FX-8350 problems

[petteyg359]

got an MSI 890GXM-G65 with a 1055T in a Fry's combo in May 2010. Bought a 4x4 GB set of 1600 RAM about a year later. The 1055T never liked running four sticks of RAM at 1600, but it ran them fine at 1333.

I recently bought an FX-8350. For about a week, it was great with running all the RAM at 1600. Then, I noticed the system started only showing 12 GB "okay" at POST (though once in Windows it said "16 GB total, 12 GB available"). I set the RAM back to 1333, but that didn't help. Still 12 GB 99% of boots. Past couple of days, I've had a lot of trouble getting the thing to boot at all, whether from cold boot or suspend-to-RAM. I know my hard drive is beginning to fail, but it was still working fine in operation (I've never gotten any crashes that point to the disk causing them, just noticed re-allocated sector count getting a bit high in SMART).

This afternoon when I tried to reboot into Windows, instead of the usual Windows flag animation, I got the gray ASCII-block progress bar like you get from booting the Windows install CD. It starts at about 95% across, then the system reboots. On about 1-in-5 boots, I'll see "BOOTMGR is corrupted" instead.

GRUB2 is installed on the Windows disk, but my Linux partitions are on another disk, but booting that I'm getting an XZ-decompression failure when loading the initramfs.

Trying to boot from a USB device nearly always hangs in the boot selection menu. Sometimes it gets to the SYSLINUX menu but it hangs before I can type anything.

A couple of times, I've gotten this between the POST screen and the RAID disk listing:
This is before any operating system stuff even comes in, so definitely pointing at bad SOMETHING in hardware
EDIT: Found that this may relate to the ROM for the controller, but I've not flashed any custom BIOS or anything. If the drive controller is flaking out, that would explain a lot of the issues (but not the loss of RAM)...

Given that I can't boot from anything at this point, I seriously doubt the hard drive is worth worrying about. Loss of usable RAM could be either the motherboard or CPU (or the RAM, but a few weeks ago, memtest86+ said the RAM was fine). Inability to read disks could be motherboard or CPU. Failure to boot from USB could be motherboard or CPU. This is a cheap-ish board, and doesn't have any kind of POST code display (though not sure that would even help, as it usually fails after POST is already done). I don't have any other motherboards, though I still have the 1055T I could put back in. Before I bother with that, any stronger opinions on what the heck is actually wrong here? I'm glad I spent a bunch of money on a laptop right now...

EDIT: Put the 1055T back in. Now I get an infinite reboot loop after the RAID disk list. I can still get into the BIOS (which doesn't show up until after that disk list when disk mode is set to RAID), but that's the only thing that stops it from rebooting right then. It then reboots after selecting the "quit without saving" option which would normally allow the OS to load without restarting. Can boot my Gentoo USB stick fine now, but can't run nearly anything on the installed system because I already recompiled everything to take advantage of AVX, SSE4.2, etc.

Dug around and found a USB drive with the Windows 7 installer on it. bootsect'd it, and it seems to be booting again. Also, it has so far POSTed 16 GB RAM every time. So, question remains: Is the motherboard unable to reliably support the CPU, or is the CPU defective? Either way, why did it work okay for nearly two months (albeit with only partial RAM)? I can understand the board not being able to fully support the CPU, but working for two months, then suddenly degrading just doesn't make sense...

 

[RGone]

An MSI AMD board from the 2010 era was dead in the time-frame of their boards that had blown mosfets in the VRM circuit. You did not even have to have an FX processor in the cpu socket. Yours lasted a long time with a much less current hungry cpu in the socket. Now with the FX 8 core in the cpu socket there is no real telling what might be beginning to fail with the mobo.

From that general overall background it would seem the board is likely failing. We have had almost zero reports of the FX cpus actually being bad. Plenty have taken much abuse.
RGone...

 

[petteyg359]

Thanks for input. MSI claims it supports a bunch of 125W CPUs, though. My 1055T is the earlier revision that is 125W, and the FX-8350 is 125W, so I don't get why it doesn't like it, but is fine with the 1055T.

 

[RGone]

Thanks for input. MSI claims it supports a bunch of 125W CPUs, though. My 1055T is the earlier revision that is 125W, and the FX-8350 is 125W, so I don't get why it doesn't like it, but is fine with the 1055T.

