EPTAnother features that would be nice for future implementation is a way to specify whether a certain field should be maximized or minimized but I expect that this would be much harder to implement.
Zerg Build Order optimizer. - Page 24
| Forum Index > SC2 General |
|
Traceback
United States469 Posts
Another features that would be nice for future implementation is a way to specify whether a certain field should be maximized or minimized but I expect that this would be much harder to implement. | ||
|
Adeny
Norway1233 Posts
This confirms my suspicions a little, but it would be great if I had the exact timings of a 4 gate, so I'm eagerly awaiting the protoss version of this program. Anyone else able to defend 4 gate?  | ||
|
a176
Canada6688 Posts
| ||
|
NATO
United States459 Posts
On October 23 2010 02:11 Adeny wrote: So after about an hour it still hasn't come up with a build that adequately defends against a 4 gate rush, and can continue to match the production of the protoss player (i.e. if he just sits on his 4 gates for a while I won't die). This confirms my suspicions a little, but it would be great if I had the exact timings of a 4 gate, so I'm eagerly awaiting the protoss version of this program. Anyone else able to defend 4 gate? You don't need the same production as a 4 gate to defend it since zerg units are better than toss gateway units. | ||
|
LaughingTulkas
United States1107 Posts
On October 23 2010 01:02 tetracycloide wrote: So I seeded the gernerator with the following: Drones: 200 Deadline: 20:0 It's been running for 2 hours (I'm going to let it keep running since it's still updating the build every minute or two) but so far the results are interesting. ---Final Output--- At time: 10:45 Minerals: 4014 Gas: 0 Supply: 214/214 Drones: 200 Overlords: 25 Queens: 7 Hatcheries: 7 Spawning Pools: 1 It does not use any kind of extractor trick or anything to get to 214/214 so I'm pretty sure the supply cap of 200 simply isn't coded. I know planning out a build to 200 supply is stupid and the program really doesn't need to know there's a supply cap to do any of the things it can do well but it's still, technically I think, a bug. Might be worth noting that, with this seed of essentialy 100% economy no units, the build it's trying variations of after 2+ hours is 9 overlord, 14 hatch, 15 pool. I like this since it's pretty close to the build being run at the pro level that, essentially, builds nothing but drones until units are absoultely needed, then builds units. It's still updating the build once or twice a minute though, even after 2 hours of running. I must be missing something. If you're just getting 200 drones, why would the program build a pool at all? That's just a wasted drone and minerals for no real purpose? Is the program making a mistake or did you ask for 200 drones and a pool? | ||
|
Artificial
United States4 Posts
I must be missing something. If you're just getting 200 drones, why would the program build a pool at all? That's just a wasted drone and minerals for no real purpose? Is the program making a mistake or did you ask for 200 drones and a pool? It makes a pool for queens. | ||
|
Lomilar
United States130 Posts
Restart the app every time you try a new build or: enter 0 into all waypoints and final under 'roach warren' between builds. Also fixed will be the 7 drone bug, had a spawning pool dependency for the queen that was always stuck in, so all builds 'had' to have a spawning pool. If you see issues like this, please report them in the issue section on the project page. That is the first place I go to fix problems. Thanks! Obviously there are still things to be worked out. :-) v0015 is up on the website. | ||
|
Cirqueenflex
499 Posts
I wanted to get 10 roaches + tunneling claws + burrow ASAP it took my computer about 5 minutes to figure out the best build next i requested it to get a queen (to see if a queen is worth getting if you want to roach-burrow-rush) after 5 minutes i got the answer that it takes 3 seconds more if i want to get a queen in this build, also i get 4 more drones and way more ressources. So 3 seconds more waiting time for roaches means a way better economy (thats why you should never just eat the results of that calculator but do some all-around-testing) i then noticed that it got the Queen after Lair So i wanted to force it to get the Queen before Lair to see if there is an economic advantage. With 1 additional waypoint it really made the calculations way more complicated. It started off testing 5k games where it would make 1 Queen for the first waypoint and then 3 additional Queens after the waypoint (for apparently no reason). Then it calculated >25000000!!! Games until it noticed that it had to start the tunneling claws upgrade at some point. Until 25kk games it just calculated games where it never even "thought" of the upgrade, went through the builds and then suddenly wondered why it didn't get the required upgrade in time etc... The next 5kk games brought out 1 build to meet the requirement. Until my work is done i will let it calculate the result but i feel like this takes days | ||
|
Shikyo
Finland33997 Posts
On October 23 2010 02:11 Adeny wrote: So after about an hour it still hasn't come up with a build that adequately defends against a 4 gate rush, and can continue to match the production of the protoss player (i.e. if he just sits on his 4 gates for a while I won't die). This confirms my suspicions a little, but it would be great if I had the exact timings of a 4 gate, so I'm eagerly awaiting the protoss version of this program. Anyone else able to defend 4 gate? The fastest 1-gas 4gate will have the warpgates up so that he'll have about 8 2-supply units(depends on what he chooses to make) entering your natural at about 6:30. Might be a good idea to try something like, spine and ~6 Roaches by 6:30, or a spine and ~28 speedlings at 6:30, with as many drones as possible. | ||
