Regarding the possible heating problem. We, Excito, have tried to reproduce the situation that might have caused the situation. I.e. a heavy load produced by compiling large "projects", gcc in this case.
This far we have been unable to even get our test environments near any critical temperatures. We have recompiled the kernel, GCC 4.5.1 with all languages enabled, several times. And a recompile of GCC on B3 takes several hours with a high load so this should be a good candidate for stress testing.
During this we sample CPU and disk temperature, this done in normal
room temperature here ~21C. While running the above tests the disk never
exceeded 48C and the CPU never rose above ~81C. The 81C on the CPU was
measured on the capsule and should correspond to a ~100C die temperature
which is well within working temperature according to Marvell specs.
So with this in mind we do think that it is some other failure with Michaels unit such as a faulty disk, power supply or perhaps a bad mainboard. This is however just specultion since we havn't looked at that specific unit.
We are talking to Michael about this and asked him to do some tests to try determine what might have failed. After this we will most likely exchange the unit to get a closer look at it.
But to summarize, i would say that it is still very much more likely that the unit in question failed for other reasons than overheating.
I've been meaning to ask about your testing methodology, so your post came right on time.
Thanks for the updates.
Tags for this Thread