Stardate 20020829.1725

(Captain's log): The new computer, which I ordered in May, received in July, spent three weeks ordering upgrade components for, and finally gave to a local dealer so they would install some of the pieces for me (and order yet more upgrade components for me) just came home from the shop, and it is finally fully configured the way I originally wanted it to be last May.

So you can expect posting around here to be rather light for the next few days.

The raw numbers: Intel 860-based motherboard with two 2.4 GHz Pentium 4 Xeon processors (Northwood) and 1 GB of 800 MHz RDRAM. 160 MHz SCSI RAID controller with four drives. Two 18G drives in mirrored mode == 18G. Two 70G drives in mirrored mode == 70G. And a lot of other goodies. It has a Radeon 7000 in it, but there's an All-in-wonder 8500 waiting to be used once I've got everything else working. (I'm not in any particular hurry to put that in, however.)

It currently has Win2K installed on it, from before the most recent upgrade, and to begin with I'm going to let that installation do all the upgrading and loading of new drivers that it wants to, in part so as to identify everything required and have it all ready. Then I'm going to wipe everything and repartition all the drives and install from scratch. Partly that's because right now Win2K is on a 16G partition, and when I'm done it will be on a 35G partition, and partly that's because I think that it's more likely to be reliable if the OS installs from scratch. (Sometimes upgrades leave parts of the older code around and the result can be odd.)

The very first question, however, is whether the aftermarket CPU I bought is actually the same stepping as the one HP shipped, or close enough so that SMP works reliably. I'm assuming that if there are "issues" then they'll manifest within the first billion cycles of SMP operation (i.e. during the OS boot process). SMP is low-level enough in the system that any problems are likely to show up really soon. But the steppings might be the same in which case there should be no problem at all, and I'll be able to find that out from the BIOS. (Which will also be a good preliminary SMP test, since the BIOS can't get the steppings from both CPUs without running them both.)

Win2K almost certainly won't use the second CPU until I tell it to, and it's virtually certain it will have to load some code in order to do so.

I will also need to do a bit of research about hyperthreading. (Google, here I come!) The guy at the store said that this system could use it, and that with the right software it could lead to a substantial performance increase: 20-30% in Linux, probably quite a lot less in Win2K. He also said that there wasn't any drawback to it, so there's no reason not to use it. It's something I can enable in the BIOS, so I'll experiment with it in the current copy of Win2K and make a choice. Apparently to the OS it looks as if each CPU is actually two, which means that the OS would report that it had four processors in my dual system.

Once I've finished the clean OS installation, the fun part starts: installing all the apps that I can't live without, and then downloading patches to bring them all up to date. The last time I did all this it took about three days to get the main stuff in and everything set up the way I want it to be set up.

All of which means that I'm unlikely to be thinking profound thoughts (or even prosaic thoughts) about the human condition for a while.