Threaded Mode 2023 May FBS Update!
Added 2023-05-02 04:43:50 +0000 UTCGreetings Everyone!
As you may have heard, we launched GearCity: 2nd Gear last month. The launch was a little rocky, some features I don't use went untested, resulting in some instability. I resolved those fairly quickly, but additional fixes to smaller issues caused some larger issues. In any event, the expansion stabilized now. I haven't heard much positive or negative about the features, but fundraising went up slightly, so that's a good sign that folks were largely happy with it.
Fund Raising / Milestone
Speaking of fundraising, another good month for us. We brought in $556.07 via Patreon and $145 on Paypal. This brings us to $1542.77 of our $3500 goal. At the current rate, we should reach that in 3-4 months. As a reminder, the next Milestone will be art-only. So voting will be for new or redone vehicle types in the new designer system. We'll also hold a vote for any remaining funds after you select a new vehicle type.
Because of the number of uncast votes, going forward, art bounties will be voted on before we do non-art bounties. Fully funded bounties (with more votes than bounty amount) will be approved from largest to smallest while milestone funds are available. I'll go into more detail as we approach Milestone #2 funding.
Over the past year, I updated the website milestone funds as they came in. This is changing because we now charge new Patreon when they sign up. Instead, I will update the website tallies at the start of the month, when I process voting points. (Now.)
GearCity
On the GearCity front, I've now switched to maintenance mode. I will now spend my Mondays (3-6 hours) on GearCity-related tasks. The first order of business is GearCity Classic v2.0.0.9. Fixes made to this build will also make their way into GearCity: 2nd Gear v2.1.0.1, along with bug fixes for 2.1.0.0.
AeroMogul
On the AeroMogul side of things, I've only been working full-time on it again for the last 2 days. I'm pleased to report that we've taken some steps ahead. The last time I reported on AM's progress, I was working on passenger demand optimization. I had worked the code down from 40GB Ram and 4 minutes processing time to 4GB and 4 seconds (source) for a single demand variable utilizing 1,440,000 data points. I'm pleased to report that I have made some architectural improvements to work this down to 4GB for 4 sets of passenger demand (one for each class) at 10 seconds processing time using 1,800,000 data points. That's 4 times the information generated from 25% more data at only a 2.5x increase in process time.
While that sounds like a step back in speed, it is the full demand numbers we need to calculate passenger traffic. In the previous system, this would eat 16GB of ram. Unlike the previous system, this new design scales with the number of airports. So, when I remove some airports (Helipads, private strips, military bases, etc.) the numbers and processing time will reduce linearly. It also means when you play with fewer airports, the game will use less ram. I can also reduce these numbers with our datasets. I may create data for low-end, medium-end, and high-end machines. I would consider this current dataset as medium-end. There are also a few more optimizations I can do, such as changing the hashmap implementations, threading, and using SIMD talked about previously. But I will do those when needed.
I also implemented an airport service area system in AeroMogul. The airports use this servicing area to draw from the passenger demand data. I will have to play with these ranges in the future, but the system is working.
Currently, I'm working on making the demand data a little more realistic with the few data points I have. (I will be using a lot more data to generate passenger demand in the future.) When I am satisfied, I will switch to AI and implement basic AI aircraft procurement and route creation. At that point, I can start working on the routing system and passenger selection/allocation, the game's backbone regarding gameplay, realism, and performance.
Hopefully, I have some good stuff to report back in 2 weeks. 
Until then!
Comments
Unused votes roll over. We only consume votes when cast on a bounty during voting. So if you're not interested in any vehicle types, I suggest hanging on to your votes and using them when there are funds for programming bounties. Likely, what I will do is a "tick" / "tock" approach. Art Bounties for milestone #2, then any bounties for #3, then Art Bounties for #4, and so forth. But we'll see. I had to skip art bounties for Milestone #1 because of the war in Ukraine, so we must catch it up with the rest of the system. That said, if enough people are not interested in updated vehicle artwork and a new car designer, I can remove that requirement, but on the flip side, I can't keep doing 16 bounty milestones with 12 discounted. The FBS is primarily designed to improve the artwork of the game (half the bounties are art related). It kinda only works on my end if the artwork takes up a good portion of the milestones funds until the discounted items are cleared out. Otherwise, it over loads me with work with all the discounted stuff. That's why I think the "tick" "tock" approach is probably the best. On "ticks," we can do a vehicle art bounty and a few hundred in programming bounties. On a "tock," I can keep us at around $2000/$3000 in milestone funds until we clear out a few cheaper bounties before raising the milestone goals. (Mainly trying to avoid having 40+ bounties funded at once.) As for full voted bounties, the other issue we have is that there are now more voting points out there than funds. So theoretically more bounties can be fully voted on than the funds in the milestone, but I already did the previous milestone amount of work. Because of that, full voted bounties will go in order from highest bounty to least while there are funds available.
Eric Jones
2023-05-02 05:11:50 +0000 UTCUm, will it be possible to not vote on anything and role over the unused votes to the next milestone? I am paying quite a lot of money for the sole purpose of being able to eventually single-handidly vote in at least one bounty that makes running an EV company nicer, and there aren't any art bounties that I know of that work well for this purpose, and furthermore they are so expensive that even if I did find one, winning seems rather unlikely.
happysmash27
2023-05-02 05:00:06 +0000 UTC
