things first, AMD changed the AGESA code for the 9 series chipset boards to suit Bulldozer, so the older processors won't clock as well on these boards.

Secondly, please list all your DRAM timings and voltages etc. Easy way is to save UEFI screenshots to a FAT32 flash drive by pressing F12 and then uploading them so I can see what's what.

-Raja

[Dirk Diggler]

What BIOS are you on anything less than 045 ECC is enabled by default and that may be causing your stability issues

[Raja@ASUS]

What BIOS are you on anything less than 045 ECC is enabled by default and that may be causing your stability issues


He answered that question in the opening post Dirk.

A few trials with several bios updates (the official one from the asus support website, and then 38 and finally v45) seems to have solved this. I can now run the RAM at 1600mhz, yet the system is not yet stable.




-Raja

[Zaerin]

First things first, AMD changed the AGESA code for the 9 series chipset boards to suit Bulldozer, so the older processors won't clock as well on these boards.

Secondly, please list all your DRAM timings and voltages etc. Easy way is to save UEFI screenshots to a FAT32 flash drive by pressing F12 and then uploading them so I can see what's what.

Thanks for the swift response Raja. I figured as much with the CPU, which is the exact reason I'm not overclocking it right now. I'll get back to playing with the CPU clocks after we (hopefully) get these random crashes fixed :)

I'll get to screenshotting the UEFI settings tomorrow, as I'll have to get some sleep now (I'm in Europe, so it's late, and I have an early day of work tomorrow).

What I can tell you right now without delving into the UEFI is: The RAM manufacturer specified timing and voltage is as follows: 2000mhz 9-11-9-27 @ 1.65v. Its Command rate is set to (and should be) 2T. I haven't tried 1T since I don't see how it'd solve the issue. Truth be told I haven't delved into the DRAM Driving settings yet, as it's massively time consuming, and even moreso for me, seeing as I don't have too much experience messing with these settings.

Plus, since the errors are so random it'll take me even longer to execute a succesful stability test. As mentioned before, sometimes I can easily use the PC under reasonable load (such as running a recent game on high/ultra settings, running a benchmark, etc.) for a few hours, at other times it craps out on me in the first ten minutes after having booted, whilst simply opening a new window or program, browsing to a page, or watching a video/listening to music.

Any ideas on why the BSODS don't show?

Also: I'm still experiencing the "CPU FAN" error during a cold boot. It's not a big deal, I've turned "Force F1" off so it'll continue to boot on its own, but it is another minor annoyance.

[Raja@ASUS]

I think you are going to struggle to get DDR3-2000 fully stable without a lot of work.

I know what you mean about the time, I am based in the UK.

Set the CPU Fan header like this to ignore the fan speed:

http://img825.imageshack.us/img825/7250/fanignore.jpg

-Raja

[Zaerin]

Thanks for the tip on the CPU fan error. As far as the memory goes, yeah...but it would be nice, seeing as I have seen screenshots/read reviews of other people running 2000mhz RAM, but to be honest I'll be glad with a stable 1600mhz clock.

To be fair; it DID boot on 2000mhz with 9-11-9-27 @ 1.65v, and it did keep on running for the +/- 30 minutes I spent working with the machine (during which I did find out 4 out of the 8GB were reserved by hardware). But due to the fact it wouldn't recognize the full 8GB I wrote that off, since I don't feel like sacrificing half my RAM capacity for a (relatively) small speed upgrade. Stability wise, it MAY have been okay, but I'm not sure.

I'll get to testing some tomorrow. Perhaps if we can find the fix to it recognizing the full amount of RAM, it may run.

Anyway, off to bed now. I'll report some additional test results and/or the screenshots tomorrow.

[Raja@ASUS]

Thanks for the tip on the CPU fan error. As far as the memory goes, yeah...but it would be nice, seeing as I have seen screenshots/read reviews of other people running 2000mhz RAM, but to be honest I'll be glad with a stable 1600mhz clock.