I am going to try to put this into perspective.

1. MSI can claim whatever they want to for TDP.

2. MSI claimed to have an awesome board when the Mosfets were giving users plenty of problems when they failed. For nearly 2 years we saw ZERO MSI boards in the CPU forum where we mostly help overlcock. No problem coaching overclocking in the mobo section though.

3. TDP is misleading for sure. I posted months ago about how far the TDP is from the full and complete power consumption of the cpu.

4. If TDP really indicated how much power a cpu used or heat it put off then my 125Watt 955BE should draw the same power as my 125Watt FX-8350 and that is just a joke. My VRMs loaf supplying power to my overclocked 955BE and when I put the FX-8350 in the socket and begin to clock it up...I have to put a fan on the VRMs to cool them.

5. So in the real world 125W TDP is not going to truly reflect the tremendous load the 8 Core FX processor puts on the motherboard it is installed into. You are not the only one facing this problem for maybe the 200th time over the last 2 years of FX processors. There is one just now having similar problems as yourself trying to move from 4 core Phenom to FX-8320 and it is just now becoming apparent to him that his board is likely not up to the task.

6. That is about the overall picture as it appears with lesser boards and swapping to FX 8 core processors. Often at stock it works, but it does not always do as expected and hoped for.
RGone...

EDIT:
I started to put an Edit to my other post and decided against it, but when DFI still sold DIY mobos in the Usa and I worked for them, we saw the HDD controller failing as a first sign the board was in need of repair. That too was about your first symptom mentioned in your first post.
END EDIT.

 

[RJARRRPCGP]

got this board with a 1055T in a Fry's combo in May 2010. Bought a 4x4 GB set of 1600 RAM about a year later. The 1055T never liked running four sticks of RAM at 1600, but it ran them fine at 1333.

I recently bought an FX-8350. For about a week, it was great with running all the RAM at 1600. Then, I noticed the system started only showing 12 GB "okay" at POST (though once in Windows it said "16 GB total, 12 GB available"). I set the RAM back to 1333, but that didn't help. Still 12 GB 99% of boots. Past couple of days, I've had a lot of trouble getting the thing to boot at all, whether from cold boot or suspend-to-RAM. I know my hard drive is beginning to fail, but it was still working fine in operation (I've never gotten any crashes that point to the disk causing them, just noticed re-allocated sector count getting a bit high in SMART).

This afternoon when I tried to reboot into Windows, instead of the usual Windows flag animation, I got the gray ASCII-block progress bar like you get from booting the Windows install CD. It starts at about 95% across, then the system reboots. On about 1-in-5 boots, I'll see "BOOTMGR is corrupted" instead.

GRUB2 is installed on the Windows disk, but my Linux partitions are on another disk, but booting that I'm getting an XZ-decompression failure when loading the initramfs.

Trying to boot from a USB device nearly always hangs in the boot selection menu. Sometimes it gets to the SYSLINUX menu but it hangs before I can type anything.

A couple of times, I've gotten this between the POST screen and the RAID disk listing: View attachment 130900
This is before any operating system stuff even comes in, so definitely pointing at bad SOMETHING in hardware
EDIT: Found that this may relate to the ROM for the controller, but I've not flashed any custom BIOS or anything. If the drive controller is flaking out, that would explain a lot of the issues (but not the loss of RAM)...

Given that I can't boot from anything at this point, I seriously doubt the hard drive is worth worrying about. Loss of usable RAM could be either the motherboard or CPU (or the RAM, but a few weeks ago, memtest86+ said the RAM was fine). Inability to read disks could be motherboard or CPU. Failure to boot from USB could be motherboard or CPU. This is a cheap-ish board, and doesn't have any kind of POST code display (though not sure that would even help, as it usually fails after POST is already done). I don't have any other motherboards, though I still have the 1055T I could put back in. Before I bother with that, any stronger opinions on what the heck is actually wrong here? I'm glad I spent a bunch of money on a laptop right now...

