(March 20, 2013 at 2:25 pm)downbeatplumb Wrote: Is this the talk in Klingon thread?

I used to know computers back when a 386 was the bees knees. I guess things have moved on a tad.

nah... they're still called i386

(March 20, 2013 at 2:43 pm)pocaracas Wrote:

(March 20, 2013 at 2:25 pm)downbeatplumb Wrote: Is this the talk in Klingon thread?

I used to know computers back when a 386 was the bees knees. I guess things have moved on a tad.

nah... they're still called i386

They weren't back then.

(March 20, 2013 at 2:48 pm)Cthulhu Dreaming Wrote:

(March 20, 2013 at 2:43 pm)pocaracas Wrote: nah... they're still called i386

They weren't back then.

To clarify.

http://en.wikipedia.org/wiki/PC-MOS/386

when to load a disk you had to go onto MS DOS and type lots of gibberish.

(March 20, 2013 at 3:01 pm)downbeatplumb Wrote:

(March 20, 2013 at 2:48 pm)Cthulhu Dreaming Wrote: They weren't back then.

To clarify.

http://en.wikipedia.org/wiki/PC-MOS/386

when to load a disk you had to go onto MS DOS and type lots of gibberish.

I was referring to i386 - back in the day, we didn't need no stinkin' i, it was just the good old 80386. Then Intel discovered they couldn't trademark numbers.

Ha! Silly mortals and their printer problems. Try using RT Linux on a neural net built from 5 PXA255 based PC104 boards to do real-time video processing and target acquisition/designation. Can you say write your own damn drivers?

And no… I just dealt with the hardware. The 40 lb. heads wrote the intelligent agent based code that made it work. Worked pretty good too until we started moving the camera. Once we threw that into the picture our artificial fly brain had a mental break down.

Neural nets handle repetitive stimuli rather well and panic at undefined stimuli. Makes them great learning systems for a highly limited set of input.

Can't imagine target acquisition ever working well unless the data is heavily processed into a mostly tamped down state with few major changes.

Else it might go crazy.

---

Would enjoy hearing more though popeyespappy. Any papers by chance?

Everything got reduced to hexagons. It couldn't tell the difference between friend and foe, but it did get pretty good at distinguishing between people and trees. The thermal imaging helped there. It also ate up the lion's share of our $75K budget.

ETA: Sorry. Everything was reduced to polygons not hexagons.

I probably shouldn't have, but Newegg was throwing in 8 GB of free DDR3, so I bought me some new guts. (My old box is 5 years old, and with only 4 GB of DDR2, and no VT-x bit, it would have just been throwing good money after bad.)

This will keep me for another 5 years. Not sure what I'm gonna do with the old box. Either an HTPC, or see if I can get $100 for it.




(My friend's on irc were trying to get me to bump for an i5. But I'd already bumped one CPU class from an Ivy G2020, so I figured I was as high as I needed to go. [I have very modest processor requirements.])

Go for an i7. If you're going to update, get a real good cpu.

The majority of programs are still cpu-limited. Furthermore, virtualization is pretty processor intensive if you want to do rude things like trying to create paravirtualizing partial gaming rigs.

---

And we're seeing an increasing amount of multicore friendly apps.

At a bare minimum, with more cores you can constrain apps to occupy their own core and not worry too much about the scheduler fucking things up.

I don't game or do anything computationally intensive. I spend most of my time in a web browser, listening to winamp.