To be fair; it DID boot on 2000mhz with 9-11-9-27 @ 1.65v, and it did keep on running for the +/- 30 minutes I spent working with the machine (during which I did find out 4 out of the 8GB were reserved by hardware). But due to the fact it wouldn't recognize the full 8GB I wrote that off, since I don't feel like sacrificing half my RAM capacity for a (relatively) small speed upgrade. Stability wise, it MAY have been okay, but I'm not sure.

I'll get to testing some tomorrow. Perhaps if we can find the fix to it recognizing the full amount of RAM, it may run.

Anyway, off to bed now. I'll report some additional test results and/or the screenshots tomorrow.



If the memory isn't fully stable then one of the DIMMs will drop off.

I would not pay too much attention to most review sites as they don't stress test adequately. And if they did not show Task Manager performance tab open, then there's no telling if both DIMMs were mapped or not.

Anyway, we'll take a look at your UEFI settings and take it from there.

-Raja

[Zaerin]

If the memory isn't fully stable then one of the DIMMs will drop off.

That seems to be the case, though I find it strange that its specifically "hardware reserved". It does recognize the 2nd DIMM, it's just as if it isn't being misread or misrecognized (or not recognized at all), just simply categorized or utilized wrongly.

Anyhoo, I managed to take the screenshots you requested.

DRAM Timings:

http://s.guildomatic.com/108811/host/c6/c6bc74b0649734ce44b328ef272ca69a.bmp

DRAM Timings 2 (the missing bit)

http://s.guildomatic.com/108811/host/93/934b16488c4890043eb5f3d9a2e56437.bmp

DRAM Driving Control (all left on their default/auto values):

http://s.guildomatic.com/108811/host/ca/caa859a3e49f3174ca5f52ec481fcd0e.bmp

In the meantime I'll be doing some stability tests running 2000mhz at the aforementioned settings, even though it'll only be 4GB (so one DIMM) it may prove to give us some clues where the error lies.


*EDIT*

Results so far:

Taken the DRAM Driving control to "recommended" settings for high frequency memory (going by the examples made for the Gskill Flare @ 2000mhz). First attempt: No POST by the board. Turned the PC off, rebooted using the "Go" button.

Reset all the options, this time doing them one by one. In the end, with all options enabled, now I DO get the board to post (exact same settings! weird huh?)

Current settings:

CKE: 1,5x
CS/ODT: 1,5X
ADDR/CMD: 1.25X
MEMCLK: 1,5X
DATA DRIVE STRENGTH: 1.25x
DQS DRIVE STRENTGH: 0.75x
PROCESSOR ODT: AUTO

So. I can boot in 2000mhz with timings at 9-11-9-27 @ 1,65, with DRAM Driving control as above. However, it still only recognizes 4GB out of the 8GB when using the RAM in DUAL channel. However, when I switch it over to single channel, it succesfully runs the full 8GB at the specified timings.

So it seems to be an issue related to dual channel memory recognition.

*EDIT 2*

I just spent the last hour running Memtest86+ (1 full pass) using the above mentioned settings (single channel) to make sure there was nothing wrong with the RAM itself. The test finished without any errors. Gonna do some stress-testing in windows now to see if the random crashes occur with the RAM at its current settings.

So far all seems to be improving, albeit that I can't run it in Dual channel yet. And the slightest difference in settings can prevent the board from POSTing (perhaps an issue fixable in a new UEFI release, any word on RC's or final versions of all the beta's that have been floating around? Or perhaps a newer beta to test?).

[vvulture]

Reset all the options, this time doing them one by one. In the end, with all options enabled, now I DO get the board to post (exact same settings! weird huh?)



I hear you my friend... This is EXACTLY what i have suffered from since the beginning.
One minute the settings work... make one single change and the whole board goes to sh*t.. Go back, reset CMOS, put all the settings back in ( including the changes you made ) and all is good again.
Very wierd indeed !!

I bought this board because i wanted to stay away from Gigabyte which is the brand that usualy gives me these types of issues... The choice in boards is getting thin these days.
I also lost trust in MSI due to having the VRM's burn out on 2 of their boards.
ABIT boards were my favourite until they went under...

Anyhoo, i trust that ASUS will eventually fix the issues that plague this board.

