Zerg Build Order optimizer. - Page 52

Post a Reply

Prev 1 50 51 52 53 54 58 Next

Subversion

South Africa3627 Posts

November 11 2010 00:06 GMT

#1021

Is the Zerg one getting a GUI as well?

Or wait, does it have one already hehe.

Dionyseus

United States2068 Posts

November 11 2010 00:27 GMT

#1022

On November 11 2010 09:06 Subversion wrote:
Is the Zerg one getting a GUI as well?

Or wait, does it have one already hehe.

The zerg version already has one.

Jaeger

United States1150 Posts

November 11 2010 00:29 GMT

#1023

You could make it more accurate by pulling numbers about minerals, gas, travel times, etc from actual maps no?

It would seem you could combine these GA's with some of the SC2 map analyzer code ( http://www.teamliquid.net/forum/viewmessage.php?topic_id=132627 ) to do this and possibly allow for other optimizations like grinding out optimal building placements/timing for fast tech or fast expansion builds and the like.

Either way really cool projects thanks, Lomilar and CarbonTwelve.

CarbonTwelve

Australia525 Posts

November 11 2010 00:38 GMT

#1024

On November 11 2010 09:29 Jaeger wrote:
You could make it more accurate by pulling numbers about minerals, gas, travel times, etc from actual maps no?

It would seem you could combine these GA's with some of the SC2 map analyzer code ( http://www.teamliquid.net/forum/viewmessage.php?topic_id=132627 ) to do this and possibly allow for other optimizations like grinding out optimal building placements/timing for fast tech or fast expansion builds and the like.

I was planning on making settings like this editable in the config for my app. Probably will be a fair way down the track though. Not sure I'd go as far as calculating bulding placement though - it's a lot of overhead for little gain. Probably something for a different app to do (ie, this app calculates build order, and another app takes a build order and tells you where to place things).

Either way really cool projects thanks, Lomilar and CarbonTwelve.

Lomilar deserves all the credit. Besides which, I haven't even released anything yet

Karliath

United States2214 Posts

November 11 2010 02:39 GMT

#1025

Hello!

I don't mean to be a dick, but when can we expect the Protoss version?

I'm really excited.

ChoboCop

United States954 Posts

November 11 2010 02:54 GMT

#1026

As a Software Engineer, I am really appreciating keeping up with this thread/project. The collaboration level of public need(business managers - requirements gathering) with the developer and constant feature releases(yes!) shows the true power of agile software development! This project can do nothing but succeed if things keep up!

michaelhasanalias

Korea (South)1231 Posts

November 11 2010 04:10 GMT

#1027

Is there a way to customize gas mining, such that I can simulate pulling drones off once I get the gas I want.

I am working with a metabolic boost (100gas) build order, and I want to see at time finished what the excess minerals are (to time an expansion), but I have all this gas I wouldn't have in an actual game.

Thanks so much for putting this program together, as I have been using it daily since I downloaded it.

edit: read the changelog and found that it's already in there... thanks!

Chopin

Norway55 Posts

November 11 2010 04:36 GMT

#1028

Very well done! Must have been fun to create

Thanks!

Loxon

United States20 Posts

November 11 2010 04:45 GMT

#1029

This has been my first experience watching open source programming from the ground up, and suffice it to say, I'm massively impressed. Nice work to all of you folks, and I hope to continue to assist with testing on both projects.

mango_destroyer

Canada3914 Posts

November 11 2010 07:32 GMT

#1030

At first I didn`t like the idea but I can see it`s uses for the early part of the game where there isn`t too many variables. I find this tool useful for the first 7 minutes or so of the game. Any later wouldn`t benefit too much because a whole slew of things can happen which would change any build plan, and for that reason that is why I think pro-gamers wouldn`t find it that useful because they probably already master build orders in the early portion of the game.

Anyways keep up the good work!

boom1

United States1 Post

November 11 2010 08:26 GMT

#1031

This is a very good starting tool to get an idea of how to start some builds.
keep it uP!

