Flash player sometimes causes XP to reboot

[tain]

I have an nLited XP SP3 machine that runs rock steady almost all the time. The only time it crashes is when playing Flash content, and even then only occasionally. Flash is the key component of the crashing. It doesn't matter if I use IE, FF3, or Chrome to invoke the Flash player. I've tried Flash versions 9 and 10 (using Adobe's uninstall utility) with no change in results.

Unfortunately the crash goes straight into a reboot. No BSOD is given and no events are logged. So I'm having trouble nailing down the exact issue. Not sure if my video drivers could be involved in this, but I'm running a GeForce 7600 GS with nVidia's current drivers.

Any ideas on what is causing this? Or how I can get some detailed crash information to diagnose the issue?

[DigeratiPrime]

My guess: problem with the cpu. A usermode app cannot hard reset the system, closest thing is a BSOD. Try running Prime95 a little and seeing if you can crash the system, also monitor the temps, remotely if you can. If it was RAM I would expect a BSOD.

Some other apps:

Core Damage http://damage.vigilantesoftware.com/

OCCT http://www.ocbase.com/perestroika_en/index.php

[tain]

We have a winner! Or a loser, is more like it

Prime95's default stress test ("Blend") tests all 8 of the Core i7's logical cores and almost instantly reboots the machine. The other two modes (Small and Large FFTs) do not crash the machine, or at least not when run for a few moments.

Local machine temperature monitoring does indicate that it is a heat issue.

I've never encountered a problem like this one. What does it mean, exactly? What can be done to remedy it?

[DigeratiPrime]

Next steps:

Check the CPU voltage to make sure it is correct before doing anything else.

An infrared thermometer can tell you if the heatsink is getting hot, without having to touch it.

If the heatsink is not absorbing heat, check the thermal adhesive between the cpu and heatsink.

If the heatsink gets too hot, then upgrade the heatsink or fan(s).

If the problem is unique to that CPU then replace it.

[tain]

CPU voltage is fine at 1.21v (I don't overclock).

No infrared device, but HWMon reports idle temps around 41C per core. The crash happens from an idle state, so heat does not appear to be an issue.

Memtest86+ reports that the 3GB of RAM is all fine.

Please help me fix this without buying a new CPU

[tain]

Could certain BIOS settings perhaps change this behavior? VT, perhaps?

[DigeratiPrime]

Unfortunately I think this is a problem with the cpu, I looked around a little and I see people running Prime95 on the Intel i7 without problems, also hard reboots with Prime95 are indicative of cpu problems. I am moving this to the hardware forum, to see if we can get some more ideas. Other possibilities could be the power supply failing, or the chipset faulting.

[puntoMX]

I would blame the BIOS over anything else. Try to set the Vcore to 1.3v just to see if the CPU isn't getting enough juice. Also, about what motherboard are we talking about here? Did you check for a BIOS update and its release notes? I've seen none defective CPUs lately so the change that the CPU is defective is… I would say… 1 in 10.000 or so, but that's not out of the box.

[tain]

Intel says that 1.3v is within the chip's tolerance so I'll give that a shot.

The mobo is an Asus P6T Deluxe V2. The BIOS is not the most recent, but I have reviewed the changelogs and there is no reason to upgrade.

[puntoMX]

The BIOS is not the most recent, but I have reviewed the changelogs and there is no reason to upgrade.

If it's any thing under Version 1003, you need the update. I would update to the latest version first any way on any "newer" motherboard.

[tain]

I bumped up my CPU voltage to 1.3000 in the BIOS and it reads as 1.2888 in monitoring software now. After a fresh boot, I ran Prime95 again. It still dropped into a reboot, but did seem to survive for perhaps 2-3 seconds longer than before.

Where did you get that BIOS version number? I'm currently running 0502 from AMI, dated 5/5/09. The latest version on the Asus product page (under downloads) is 0504, dated 5/25/09. I don't know where you got a version 1003 from.

Could disabling any CPU features in the BIOS make a difference? Features such as VT, VT-d and Execute Disable Bit while SpeedStep is disabled.

[puntoMX]

Where did you get that BIOS version number? I'm currently running 0502 from AMI, dated 5/5/09. The latest version on the Asus product page (under downloads) is 0504, dated 5/25/09. I don't know where you got a version 1003 from.

My fault, I see you have the "v2" so yes, 504 is the latest version.

It should work fine on BIOS defaults, so if you checked your RAM with memTEST or something and you updated to the latest drivers, I would blame indeed a faulty CPU... But like I said, the change is small.

[tain]

Well that is bad news

Just for giggles, I loaded up another boot disk with XP and mostly default drivers. I get the same rebooting behavior upon launching Prime95/blend. I fully tested the RAM a few posts back so I suppose this just confirms the bad news.

This sucks.

Pretty please make it be fixed without acquiring new hardware

[tain]

After much coordination and jumping through hoops with my vendor's tech support I finally convinced them to send me a new chip. All fixed now! The new chip runs great and is clear evidence that my old chip was defective.

Thanks guys! Chalk up another win for MSFN tech support

Here are some notes from my testing of the bad chip in case it helps anyone else:

I changed the timing from 9-9-9-24 to 8-8-8-24 as you requested. The system still goes into a hard reboot almost instantly after starting Prime95 in the default Blend mode.

Changing the memory speed to 1333 caused the system, after saving/exiting BIOS, to do a hard reset...the fans stopped for a bit, then it picked back up. This appears to be normal for whenever you reset the memory speed. Anyway, this didn't fix the problem.

Then I tried swapping the RAM modules and that is when it got weirder. Removing two of the modules caused the BIOS to give me an error to the tune of "Ovcerclocking failed...hit F? to load defaults". Loading the defaults did seem to help. The BIOS set everything to AUTO and it picked 1066 as the memory speed. This let Prime95/Blend run for about a minute or two before doing the usual hard reboot. So it may have helped a bit, but the result is the same.

I'm leaving everything on AUTO for now since it may increase system stability.

This still looks like a CPU issue to me.

Ran lots of tests today. This is mostly about swapping RAM and running Prime95 in Blend mode.

BIOS at defaults (AUTO):

-Each stick separately seemed to run stable

-All three sticks together seemed to run stable

-----(previously reported crash unexplained)

BIOS with RAM at full speed (9-9-9-24/1.64/1600)

-Each stick of RAM crashed almost instantly

-All three sticks crashed almost instantly

An interesting thing happened on two of the Prime95 tests. I twice saw Worker#5(logical core #4 of 7) report a rounding error just prior to the system crashing. So I went into the BIOS and found a setting for which CPU cores to enable: All, 1 or 2. When I changed it to 2, the system still crashes almost instantly. But when I run it with just one core enabled it seems to run fine, even with the RAM at full speed.

Right now I'm running RAM at full speed with just the first physical core enabled (2 logical cores). Not sure when you will get back to work but I'll try running this configuration for a while to see how it holds up.

What do you think? This still smells of a bad CPU to me.

[cluberti]

At this point, I'd be harassing the OEM to RMA it for a new chip. If it's DOA like that, it's not worth risking a part that wouldn't last as long as a "good" part even if you do get it "working".