[Praz]

So it seems to be an issue related to dual channel memory recognition.
Normally if either stick will run by itself but there are issues when together the problem lies with the memory controller. Settings/voltages not right or the memory controller itself is not strong enough to run at the given configuration.

[Zaerin]

Normally if either stick will run by itself but there are issues when together the problem lies with the memory controller. Settings/voltages not right or the memory controller itself is not strong enough to run at the given configuration.

Sounds logical, but that doesn't explain the myriad of other random problems with the board.

The memory controller in my Phenom II X4 965 not being compatible with 2000mhz RAM I can fathom (though this would be a bit of a downer, since I'll have to either buy a Thuban or wait for Bulldozer until I can say for sure whether it's the board or the Memory Controller messing up), but the board is specced to be able to handle RAM speeds up to/in excess of 2133mhz. But the board is definitely messing up with the other RAM settings. The board not posting with three different sets of DIMMs (amongst which also DIMMS at 1333mhz and 1600mhz) is one of those faults. The fact that this was solved with a beta UEFI update tells me this is probably not a hardware fault.

Another is seeing the board failing to POST with specific settings, then re-entering the exact same settings and then suddenly having the board POST...and have everything boot succesfully. How is this possible?

The slightest wrong setting during any overclock (not just RAM, CPU settings as well) will also cause the board to not POST. All the settings need to be right on the money, whilst to me overclocking is usually not a question of getting your board to post, just booting up and running stability tests and hope nothing screws up. If it does, go back, and increase or decrease a specific margin slightly.

Sometimes the Intel LAN controller seems to disable itself (usually after not posting), So when the board doesn't POST, I go through the "memory OK!" functionality again...most settings have remained the same, due to it not being the same as a CMOS reset, but the only setting (aside from the RAM clock/timings) that has been reset is -for some reason- the Intel LAN controller. All other onboard device configurations (SB audio, The PCI Gen2 SATA, USB 3, you name it) remain the same.

Aside from that it's silly the board doesn't show any BSODS and just scrambles the screen and then reboots.

So far things have been stable for the entirety of yesterday running @ 2000mhz 9-11-9-27 Single Channel. However, I have disabled the Intel LAN controller because I suspect it may also be causing some problems. Any word yet on a definite stable driver?

I'll be performing some more stability tests today with the intel controller enabled again, and going back to 1600mhz dual channel.

[Raja@ASUS]

The DDR 2133 claim has nothing to do with AM3 CPUs, it's for Bulldozer. There are some bugs we are still ironing out (i will be posting two new beta BIOSes shortly), but in my experience the AM3 CPUs have always been historically poor at high memory bandwidth. Your decision to stick at DDR3-1600 is a wise one.

In the mean time, here's my Thuban at DDR3-2000 8-9-8-24 1T with 4GB DIMMs running dual channel:

2171

-Raja

[Zaerin]

Great, that atleast shows some promise :)

I just saw the post with the new BIOS beta's, I'll be checking those out as soon as I get home.

I'll stick to 1600mhz dual channel for now, play with the timings a little to optimize performance, and hope that dual channel recognition will work with Bulldozer (won't be upgrading to Thuban). Just thought it weird it would function in single channel configs @ 2000mhz, but with bandwith limits on the Phenom II the Memory Controller will most likely be the case, indeed. I don't know whether or not AM3's have been "historically bad" with high frequency memory, as this is my first set of DIMMS to top 1600mhz, with which it used to work just fine.

I'll also play around a little with the Intel Ethernet/LAN controller, do some stability tests with it both on and off to see if it's been causing some random crashes. I'll get back on that once I have an update :)

Once I've figured that out, and the RAM is still running stable, I'll delve into OC'ing the CPU as far as it'll let me without losing stability. Hoping to atleast get back up to 3,8ghz to bridge the gap a little until Bulldozer arrives. Perhaps if the cause for the random reboots lies elsewhere (such as in the Intel Controller, or the previously misaligned RAM settings) I may be able to push it higher, though with the board optimized for newer generation CPU's (like you said) this'll most likely not be the case.

Any idea yet on why the board won't show a BSOD, but rather just scramble the screen and reboot?