|
teemh
Canada120 Posts
| ||
|
Adeny
Norway1233 Posts
On October 23 2010 02:25 NATO wrote: You don't need the same production as a 4 gate to defend it since zerg units are better than toss gateway units. It's caps lock day not opposite day. | ||
|
Lomilar
United States130 Posts
On October 23 2010 02:32 Cirqueenflex wrote: Extractor trick is in the newest version. I wanted to get 10 roaches + tunneling claws + burrow ASAP it took my computer about 5 minutes to figure out the best build next i requested it to get a queen (to see if a queen is worth getting if you want to roach-burrow-rush) after 5 minutes i got the answer that it takes 3 seconds more if i want to get a queen in this build, also i get 4 more drones and way more ressources. So 3 seconds more waiting time for roaches means a way better economy (thats why you should never just eat the results of that calculator but do some all-around-testing) i then noticed that it got the Queen after Lair So i wanted to force it to get the Queen before Lair to see if there is an economic advantage. With 1 additional waypoint it really made the calculations way more complicated. It started off testing 5k games where it would make 1 Queen for the first waypoint and then 3 additional Queens after the waypoint (for apparently no reason). Then it calculated >25000000!!! Games until it noticed that it had to start the tunneling claws upgrade at some point. Until 25kk games it just calculated games where it never even "thought" of the upgrade, went through the builds and then suddenly wondered why it didn't get the required upgrade in time etc... The next 5kk games brought out 1 build to meet the requirement. Until my work is done i will let it calculate the result but i feel like this takes days One of the downfalls of the evolutionary side of things is that if you demand a REALLY REALLY tight build, and then have some other destination, it may be able to achieve that tight build, but it takes a really long time to figure out what to do next. This is because everything hinges on one or two actions that need to be mutated to the new appropriate values. That's a thing I am currently trying to find out how to fix. | ||
|
Cirqueenflex
499 Posts
Also as i already mentioned please let the user set a maximum of everything. This might be the same function as above and lets you somehow control the system way better (like stay on 1 base only) i think this won't be that hard to implement. You have to add/subtract to the pool of available stuff that can be done at one point in time depending on what already exists/what is required/what is denied. | ||
|
icezar
Germany240 Posts
| ||
|
Lomilar
United States130 Posts
Though that isn't obvious. Perhaps I will make that a UI component, to show all the possible actions. | ||
|
Galois
United States124 Posts
Haven't tried using the program for the attacks, but they might be worth checking as well (even though they're both 100/100 anyway). | ||
|
Cirqueenflex
499 Posts
still this takes forever and my build isn't really THAT complicated. EDIT: well, even if limiting is done it still seems way too much. The whole waypoint system doesn't really satisfy me at all. A priority in build order (Queen before Lair) would be easier to implement i guess instead of calculating n^n builds (n builds until waypoint, n builds after waypoint, together n^n) also how often does the system check for available actions? I guess right now it's just a loop going through or is it every second? If not every second, why not make it? | ||
|
icezar
Germany240 Posts
On October 23 2010 04:02 Lomilar wrote: Cirqueen: The limiting of possible actions is already being done. Besides the roach thing (a bug. :-x), the only possibilities are those which can be used to get you to your destination. Though that isn't obvious. Perhaps I will make that a UI component, to show all the possible actions. If this is done why if you give it 7 drone by 0:20 sec it will give you the build but is trying other things also, What is it calculating? I let it 10 min with this simple build and it was looking for something better?! | ||
|
a176
Canada6688 Posts
On October 23 2010 03:41 Cirqueenflex wrote: Also as i already mentioned please let the user set a maximum of everything. This might be the same function as above and lets you somehow control the system way better (like stay on 1 base only) i think this won't be that hard to implement. You have to add/subtract to the pool of available stuff that can be done at one point in time depending on what already exists/what is required/what is denied. I was about to ask if the program was truly following the requirements strictly. I'm guessing not, as I'm coming up with 'satisfied' builds that have more units than I'm requiring, or more or less than the buildings I'm requiring. I'm assuming the requirements aren't strictly followed (I submitted the 4-extractor 3-extractor issue). Perhaps the program will come up with the proper final solution after a long period of time, but I think it shouldn't show the build as satisfied if its not following the requirements strictly. I think I've tried to read all your posts (lomilar), but if there's something I don't get, please do let me know as the above. | ||
|
Lomilar
United States130 Posts
On October 23 2010 04:08 Galois wrote: Found a bug: The cost of Carapace is listed as 100/100. Its actually 150/150. Haven't tried using the program for the attacks, but they might be worth checking as well (even though they're both 100/100 anyway). Will be fixed in v0016. Thanks. | ||
| ||