The Touch

United Kingdom667 Posts

November 11 2010 09:50 GMT

#1032

On November 11 2010 16:32 mango_destroyer wrote:
At first I didn`t like the idea but I can see it`s uses for the early part of the game where there isn`t too many variables. I find this tool useful for the first 7 minutes or so of the game. Any later wouldn`t benefit too much because a whole slew of things can happen which would change any build plan, and for that reason that is why I think pro-gamers wouldn`t find it that useful because they probably already master build orders in the early portion of the game.

Anyways keep up the good work!

I think at the very least the program can tell you the rough order to build things in. Say you want to get 12 roaches and 12 hydras with 1/1 upgrades, roach speed, and hydra range, as fast as possible. The program will be able to tell you whether you should be building the evo chambers first, at roughly what point you should start which upgrades, and roughly how late to leave the hydralisk den.

That way, if you get knocked off track, you still know roughly the best route to get back to your game plan.

Babaganoush

United States626 Posts

November 11 2010 09:51 GMT

#1033

Much thanks to the OP for making this thing. It's pretty interesting. I agree that it's probably only useful for the early game. Too many factors would affect anything further

Just to test out, I tried testing whats the optimal build order was for 2 mutalisks. The result was to build your lair at 10, then expand at 12, and then get your mutalisks around 28 supply... and they'll pop out at 8:50. ...Doesn't sound like something anyone would do, lol.

I changed it to 6 mutalisks then it seemed to calculate that better, stating that it'll take 7:12 to pop those out. I'm curious to why it couldn't figure out something optimal for the two mutalisks.

Maybe I just needed to run it a bit longer.

Edit: I noticed that Gas Extractors was part of the goal.

Will the optimizer not estimate how many extractors are needed for your build order?

CarbonTwelve

Australia525 Posts

November 11 2010 10:23 GMT

#1034

Ok, I've just uploaded the first version of my app to GoogleCode: http://code.google.com/p/scbuildorder/

Please please forgive me for the rough code, wildly varying coding styles (some of the memory pool stuff I think I wrote about 8 years ago), and the very very basic GUI. It's all a work in progress and it's a very early version

Anyway, there's an executable there too, so please feel free to try it out and if you notice any issues with the build orders it produces please let me know. Feature requests might have to be on hold atm as there's a crapload left that I want/need to do before niceties are done

Note: you'll need the 32 bit visual studio 2010 redistributables to run this. You can get them free from Microsoft (~5MB download): http://www.microsoft.com/downloads/en/details.aspx?familyid=A7B7A05E-6DE6-4D3A-A423-37BF0912DB84&displaylang=en

dcemuser

United States3248 Posts

November 11 2010 10:28 GMT

#1035

On November 11 2010 19:23 CarbonTwelve wrote:
Ok, I've just uploaded the first version of my app to GoogleCode: http://code.google.com/p/scbuildorder/

Please please forgive me for the rough code, wildly varying coding styles (some of the memory pool stuff I think I wrote about 8 years ago), and the very very basic GUI. It's all a work in progress and it's a very early version

Might want to add a note that users need the Visual C++ 2010 Runtime for this to work.

Also, you may want to start your own thread, because this is a different project than EC.

Lastly, did you talk to the EC developers? What did they think of your project?

Edit: I told it I wanted 1 Dark Templar, and here's what it came up with.

