Science/Engineering/Technical forums

I'm Rand Huso and I'm addicted to science. (everybody say "Hello Rand")

I did some degrees in Engineering-Physics (Electrical), Mathematics, Meteorology, and Numerical Analysis back in the '70s and '80s. I started writing software in '72 and it's kept me off the streets. I'm working at GNS Science in Avalon, New Zealand. I live aboard a small yacht in the Wellington Harbour during the week and commute to home on the weekends, or as often as I can. I'm trying to write my own climate forecast model in my spare time.

I chose to live here because I believe New Zealand has the best chance of any country in surviving the climate change and the post-peak-oil economic environment. Besides, I like yachting, and this is one of the best places in the world for that.

Hello, Rand!
Yes, JohnF has been showing us some really nice pix of your area of the world, and frankly, were I ever to leave this place -- that's where I would go, for a lot of the reasons you mention, as well as others (your government probably makes more sense than mine most of the time, not that it's difficult, but quite a few are even worse then USA these days). I now have to look at my pix from here in the climax part of the year to not be so jealous. It's pretty nice here too, if you're far from the crowds, and I certainly am, but rats, there's no decent place within 500 miles to do boats...I don't count big lakes. Being far from the big water, we have 4 real distinct seasons here, and to survive here you have to have a taste for "weather" in general and not be too hooked on any one kind. Spring is pretty orgasmic though, after a usually hard winter, and once you go out, you don't want to come back indoors for a long time.

Enjoy the forums, we look forward to seeing you post on the things here, or if I left out categories -- new things (I'll make new categories as interest shows we want them).

Doug
Rand is an ex yank
I worked with him many years ago when he came to NZ to work with me and others designing a portable cellular base station. He then returned to the states around 2000 coming back here from Seattle Nov 2008
and yes I got him the job here

Yep. But I think I had something to do with landing the job too.
I was born in Minnesota, or so I'm told. Raised just east of Seattle. Spent 3 years in Italy. Visited over 30 countries, read something about all the others.
If you look at the temperature changes here: http://earthobservatory.nasa.gov/Featur ... a-features you'll notice that NZ is still relatively stable. My decision to live here was based primarily on concerns about sustainability, not the least of which was population. That's potentially depressing. What's not is that we've just run one of my programs on a 2000 node 8 core per node computer in India to see if it saturates the I/O. No verdict yet because they don't have their cluster configured properly.

Being south of the Mason-Dixon line here, a lot of my neighbors wouldn't like to be classified as "yanks" -- but I'm not one who cares about such things. Even though my middle name was that of a famous confederate general in that war (and the first from another later war general, easy to guess which...).

Rand, have you played with any of the Nvidia stuff yet? I've got some of the hardware (tesla boards), but haven't had time to really wring it out. Looks pretty good if you know your stuff, though I understood that at least at one time, the Cuda environment made a host thread spin till a task was done (per task in the Nvidia stuff), which sounds pretty dumb. Don't they know what interrupts are for?
Of course, doing anything much in ring 0 of windoze isn't so much fun as it used to be...I'm a linux guy now. I'll get around to that someday. I was wanting to use it for MLFN neural net training for my stock trading....once I get some other stuff going, like parsing the dumb XML interface for realtime data my broker gives me.

Nvidia, Cuda, OpenCL, all not quite there. We tried porting a migration algorithm to use on video cards but although the amount of memory is nice, the type of memory isn't. I recall there were limits on the sharing of memory between groups of threads. Anyway, I've been doing MPI (MPICH2) for a year now and wrote a nice layer / framework for registering callback events when the appropriate type of packed is received. It allows me to Pull work to the nodes instead of Pushing work, which keeps the CPUs very near 100% on all the worker nodes. I've also integrated OpenMP into the mix as well. We're using this framework for a second application now to do another algorithm usually just called SRME (Surface-Related Multiple Elimination). The previous was 3D Kirchoff Time Migration.
I'll be out for a couple weeks. Happy Holidays, everyone!