What Will Be Done

My decision as it now stands: I'm stayin' the course. I still feel I need to commit to more rigorous improvements in my art, and I need to do it soon.

Since there have been a few recommendations along the technical I've received from strangers, I want to make something clear: the SGI is my last computer. I intend to use it until I no longer can upgrade/repair it, something to be treated as a perfect brush.

Part of the reason is that this is what I know. The larger part is that I simply don't want to be stuck using and maintaining mediocrity for the the rest of my professional life. Learning a new OS is, from my perspective, a waste of time. The old hardware and software still works, and works as if it were new, and I have never had an occasion to throw away a perfectly usable brush.

(Then again, no brush has ever frustrated me like a computer manages...)

So. I'm back.

The Failed Reunification of Gondwanaland

Yesterday I assembled the two nodes on my Origin 200, and discovered something annoying. One of them was very dead. Since the logic appears to be getting power (even if it doesn't completely power up), and the cable is probably not suspect (each node independently confirms that one node runs perfectly fine), I'm ruling out everything but the logic carrier, which appears to be hosed.

This leaves me with a dual-processor Origin 200, and a fifty-pound doorstop for parts. Not great, but not a total loss, either. And I think it just sort of emphasizes the necessity of obtaining more current iron for an IRIX-based renderfarm. The Origin200 is 1996 vintage, and, on reflection, none of my IRIX machines break the 2001 mark.

Why don't I go to the PC? It's not a jump I'm willing to make. It will involve learning a new OS (Linux), and I feel I sacrifice far too much of my creative time to technical endeavors as it is. Migrating over to the PC therefore makes no sense to me.

On Computer Renderfarms

My recent animation experience has taught me that I need every scrap of CPU time I can get my hands on, and it needs to be MIPS-based (it has to do with the different ways different CPU vendors do floating-point calculations--believe me, there is a reason there are no mixed-cpu render farms, where some CPUs are Intel, some are IBM/Motorola, and some are Sun SPARC--those differences can add up to peculiar flicker in procedural textures ).

So even as I relax this weekend, I am taking my Origin 200 server, Gondwanaland apart from the state of storage it's essentially been in for the past two or three years. At this point, I have no idea whether it will even power on. But I can't say no to 4x180 MHz R10000s. This system is positively ancient, certainly no more recent than 1997.

But it is still useful.

Additional adds: an R4400 Indigo² (Carcharodon), another Octane and possibly an R5000-based Indy. If my living situation were different, I'm certain the electricity costs alone would not be worthwhile. But it might be interesting to see if systems from 1994 can still be used in this environment.

If I'm successful, I may write about the composition of the network later.

CG History

The computer (I wish I knew who it was who originally said) is a stupid pencil with millions of colors. I'm rather late to the game in even appreciating the value of CG to filmmaking (that is, it took Jurassic Park to get my eyes clawed open).

After that, I was all about those awesome tools they used (I acquired one first in 1996; an Indigo R4000.) I still maintain and use a number of SGI machines: Octanes, an Indigo R3000, an Indy (which has an amazingly fast boot time), and an Indigo². (All SGI IRIX machines are named after animals like Sarcosuchus_imperator, Paradoxides, Helicoprion, Carcharodon; the Macs in my employ must make do with names like Tsuburaya, Miyazaki, and Harryhausen.) Modern SGIs make use of conventional CPUs and are Linux, rather than IRIX-based, so I tend to pay far less attention to such things (and there is nothing more demotivating to the acquisition of new hardware like the feeling you're getting something that can be found in a common PeeCee). I now concentrate on the systems I have (one of the Octanes and my G4 PowerBook) in order to produce. I have a long way to go. But, CG has come a longer way, so I don't feel so bad.

Some good stuff (good == historical):
Pencil Test. Made entirely on (several) Mac II computers. (For some reason, I'm having trouble locating a similar project done entirely on the PeeCee. Anyone know of one?)