+ Show Spoiler +

0.00: 50M 0G 0E 6/ 10S - Build Probe
17.00: 73M 0G 10E 7/ 10S - Build Probe
34.00: 107M 0G 19E 8/ 10S - Build Probe
44.44: 116M 0G 25E 9/ 10S - Chrono Nexus
44.44: 116M 0G 0E 9/ 10S - Build Pylon
51.07: 50M 0G 4E 9/ 10S - Build Probe
74.59: 150M 0G 17E 10/ 18S - Build Gateway
82.46: 50M 0G 21E 10/ 18S - Build Probe
94.28: 75M 0G 28E 11/ 18S - Build Assimilator
101.64: 50M 0G 32E 11/ 18S - Build Probe
111.19: 75M 0G 38E 12/ 18S - Build Assimilator
118.64: 53M 0G 42E 12/ 18S - Build Probe
118.64: 3M 0G 42E 13/ 18S - Chrono Nexus
124.28: 51M 0G 20E 13/ 18S - Move Probe To Gas
124.28: 51M 0G 20E 13/ 18S - Move Probe To Gas
124.28: 51M 0G 20E 13/ 18S - Move Probe To Gas
139.59: 156M 29G 29E 13/ 18S - Build Cybernetics Core
139.59: 6M 29G 29E 13/ 18S - Move Probe To Gas
139.59: 6M 29G 29E 13/ 18S - Move Probe To Gas
139.59: 6M 29G 29E 13/ 18S - Move Probe To Gas
149.86: 50M 63G 34E 13/ 18S - Build Probe
149.86: 0M 63G 34E 14/ 18S - Move Probe To Gas
189.59: 186M 214G 57E 14/ 18S - Build Twilight Council
192.94: 50M 127G 59E 14/ 18S - Build Probe
192.94: 0M 127G 59E 15/ 18S - Chrono Nexus
204.50: 50M 171G 40E 15/ 18S - Build Probe
204.50: 0M 171G 40E 16/ 18S - Chrono Cybernetics Core
213.43: 50M 205G 20E 16/ 18S - Research Warpgate
239.59: 163M 254G 35E 16/ 18S - Build Dark Shrine
239.59: 13M 4G 35E 16/ 18S - Chrono Cybernetics Core
246.15: 50M 29G 13E 16/ 18S - Build Probe
263.83: 100M 96G 23E 17/ 18S - Build Pylon
271.70: 50M 126G 28E 17/ 18S - Build Probe
288.83: 109M 191G 37E 18/ 26S - Build Probe
288.83: 59M 191G 37E 19/ 26S - Chrono Nexus
300.16: 148M 234G 19E 19/ 26S - Build Probe
311.11: 192M 276G 25E 20/ 26S - Chrono Cybernetics Core
312.83: 207M 282G 1E 20/ 26S - Build Probe
328.97: 306M 344G 10E 21/ 26S - Convert Gateway To Warpgate
339.59: 411M 384G 16E 21/ 26S - Build Dark Templar
344.59: 410M 253G 19E 23/ 26S - Satisfied.

A few parts of that deserve some questioning, in my mind. I let the program run for about 5 minutes, about 1 billion games played.

First, it moves 7 probes to gas. This makes no sense.

Second:
" 204.50: 0M 171G 40E 16/ 18S - Chrono Cybernetics Core
213.43: 50M 205G 20E 16/ 18S - Research Warpgate"
Why does it Chrono the Cybernetics Core 8.93 seconds before it actually uses the Cybernetics Core? To save energy? That doesn't make sense either...

All in all, though, a great program. I can't wait to see it further along the line.

CarbonTwelve

Australia525 Posts

November 11 2010 11:16 GMT

#1036

On November 11 2010 19:28 dcemuser wrote:
Might want to add a note that users need the Visual C++ 2010 Runtime for this to work.

Yep, added that shortly after you posted, thanks.

Also, you may want to start your own thread, because this is a different project than EC.

Possibly, will get a few opinions on that one. Once it's a bit more polished I might move it to another thread.

Lastly, did you talk to the EC developers? What did they think of your project?

I originally sent a PM to Lomilar but I think he was away for a bit. Anyway I've been on IRC with them last day or so.

Edit: I told it I wanted 1 Dark Templar, and here's what it came up with.

+ Show Spoiler +

A few parts of that deserve some questioning, in my mind. I let the program run for about 5 minutes, about 1 billion games played.

First, it moves 7 probes to gas. This makes no sense.

Not according to the calcs done by PiousFlea: http://www.teamliquid.net/forum/viewmessage.php?topic_id=140055

I used these measurements for my app, which I think Lomilar was doing too but did a bit more simply. Anyway, I assumed 1 gas geyser is close and 1 is far, so one only benefits from 3 workers, the other from 4. In most builds it'll only use 3 per geyser, but for gas intesive builds it'll use 4 on the 2nd one.

Second:
" 204.50: 0M 171G 40E 16/ 18S - Chrono Cybernetics Core
213.43: 50M 205G 20E 16/ 18S - Research Warpgate"
Why does it Chrono the Cybernetics Core 8.93 seconds before it actually uses the Cybernetics Core? To save energy? That doesn't make sense either...

Hmm, not sure on that one, doesn't make much sense to me either. Will look into it.

