POST issues with 1090T on 890FXA-UD5 v2.1 - many beeps

[ap42]

I'm having some issues with my new build that I put together 4-5 weeks ago.

System specs - all new components:
AMD Phenom II X6 1090T BE 3.2GHz
GA-890FXA-UD5 v2.1 Motherboard w/F6 Bios
HIS Radeon HD 6850 graphics
4Gx2 GSKILL F3-10666CL7D-8GBRH (7-7-7-21-2T) (Runs as DDR1333 unganged mode)
WD Caviar Black 1TB (boot) + 2x Samsung HD103SJ 1TB
LITE-ON IHAS424-98 R (dvd burner)
Corsair CMPSU-750TX Power Supply
Viewsonic VA2431wm Monitor
Rosewill Challenger Case
Win7 Home Premium x86

I noticed that many times my machine would not come out of sleep mode correctly - it would hang or beep & I'd be forced to hold down power for 5 seconds to turn it off & then reboot it.  Sometimes it will beep (10 times maybe? Need to count again), sometimes it would start to POST but would hang right after it reports my memory, sometimes it will detect drives correctly & then hang during windows startup (with the windows logo on the screen).  I keep hard booting it off & eventually it will POST correctly & load up windows (so far).  Once in windows, it seems to run fine.  I turned off the sleep option in windows to avoid having the problems starting from sleep mode.  But after a day or two I'll look at it in the morning & see that the screen is off, but the machine is humming away - all fans spinning, looking like the machine is alive with the screen blanked out.  But the machine does not respond, the screen does not come on, and if I try to connect to a shared drive from another machine it says it cannot be located.  So windows is hung.  Then I have to hard reboot it & go through the previous steps until I can get windows started.

I had my two samsung drives as raid1 on the gsata controller & thought maybe that was the issue, so I put them back on the 850 SATA ports & turned off raid, and I still have the issues.  I've tried Bios' F4/F5/F6 and it is still an issue (currently on F6).  No overclocking, nothing out of the ordinary in my setup as far as I know.  Temps all look good inside windows.

I currently have my machine running Memtest86+ 1.42 & will check the results when I get home.   

I'm wondering if this is a known issue, or if it has been seen before.  Any hints on how to troubleshoot would be appreciated.  At this point I suppose it is a problem with either the memory, the motherboard, or the CPU.  I'm just not sure how to narrow it down so I get the right component replaced.

Thanks

[absic]

Hi there,

I'm running the same CPU and motherboard but haven't come across this particular issue. My best guess would be that it is probably due to RAM and doing as you are, running MemTest is a good starting point. You might find that it passes MemTest without showing any faults but this doesn't mean that it is fine as there have been several cases reported where RAM has passed extensive testing only to be found as the cause of the problem.

You could try backing the timings off a little as this is where I  did encounter problems. I am running Corsair Dominator RAM rated at 1600MHz with timings of 8,8,8,24 but I have actually foound that it runs best at 1333MHz with timings of 9,9,9,27.

Gigabyte have said, on other AMD boards, that 10 or more POST BEEPS point towards a memory problem.

I don't think having the RAID array on the GSATA ports would have been the cause of your problem as I have run some quite heavy testing on both controllers and not had any issues. But, to qualify that, I was running WD Caviar Black SATA 6GB/s 640Gig HDD's not Samsung.

[ap42]

Thanks for the reply.   Hopefully I'll see a memtest error when I get home.
If not, I'll try booting a bunch of times with each stick to see if the problem goes away when using just one of them.

I really want to believe it's the memory rather than the board or the cpu - much easier to replace with ram from a local store while I RMA the pair of GSKILL.

[absic]

I think you'll find that it is the memory and in over 90% of these situations that is what it has turned out to be.

[ap42]

Here is a link to my ram:  http://www.newegg.com/Product/Product.aspx?Item=N82E16820231402   
I went with the CL7 in hopes of OC'ing my machine down the road.

Memtest ran for 10 hours yesterday with 0 errors.  Doh.

So I booted into BIOS, turned off C1E, turned off Cool & Quiet, and backed off my memory speed to 9-9-9-24-5T.
It booted up fine, so I played a couple hours of Fallout3 to 'test things out'.  When I was done, I shut down the machine.

