algorithmic modeling for Rhino
First and very much foremost: thank you, once again, Wallacei team, for the amazing plug-in. Your hard work is much appreciated, and I can't wait to take 2.5 for a spin later on today.
Having said that, I set up an optimisation on Friday and came back yesterday to find 50,000 geometries waiting for me (which is obviously magnificent), but working my way through the phenotypes is a very, very slow process. This is no doubt due to my less-than-perfect university computer set up, and so I'm making a bit of noise to get a computer with better specs.
I appreciate this is perhaps not an entirely Wallacei-based question, and I understand it all depends on how one sets up the definition, how many objectives one solves for and so on, but since files with tens of thousands (or more) geometry iterations tend to become fairly heavy, I'll ask it here anyway: what, in your experience, would typically be a good-enough hardware set-up to run Wallacei comfortably? What specifically should one focus on and prioritise when specifying the hardware?
And, as a follow-up question, over and above triple checking the definition for ways to reduce the computational load and following the other instructions on the cheat sheet, have you encountered any other tips and tricks along the way that might speed things up? Checking the Draw Parallel Coordinate Plot for my 50,000 iterations just now took about ten minutes...
sheesh, 10 minutes to draw the pcp for 50,000 solutions is way too long - definitely in need of a serious hardware upgrade!
But, first things first - Wallacei doesn't save phenotypes, only genotypes - and so regardless of how large your population size is, the grasshopper definition should stay relatively light.. ofcourse if you are inputting geometry into the 'data' input of the Wallacei component then thats a different story (since you read the cheat sheet and all Im guessing you aren't doing that so I won't get into it).
So to give you a better answer, which parts are slowing down your machine? Within Wallacei, drawing the PCP in tab 3 for a large population size (such as yours) should take around 30 seconds, but besides that, what other parts are you finding slow?
Thanks for the reptile-rapid reply (as always). I'm really impressed by the dedication you show in responding so quickly and helpfully.
I'm sort of reluctant to take up your precious time with troubleshooting my computer, which is clearly, as you say, in need of a serious upgrade.
But to answer your question: of course Wallacei doesn't save the phenotypes, and the files aren't actually that big (around 3Mb for the 50,000 iterations definition). I'm not tracking any geometry as data for sure. But on this machine, almost every action becomes slow once I create a lot of genotypes and start playing around with them. In your (excellent) series of video tutorials, for instance, the solver starts trawling through solutions (visually, on the screen) almost immediately, whereas for me, it usually takes a generation before WallaceiX begins to draw antyhing. Tabs 2 and 3 are dead slow, in particular the aforementioned PCP plot. Haven't tried any clustering yet, but will later (now with 2.5!).
I'm also getting some weird results that I turned into a separate post, which leads me to believe there might be something in my definition that is messing things up. Perhaps I'm just making a silly mistake (or three) along the way.
For sure, unless silly mistakes are to blame, these are issues to do with a useless computer and not with Wallacei. My question was more of a general one: if/when I manage to talk the powers that be into giving me a new machine, is there anything in particular I should make sure to specify in order to make things run smoothly?
Just exported three generations (150 solutions). This took around 11 minutes.