All in all, though, a great program. I can't wait to see it further along the line.

Thanks

dcemuser

United States3248 Posts

November 11 2010 11:33 GMT

#1037

Yeah, the error with the Chrono boost is kind of odd, here are two more instances, in a "1 Warp Prism" test.

+ Show Spoiler +

0.00: 50M 0G 0E 6/ 10S - Build Probe
22.43: 100M 0G 13E 7/ 10S - Build Pylon
57.32: 150M 0G 32E 7/ 18S - Build Gateway
68.95: 50M 0G 39E 7/ 18S - Build Probe
68.95: 0M 0G 39E 8/ 18S - Chrono Nexus
84.97: 75M 0G 23E 8/ 18S - Build Assimilator
88.89: 19M 0G 25E 8/ 18S - Chrono Forge
114.97: 149M 0G 15E 8/ 18S - Move Probe To Gas
114.97: 149M 0G 15E 8/ 18S - Move Probe To Gas
122.32: 180M 10G 19E 8/ 18S - Build Cybernetics Core
133.33: 70M 26G 25E 8/ 18S - Chrono Nexus
133.33: 70M 26G 0E 8/ 18S - Move Probe To Gas
172.43: 213M 100G 22E 8/ 18S - Build Robotics Facility
184.76: 50M 23G 29E 8/ 18S - Build Probe
237.43: 204M 124G 59E 9/ 18S - Chrono Robotics Facility
237.43: 204M 124G 34E 9/ 18S - Build Warp Prism
248.04: 50M 144G 40E 11/ 18S - Build Probe
257.43: 40M 162G 45E 12/ 18S - Chrono Robotics Facility
270.77: 101M 187G 27E 12/ 18S - Satisfied.

The first Chrono makes perfect sense, it is the first time a probe has began production after 25 energy has been acquired.

The second one... is a on a Forge. There is no Forge. Even if there was a Forge, there is nothing in production at said imaginary Forge.

The third one occurs when nothing is building at the Nexus (for the entire duration of the boost). I think this may be an error with the "Move Probe to Gas" command or maybe just that it is trying to use its boosts at 25 energy sometimes.

The fourth and fifth boosts are perfect - Chronoing out the unit I asked for.

Hope this helps you narrow down the issue.

PrideNeverDie

United States319 Posts

November 11 2010 11:46 GMT

#1038

anyone figure out the best possible economic opening for zerg?

i saw a 13 pool 16 hatch build shown

CarbonTwelve

Australia525 Posts

November 11 2010 11:54 GMT

#1039

On November 11 2010 20:33 dcemuser wrote:
Yeah, the error with the Chrono boost is kind of odd, here are two more instances, in a "1 Warp Prism" test.

+ Show Spoiler +

Lol, hmm. Chrono forge was bugged because I added the command in at the last minute (after noticing it was missing in the available commands list), but I never actually implemented the restrictions for it, hence it used it without a forge existing. As for the Chrono Nexus, idunno. I suspect if you left it longer it'd eventually remove that as it doesn't help and the code prioritises shorter build orders that achieve the same result. Possibly a similar story with the cybernetics core from earlier. Having said that, I tried the same requirements (1 warp prism) and I got exactly the same result. That's more than just coincidence - I'll have to look into why it thinks those Nexus & Forge chrono boosts are a good idea. Part of the trouble with chrono boost is that it's so hard to help the algorithm decide how to use it :/

icezar

Germany240 Posts

November 11 2010 12:19 GMT

#1040

On November 11 2010 20:46 PrideNeverDie wrote:
anyone figure out the best possible economic opening for zerg?

i saw a 13 pool 16 hatch build shown

If you consider only minerals then the best i found was:
9 Overlord
15 Hatchery > transfer 2
16 spawning Pool
16 Overlord
19 queen
23 queen
25 Overlord
29 spawn larvae
29 spawn larvae
36 Overlord [4]
36 spawn larvae

that is using pen and paper because you cannot ask EC for best econ.
now, if you want to add gas... it depends on how much do you value it over minerals...

Prev 1 50 51 52 53 54 58 Next

Please or register to reply.

Zerg Build Order optimizer. - Page 52

Completed

Ongoing

Upcoming