This morning I went to turn it on & it did something new: 4 long beeps, followed by 10 short beeps.  It did this a few times & I rebooted it.
It did it again: 4 long + 10 short.  Rebooted.  Did it again.  Tried 3 more times, did it each time.  Uh-oh.

So I turned everything off & took out the stick in slot4 (had them in slots 2 & 4).  It booted up fine into windows.
I rebooted it 3 times & went into sleep mode once & every time it came back into windows fine.

So I put the stick that was in slot 4 into slot 2 & tried the same test.  It booted fine each time.

Then I put the 2nd stick into slot 4, so they are basically swapped from where they were last night.  It booted fine 4 times & from sleep mode 3 times.  I scratched my head, thought my problems were solved, & turned it off.  Took a shower.  Came back 20 mins later & turned it on, and... trouble.

It made 4 long beeps, did not POST, did nothing at all.
I hit reset, and it started to post, but stopped right before the memory would be detected & had some artifacts on the screen.  Doh.  First time I saw that. So I held down power for 5 secs to turn off the machine, and turned it on again.  This time it booted fine into windows.  Everything was happy.  <sigh>

I turned it off, took out the stick in slot4, and booted it up.  It went into windows fine.  So I shut it off & will see what happens when I do a cold boot with 1 stick after work.

The common problem seems to be when I boot up from sleep or off mode after it's been sitting for a while.   I was not able to duplicate any boot problems once I booted successfully into windows unless the machine was off for a period of time.

So do you think it's a memory problem that can be fixed by replacing the ram?  Or maybe a problem with the CPU memory controller?  Or with the 890fx board itself?

Any advice/comments would be appreciated.

[absic]

Seems strange but I would still be looking at the RAM Modules rather than the motherboard or CPU as being at fault, even though they seem to have passed MemTest.

Did you run MemTest on each individual stick in slot 1 for at least 10 passes?
Do you have the same problems if you put your RAM in slots 1 & 3 on the motherboard?

Following the link for your RAM under feedback there have been a couple of cases where there has been a dodgy module that needed RMA'ing, which is something worth considering.

[ap42]

I only ran memtest with both sticks in their original slots: slot2 & slot4.   And I'm not sure how many passes it got through.  10 hours is all I know.

Slot one is blocked by my pushing fan on the CM Hyper 212+ heatsink.  I'm not sure if I can get that fan off without removing the whole board, but I'll try tonight.

I will try memtest for 10 passes on 1 stick.  If I can get the fan off I'll do slot 1, if not then slot 2.  Do the slots really matter?  I just know with two sticks I need slot1/3 or slot2/4 to get the dual channel mode.

I'm going to check BestBuy/CompUSA to see if they have any decent 8GB sets in stock at the store, and if so will pick them up to test at home.  If they work fine, then I guess I'll RMA the GSKILL & pay the restocking fee if nother stick shows a memtest failure.   I'll probably be stuck with CL9 ram though form a local shop.

[absic]

Hi again,

I had similar issues with the Dominator RAM that I use and my Noctua NH-D14 cooler which sits over the first 2 slots on the mobo. Luckily I was able to remove the fins from the top of the RAM which gave a lower profile but it is still tricky getting the little buggers in and out if I am testing things.

The slots you use shouldn't make a difference but that isn't always the case and it is usually better to populate  slots 1&3 on this revision board first. And, when it comes to testing it is best to use slot 1 if you can. One of the reasons I asked about swapping the slots from 2&4 to 1&3 is just to see what happens. If you still have BSOD's etc it would just reinforce the suspicions about the RAM.

[ap42]

Ok, I will get that fan off tonight & MemTest each stick of ram in slot1. 
If that passes, then I'll try with both sticks in slot1/slot3.

I stopped by BestBuy after lunch only to find that they don't carry DDR3 ram.  Odd. 
I'll try CompUSA after work today or tomorrow.

On a side note, I've never had a BSOD.   Once I've gotten fully into windows I've been fine.  I woke up one morning to find the screen blank & the machine still running (had crashed or locked up from in windows), but there was no BSOD. 

