GearCity 1.14 Progress

[Eric.B]

Fixed bug in AI system which would not assign proper reserves to proper production lines.

[Remaggib]

I'm sure you've mentioned it somewhere or another, but will there be an option to turn down the number of AI competitors? Only reason I ask is that by 1913 my wife's computer takes literally 40 seconds to a minute to end turn, and that's on months that aren't quarterly. Feel free to delete this or move it if my post shouldn't be here

[Eric.B]

I'm sure you've mentioned it somewhere or another, but will there be an option to turn down the number of AI competitors? Only reason I ask is that by 1913 my wife's computer takes literally 40 seconds to a minute to end turn, and that's on months that aren't quarterly. Feel free to delete this or move it if my post shouldn't be here

Yes, there should be a range of options. From 10 AI to ~500ish.
In 1.13 there is 200AI of which only about 60 are operating before 1930.

I'm not sure if selectable AI will be ready for 1.14, we'll see.

[Sebastiano1942]

1.14 will be different game with much more options ,can we know approx. when can be done 1.14 ? and how much job you have to do so far.

Thanks

[Eric.B]

1.14 will be different game with much more options ,can we know approx. when can be done 1.14 ? and how much job you have to do so far.

Thanks

1.14 Full Beta (Early Access) will be the full game minus whatever is left for me to do in the suggestion parts of the forums and some artwork.

1.14 Open Beta will have the same restrictions as before (1900-1930 years)

ETA is roughly 1 to 1.5 months from now. I'm a little over halfway done. Working on balancing the economic data and AI in the late game.

[Sarchez]

ETA is roughly 1 to 1.5 months from now. I'm a little over halfway done. Working on balancing the economic data and AI in the late game.

Sleep is overrated. Work work work!
Am really looking forward to the early access, so I can build a Morris Mascot, with a badass V12 engine!

[Eric.B]

It's been a while since I updated so here it goes.

I've been working on balancing issues such as starting money, production line growth rates, costs, employees, etc. As of right now I'm pleased with 1900 and 1980. I'm about to check 1930 and 1950 to see if the interpolation between 1900 and 1980 provides good results for 1930 and 1950...


Also I've done significant improvements and bug fixes to the AI. Found a real nasty bug in the AI system that prevented the AI from selling from reserves. This means the AI was charging much higher prices than it really needed to, to compensate for the over production of vehicles. Needless to say, repairs to this bug has allowed the AI to reduce its prices 200% more...

I've also added another 100 AI to the game and enabled the AI amount selection when you name your company. For slower computers you may want to choose less AI.

Difficulty and amount of AI now effects AI's starting money. The money is to compensate for lack of competition.

[Arakash]

I've also added another 100 AI to the game and enabled the AI amount selection when you name your company. For slower computers you may want to choose less AI.

I dont know if youve mentioned it before, but when you say that slower computers will have problems with high numbers of AI, do you have a rough idea of what kind of computers might have a problem?

If you dont have a clear idea of that, what kind of computer are you testing with? that might give a rough idea.

[Eric.B]

I've also added another 100 AI to the game and enabled the AI amount selection when you name your company. For slower computers you may want to choose less AI.

I dont know if youve mentioned it before, but when you say that slower computers will have problems with high numbers of AI, do you have a rough idea of what kind of computers might have a problem?

If you dont have a clear idea of that, what kind of computer are you testing with? that might give a rough idea.

Single core speeds are what make the difference with Gearcity, as GC is not threaded, and even if it was, it wouldn't help much since most of the AI interaction is via the database. My understanding is that SQLite can only be written to one thread at a time. Which would probably create a bottle neck.


As such, my AMD Phenom 9550 averages roughly 45 seconds a turn with 200 AI and no player data. With 300 AI the same CPU does averages about 2.5 minutes a turn. This is a medium end, 8 year old CPU.


With my AMD FX4300, 200AI averages about 25 seconds per turn and 300AI averages roughly 40 seconds per turn with no player data. This is a medium quality CPU that is about 3 years old...

With less than 200 AI, the turn times are barely noticeable, but the early game is realism is noticeable. With 300AI just about every city has at least 5 companies in it, and ALL vehicle types are filled. This is not the case with less AI, although I expect the companies to be stronger fundamentally with less competition.

As I get the rest of the game settled, I'll go back to optimizing turn speeds. Dumbing down the consumers and limiting the AI choices in car design will probably go a long way into bringing down the turn times.

[Arakash]

Thanks, that cleared it up quite a bit.