EDIT: Put the 1055T back in. Now I get an infinite reboot loop after the RAID disk list. I can still get into the BIOS (which doesn't show up until after that disk list when disk mode is set to RAID), but that's the only thing that stops it from rebooting right then. It then reboots after selecting the "quit without saving" option which would normally allow the OS to load without restarting. Can boot my Gentoo USB stick fine now, but can't run nearly anything on the installed system because I already recompiled everything to take advantage of AVX, SSE4.2, etc.

Dug around and found a USB drive with the Windows 7 installer on it. bootsect'd it, and it seems to be booting again. Also, it has so far POSTed 16 GB RAM every time. So, question remains: Is the motherboard unable to reliably support the CPU, or is the CPU defective? Either way, why did it work okay for nearly two months (albeit with only partial RAM)? I can understand the board not being able to fully support the CPU, but working for two months, then suddenly degrading just doesn't make sense...

Possibly fractured solder joints on the board.
Over at badcaps.net, we have folks speculating that the new lead free solder is the culprit! (post 2005)

Some of the symptoms may be caused by RAM that's not seated properly.

 

[petteyg359]

So, my Extreme9 board arrived today. I put in the 8350, connected everything else, and am getting occasional insta-reboots with no BSODs or other visible errors. I haven't reinstalled Windows yet, but... I got one BSOD while attempting to install the USB 3.0 drivers (MSI board used Renesas chip, this board uses Etron chip). After it restarted, I tried to install the Intel network driver instead. Halfway through installation, got an insta-reboot. Tried again, got another insta-reboot. Figured it might be Windows needing to be re-installed from scratch with new chipset drivers, so booted into Linux instead, since I could just re-build the kernel sources with Intel network support there. Halfway though kernel compiling, the thing did another insta-reboot. In Windows again, just did the driver install manually without using Intel's setup.exe, and it worked fine.

Right now I am installing the chipset drivers downloaded from ASRock's site. If that doesn't help, I'll do a clean re-install of Windows. Last resort before RMA'ing the CPU is switching back to 520HX power supply.

 

[RGone]

Insta REboots are usually power related. Not always, but a good place to begin. However without a BSOD and the memory dump it is hard to do more than speculate.

 

[Mandrake4565]

Insta REboots are usually power related.

I was having problems when I was really pushing my 8350 and my gtx 580 with insta reboots while using a Seasonic Gold X 650w. When I hooked my Watt meter up to my Pc I was drawing 710 w from the wall, this was after I bought a 1000w unit. Granted this was when I had the Cpu over 5.3 and the Gpu as far as I could push, running benches. It sure sounds possible that the 520 HX may not have the juice to run it. Looking at it, the 12 V rail says 480 watts so it should be good, but if it was on it's way out, It may be the culprit.

 

[petteyg359]

sounds possible that the 520 HX may not have the juice to run it. Looking at it, the 12 V rail says 480 watts so it should be good, but if it was on it's way out, It may be the culprit.

It's on a less-than-a-month-old 760AX right now. On a tangent, pulling out the old motherboard, I noticed the "140W CPU READY!" sticker it came with (despite only having four box things on the power section, vs. the new board's dozen-ish)...

 

[Mandrake4565]

Crap sorry, I didn't read the whole sentence when you wrote "switching back to" the 520, my bad. If you haven't already, maybe you should simplify things till you find the culprit. If your running multiple drives, try disconnecting everything but the main. That's if what you said you were going to do in post # 6 doesn't work.

 

[petteyg359]

So, I updated the BIOS(UEFI?) top 1.40 via its own online update mechanism (board cane with 1.00). I got Windows reinstalled today, using UEFI and everything. It worked fine all day, didn't crash on driver installs for either Intel LAN or USB 3.0, our installing VS2012, or during a few hours of work in VS. Later, got Linux reinstalled, also using UEFI. Just now, Windows has started doing unannounced reboots again. New symptoms of weirdness:


UEFI shell option in boot selection always lists all installed storage devices then freezes with no prompt or anything
Using efibootmgr to add an entry for Linux, everything showed up fine, with entries for both Linux and Windows. Rebooted, and the Linux entry was there, but the Windows entry was gone. Used efibootmgr to add back Windows, and again both entries showed up in efibootmgr's output. Rebooted, and Linux entry was note gone.
bout into Windows to look for an EFI entry editing tool, and start getting random reboots again.


