2010

Fri, Mar 19

<16:29 EEST> The recent Linux kernel release 2.6.33.1 is personally newsworthy as it finally fixes the i8xx lid status problem. I noticed this as the upgrade conflicted with the previously applied patch, and things are a little cleaner again.

The new 64-bit Hoo is running fine, and I have finally run the benchmarks for GIMPS on both of my x86* machines. Considering my usual craze for distributed computing statistics, it is funny how this took so long. Interestingly, all of my results are notably faster than the ones listed in the benchmarks page. One obvious explanation would be that it lists the average, and there is always some inherent scattering between different measurements, but mine are around 15% faster. Note that the benchmarks refer to exact CPU models and clock speeds, so it is curious that some machines must be that much slower. Perhaps some lesser operating system is stealing cycles?-)

Sun, Mar 14

<16:46 EEST> What a better day to announce a few technical upgrades than the Π Day! On Thursday my new CPU arrived in the mail, and Hoo is now running at 64 bits with a smaller, quieter SATA laptop HD. More interestingly, during the disassembly I managed to solder a SPDIF header to the audio chip on the motherboard, so I finally have digital audio :)

The CPU upgrade from T2300 to T7200 is not quite as good as expected, in terms of power consumption. A quick estimate from the die temperatures, with the same cooling setup, shows a nearly doubled wattage, or probably least 1.5-fold. Admittedly, the CPUs are almost from the same era, using the same 65-nm process, and this model has a huge L2 cache in addition to the extra execution units of a Core 2. On the other hand, performance has increased at least twofold, and it can still be clocked down to 1000 MHz if desired.