My brand new development box (and render target), using a Gigabyte GA-Z97X-UD5H motherboard & Intel i7 4790K processor was showing a strange (read: high) CPU temperature as soon as the processor ramped up to 100%.
When running Prime95, the Corsair Link software (and I backed the following readings up with Speedfan too) was reporting an instant increase to 80 degrees C when using the Blend setting, and a CPU temperature of 100 degrees C when using the Small FFTs (maximum heat) setting. This scared me somewhat! Especially as I was using the Corsair H110i Liquid Cooler – my first time with liquid cooling and at first I thought I’d done something wrong. Like plugged the pump motor into the wrong CPU header. I didn’t do that. No, really I didn’t :-). However…
And the solution:
If you’re running into the same issue, then I’m happy to say that a BIOS upgrade (from F4 to F8) sorted the issue out. CPU temperatures are now reporting at around 26 degrees C @ idle (which is only around 0.5 degree above ambient in our office…) and 65 degrees C when maxing the CPU out using the Blend setting in Prime95. I’m guessing the problem was just a reporting error, rather than the case of the actual CPU temperature hitting 80/100 degrees, but that is a guess.
Incidentally, it’s the first time I’ve used the BIOS on a MB as new as this one, and I gotta say its a dream to use. Mouse & network support, plus plenty of configuration options and a very quick BIOS flash procedure. I like.
Yay! Finally got this box to actually do something (that it was not designed for). I managed to get it booting over PXE. I set up the network boot options on my Smoothwall box, pointing to a TFTP server on my laptop running XP. This didn’t quite work at first – the Neoware managed to get an IP address, and it was also confirming the IP address of the TFTP server, but wouldn’t load any files from it. So I installed the aftp mod for Smoothwall, copied the boot files over, and it worked like a charm. I reckon that although the PXE client was detecting the next-server option from the DHCP server, it wasn’t actually using it, and just tried to access a TFTP server on the DHCP server’s address – this is probably due to a fault in the firmware for the RTL8139 on board ethernet chipset, although I wouldn’t like to say for certain.
Now all I have to work out is why the Debian Etch installer is failing when its trying to partition the hard drive. I thought it was the drive itself at first (its untested) but it fails at the same place with another drive.