Tomorrow, I'll try the old PSU, but is the brand new board "already" broken? Boot entries aren't supposed to disappear, and the EFI shell is supposed to be a shell, not a static list of drives that requires ctrl-alt-delete to get out of.

EDIT: Same rebooting with the old PSU.

EDIT2: So far so good with the old CPU back in. Too late for NewEgg RMA, but have submitted a request to AMD.

 

[petteyg359]

The replacement CPU came today. System has been working fine with the 1055T for the past week. Put the new 8350 in, and have four no-BSOD restarts already. Don't know what to try next...


Noticed memtest86+ hanging early during test 5 (block move). Upped voltage to 1.7, and it got nearly finished with that test, but hung at 89%. How high is too high on relatively cheap Geil sticks? And why would it, at the same 667/1333 speed, suddenly need more voltage / looser timings on a CPU with a supposedly stronger IMC?

 

[Johan45]

So, I updated the BIOS(UEFI?) top 1.40 via its own online update mechanism (board cane with 1.00). I got Windows reinstalled today, using UEFI and everything. It worked fine all day, didn't crash on driver installs for either Intel LAN or USB 3.0, our installing VS2012, or during a few hours of work in VS. Later, got Linux reinstalled, also using UEFI. Just now, Windows has started doing unannounced reboots again. New symptoms of weirdness:


.

I was having issues of late with similar symptoms. I'm not saying this is your problem but first thing was the Windows install with the EFI partition and trying to run Win7 and vista in dual boot. Right now I have the old style partition but haven't re-installed Vista yet.
My second issue was with which I think may have been the root cause of a lot of my problems was the AMD SATA/AHCI driver.
My machine would always start acting screwy after the Win7 SP1 update. I wiped it and re-installed several times but this last time when installing the chipset driver I omitted the SATA controller driver and everything seems OK at this point.
It just seemed like quite a coincidence that as soon as you install Linux it starts having fits.

 

[petteyg359]

Well, I had successful a memtest86+ 5.00RC1 pass (took about an hour) with the RAM dropped to DDR3-1066 at 1.7v. This RAM worked just fine with the 1055T at 1.5v and 1333 and passed memtest and hours of compiling and gaming and transcoding and everything. I thought the FX-8350 was supposed to have a better IMC... ASRock's site even suggests several sets of similar Geil RAM tested fine on this board.

 

[Johan45]

I've never had trouble with Geil Ram either on any of my AMD's . Have you raised the CPU_NB voltage at all?

 

[Mandrake4565]

Petty as far as memory goes, I was having issues with my 8350 running over 1600 mhz. I found that I had a bad BIOS flash after a long time of messing with it.

 

[petteyg359]

I've never had trouble with Geil Ram either on any of my AMD's . Have you raised the CPU_NB voltage at all?

No, all voltages are at default except for RAM, and right now I have the RAM set to 9-9-9-27-35 at 1066, because I'd rather have it a bit slower and have a working system for a while

The SPD for 1333 is 9-9-9-24-33, and the label for 1600 is 9-9-9-28 (label doesn't specify tRC).
Oddly, there's an SPD entry for 740 (though I've never heard of anything advertised as DDR3-1480) at 10-10-10-27-37. Geil sold a CL10 variant of this RAM, but the label on that says 10-10-10-28, so no idea what's going on with that SPD setting. BIOS ignores it, anyway, and defaults to 1333. All of the SPD entries (and the label) specify 1.5V.

 

[Johan45]

It looks like you have lot's of ram not sure if it's 2x8 or4x4 but I'd set the CPU_NB voltage to 1.25 and then see if the ram will run at it's regular speed. The IMC may be improved but it also juggles 8 cores and all your ram so usually needs a bit of boost for that amount of memory. I don't think the ram needs more than maybe a .05v boost over stock.

 

[petteyg359]

It's 4x4 (says in first post). The default NB is 1.108. Closest options to 1.25 are 1.247 and 1.260. Going to attempt smaller steps, anyway. 13% increase seems overkill.

 

[Johan45]

It will create a bit more heat but it's not harmful in that range. I go up to 1.4 to run at 2700 on the NB. Have you changed the NB freq?