Long-term Memory

I’m again having problems with my new MacBook Pro (the late 2008 unibody one). The problem is most easily visible when BootCamping into Windows and running whatever 3D application (I use 3DMark06). Within 5 minutes, the machine crashes. After some trail-and-error, I found out that it appears to be a thermal problem.

Normally, when the load goes up, the fans kick in to keep the temperature down. Mine don’t… When stressing my CPU under MacOS, it happily goes up to 90ºC with the fans still idling at their standard 2000rpm. Needless to say, a bit of GPU-load increases the temperature beyond the safety-thresholds.

Resetting the SMC solves this problem only partially. On the first boot with a clean SMC, the same CPU-test brings the CPU to 70ºC with fans blazing at 4750rpm, as it should be. Once the machine goes through a sleep-wake cycle, the same test brings it straight up to 90ºC.

Just when I was about to call Apple Care, a software update pops in: SMC Firmware Update 1.2, looked promising, but the problem persisted… I’ll call Apple Care tomorrow.

Update 2008-12-16, one day later

Called Apple Care and described the problem. According to the guy I had on the phone, they don’t know of any similar problem… The Internet thinks otherwise…

Anyway, they referred me to an Apple Authorized Service Provider with a case number. To be continued…

Update 2008-12-19

Just got a call from the repair shop. I’ll try to get this confirmed, but what I understood from the call was:

• Yes, when we do the same tests (3DMark06), it crashes here as well, although not every single time
• When we do the MacOS-tests, we also see the CPU going to 90ºC; this is not a problem
• We checked the complete machine, nothing is wrong with the hardware

What bothers me the most is that the they consider this almost “expected behaviour”…

Update 2008-12-22

Went to get my MacBook Pro back. I asked to talk to the guy who did the analysis:

• The problem with 3DMark06 is reproducible, but not supported. He claims that the tests usually crashes at the same point in the test (when the fireflies enter the tree), independent on the startup temperature
• He was not able to reproduce the problem on MacOS. I asked if I could try to reproduce it right away, with success: First boot (unclean SMC) brought the CPUtemp up to 88ºC with fans at 2000rpm. Next I tried to clear the SMC, but got the same result (75ºC with 2000rpm). A second try at clearing the SMC was successful: I got 76ºC with fans increasing beyond 3000rpm
• The technician agreed that this is “strange but unknown”. He’ll contact me when Apple discloses more info on this
• He tested the hardware thoroughly with the special Apple-Authorized-Service-Provider-only test tools and found nothing wrong

So the conclusion is: My Mac is behaving strange, but there is nothing wrong with it.

Update 2008-12-22, later that day

I’ll have to drop the Mac-test from the above case. I just ran the CPUtest a bit longer. After a few minutes at 95ºC, the fans started to kick in, although not as quickly as with a clean SMC.

Update 2008-12-23

Tried to pin down the Windows part of this issue, without success. I updated my Windows to include SP3 and all patches and checked for apple-updates as well. The results:

• Fresh boot into Windows; idle for a few minutes; Firefly Forest test only: crashed after 31″
• Reboot; re-ran the Firefly test: crash after 24″
• Reboot; Firefly test: crash after 11″
• Reboot; Firefly test: crash at 12″
• Reboot; Firefly test: crash at 19″

The other tests seem to run great, even the Firefly Forest runs sometimes:

• Fresh boot; Return To Proxycon, Canyon Flight and Deep Freeze, 5 times in a row: OK
• Firefly test: OK
• Firefly test: OK
• Firefly test: crash at 60″
• Reboot; Firefly test: crash at 28″

By “crashes” I mean: my external screen goes black, then into sleep; Alt-Ctrl-Del doesn’t work anymore.

A friend of mine tried to run the standard setup (all 4 tests, sequentially): on his identical MacBook Pro (appart from a 7200 vs 5400rpm hard disk) it ran fine 4/4 tries.

Update 2009-01-02

Did some more comparison tests by swapping the hard disks between my MBP and the one mentioned above: both computers with both drives crashed on the 3dMark06 tests; not always, not at the same spot every time. My MBP with his HDD even crashed at the Proxycon test once.

I’m asuming that this is a general hardware/driver problem, maybe temperature related. I hope Apple brings out an updated firmware/driver for whatever fixes this issue…

Update 2009-01-04

Just played Flight Simulator X for more than an hour without a glitch, although the fans were doing overtime.