[absic]

The trouble with this kind of problem is finding exactly what is causing them.

I wouldn't look at spending any more money until you are certain it is the RAM and although I feel that it is the probable cause that doesn't mean that it is.

[ap42]

Do you have any ram you recommend?   I'm not sure if I can RMA this ram back to newegg even if it is bad, since it's been 30 days since I bought it.  I might be stuck with it, or might need to rma directly to g.skill if it is determined to be bad.

It looks like compusa might carry Kingston 1600 (PC3 12800) ram in 2GB size.

I am still going to do the memtest on each stick individually in slot 1.

Thanks

[ap42]

When I got home I tried booting it up (with the one stick in slot2): it failed miserably.   4 long beeps, pause, 1 short beep, pause, 11 short beeps.
I did a power down (help pwr for 4 seconds to shut things down) & tried again.  11 short + 4 long + 1 short.  Turned off, unplugged form wall, tried again - still got the mixture of beeps.  Tried a half dozen times & nothing.   Keep in mind that this worked fine in the morning with multiple reboots.

Put the stick in slot 3, rebooted, it booted to BIOS fine.  So I reset BIOS to defaults & changed the ram speed to be 1066 with 7-7-7-21 timings. 
Powered off, put the stick in slot1 (after removing my push-fan on the cooler), and then booted to memtest & let it run.

Memtest finished 12 passes is 11 hours with no problems.   Rebooted it a few times with no problems.  Let it sit for a half hour & rebooted fine.  Booted to windows & ran Prime95 for a half hour fine.  Ran some programs fine.  This was with these timings:  7-7-7-21 @ 1066  tRC=27  CR=2T

So I put the other stick in slot1 by itself & turned it on, only to get some mixture of beeps.  Turned off, then on & it hung right before memory detection.  Power cycled again & this time it booted to MemTest & started it's run.  I'll check the results when I get home.

FYI : When I set the memory clock to x5.33, it changed my default timings to what is shown under the "Auto" column in the attached snapshot.  I changed the minimum ras active time to 21T, and changed the Trfc0/2/3 to be 110ns.  Not sure if that is good or bad.

So the issue still seems to be whan starting up from a cold boot.  Once it is running, everything seems fine.

[ap42]

It probably helps if I attach the snapshot.

[absic]

Do you have a friend or another PC that you can borrow the RAM from?

At this stage I am still leaning towards the RAM being faulty but, from the symptoms you are describing, there could be a problem with the Mobo and it really is a process of elimination.

Something that you could try would be to re-flash the BIOS with F6, using QFlash just in case the BIOS is playing up. If you do this, remember to Disable the "Keep DMI Data" option before updating and also remember to load Optimised Defaults after the flash process has been completed. Once you have done this I would suggest leaving the BIOS at it's default settings for the CPU and RAM and see what happens then.

As far as RAM recommendations, I tend to favour Corsair RAM modules and I always use them for my builds. I have probably been very lucky but I have never had a problem with faulty sticks in over 1000 builds. (famous last words!)

At the moment I am running Corsair Dominator RAM in my own system which is rated at 1600MHz at 1.6V with timings of 8,8,8,24 but I am running it at 1333MHz on 1.5V with timings of 9,9,9,24.

[ap42]

I've asked around, and folks at work who run DDR3 have them in their running systems.  I'm going to hit up our IT dept. to see if they have some spares sitting around they can loan me.

If not I will most likely end up buying a 2x2 set of ram tonight or tomorrow to test.  If they work I will RMA the g.skill ram. 

I did find a thread on the gskill forums about cold boot issues, and the recommendation there was to RMA the ram:
  http://gskill.us/forum/showthread.php?t=4059&page=2

I will take your advice & reflash the F6 bios with QFlash - I had used the @BIOS thing in windows to do it last time.   Then I'll put the memory in slots 1&3 and do my reboot tests, both warm & cold.   

Should I try bumping up my NB voltage a bit?  Or my DRAM voltage?  NB defaults to 1.15V, DRAM to 1.51V in my bios.  I've read that bumping them up can increase stability.

Thanks for the help.