3: 2 needs what?!
1: 60'000 hours of compute time, and I think he(?) has exhausted all binary optimizations and won't be able to bring down the computations to even 30'000, let alone 4000.
3: And this radical transformation just to see why his idea won't work?
1: It may work ...
*looking at each other*
1: ... alright, it won't work, and he knows it too. That is why he is not pressing the issue, but 2 can't let it go, and we can't try realtime on every suspect slice, can we? The results are not compatible to be filling the tables immediately. They still need post processing, and, it is still about 60'000 hours for a complete set with no chance of optimization if we go that route.
3: 2 needs to let this go, man(?), ...
1: I know. He knows too. Maybe in time he will. .. In comparison Zeah's hunches are better.
3: Zeah's ideas are better, yes! She just got too many of those! Too much work to get a handful of semistable setups!
1: Yeah, Friday was yet another disaster.
3: I heard.
1: If I was her, I would have given up on Friday's expansion sets. You got to admire her persistence.
3: You mean stubbornness?
1 *giggles*
3: How is your own project going?
1: Stuck. I keep going over and over all components. They all work independently. Together, though, they don't produce the I/O Dough asked for.
3: Don't give up. You will figure it out.
1: Yeah. In one of two ways, magic, or donothing. This is why I sympathize with 2, and at least he has somewhere to release his frustration.
3: uh, uh, 60'000 hours to all of us.
1: And he is on the explosive exponential path. Expect the next step to need 200'000 hours.
3: He can optimize the next step, can he?
1: To what useful degree? How many steps needed? Next, rebuilding the whole PiPrime set from scratch?
(scene may continue if needed, or be expanded to include sponsor content and life/romance/death for a more complete picture)
No comments:
Post a Comment