EPT@Psilo: No overlay, it calculates build orders. You say what you want to have and it caculates the fastest possible way to it.
SCFusion - WoL, HotS & LotV Build Order Optimizer - Page 56
| Forum Index > SC2 General |
|
TaLuS
Germany17 Posts
@Psilo: No overlay, it calculates build orders. You say what you want to have and it caculates the fastest possible way to it. | ||
|
Borskey
160 Posts
Basically, I want to modify an existing build but only after, lets say, the 44 food mark. So I set a custom initial build order, but started running into problems. Specifically, things break at the part where the factory is supposed to be made, because workers never go into gas. I've tried several different things: even your example builds don't work (say, pick the 1-1-1 and make it spit out a banshee; it won't work) Here was my attempt at + Show Spoiler + 10 Supply Depot 13 Barracks (Naked) 15 Orbital Command 15 Marine 16 Command Center 17 Marine 18 Calldown MULE 18 Refinery 18 Refinery 19 Supply Depot 19 3*Move SCV To Gas 19 3*Move SCV To Gas 19 Bunker 21 Factory (Naked) 21 Barracks Reactor 21 Orbital Command 22 Calldown MULE 22 Supply Depot 23 Calldown MULE 25 Starport (Naked) 25 Lift Factory (Naked) 25 Lift Barracks (Reactor) 25 Land Factory (Reactor) 26 Land Barracks (Naked) 27 Barracks Tech Lab 27 Hellion 27 Hellion 33 Supply Depot 35 Hellion 35 Hellion 35 Calldown MULE 39 Lift Starport (Naked) 39 Lift Barracks (Tech Lab) 39 Land Starport (Tech Lab) 40 Land Barracks (Naked) 44 Barracks Tech Lab 44 Banshee No matter what I tried, Fusion either told me it didn't understand, or if it accepted the build order it would just never build a factory. So I decided to use the build order up till the bunker (I deleted everything after 19 supply, including the gas commands), and then set a way-point that could only be reached if it actually went in the gas and did the switching to get the reactor on the factory. Then I had my target as 2 marines 6 hellions 1 banshee with constant SCV production. In simple, this is what your program spit out: + Show Spoiler + 10 Supply Depot 13 Barracks (Naked) 15 Orbital Command 15 Marine 16 Command Center 17 Marine 18 Calldown MULE 18 Refinery 18 Refinery 19 Supply Depot 19 Bunker 19 Move SCV To Gas 19 Move SCV To Gas 19 Move SCV To Gas 19 Move SCV To Gas 19 Move SCV To Gas 19 Move SCV To Gas 20 Factory (Naked) 21 Barracks Reactor 23 Supply Depot 23 Calldown MULE 24 Supply Depot 27 Starport (Naked) 27 Lift Barracks (Reactor) 27 Lift Factory (Naked) 28 Land Factory (Reactor) 29 Hellion 32 Hellion 37 Starport Tech Lab Waypoint 1 satisfied: 6:17.33: 332M 251G 47E 37/ 54S Income: 1122M 228G Buildings: 1 Command Center 1 Orbital Command 2 Refinery 4 Supply Depots 1 Barracks 1 Bunker 1 Factory 1 Starport 1 Reactor Units: 29 SCV 2 Marine 2 Hellion Research: 38 Hellion 40 Hellion 44 Banshee 49 Calldown Extra Supplies 50 Hellion 52 Hellion Waypoint 2 satisfied: 7:40.22: 937M 466G 44E 58/ 62S Income: 1378M 228G Buildings: 1 Command Center 1 Orbital Command 2 Refinery 4 Supply Depots 1 Extra Supplies 1 Barracks 1 Bunker 1 Factory 1 Starport 1 Reactor 1 Tech Lab Units: 39 SCV 2 Marine 6 Hellion 1 Banshee Research: So, I copied and pasted everything from that build up until it switches the factory and barracks, with the same waypoint and same target.The result: it still breaks because the SCVs again don't go into gas so it never builds a factory. =( | ||
|
matrius
100 Posts
I just saw this output while playing around, and not sure if its right or not: 6 3*Probe 9 Pylon 9 2*Probe 11 Chrono Nexus 11 Probe 12 Chrono Nexus 12 Probe 13 Gateway 13 3*Probe 16 Chrono Nexus 16 2*Probe 18 Chrono Nexus 18 Pylon 18 Probe Waypoint 1 satisfied: 3:39.78: 819M 0G 24E 19/ 26S Income: 706M 0G Buildings: 1 Nexus 2 Pylon 1 Gateway Units: 19 Probe Upgrades: What is income? Can it be less than the amount of minerals left in bank at end of a build? Can it be less more than 50 minerals (the amount you start with)? Thanks! | ||
iHirO
United Kingdom1381 Posts
| ||
|
t0b4cc0
Austria12 Posts
I just wanted to figure out some timings for me and my mate for planing pushes for 2v2, to see at wich point we booth have the push ready. € the problem is, that the program always tell you that its good to make a 12 hatch or 14 cc. But the problem is especially in 2v2, you will just get rushed to death when you booth fastexpand. (the program even tells me to doubleexpand for 40 banelings, wihtout any unit to defend it...) Is there any way to tell the program to stay on on base (easy solution), or to expand after a certain point (best solution)? The program is cool dont get me wrong, but its getting completely useless because u cant doubleexpand with a mate who expands too in 2v2, so that you ahve 5 bases but not even 1 unit to defend it. regards t € i found a complicated way around this...... i always make waypoints when i want to have a certain thing and look how long it need to come to this waypoint, lets say i need 4 marines and a bunker early (so that the program doesnt tell me that i need a fast expand cause i need marines and bunkers) then i cut down the time as much as possible until ive got it the next thing waypoint is factorys with teclabs more marines then tanks and siegemode until i have my build that works quite ok with terran, but i have found a problem with zerg with zerg it doesnt understand how to build workers constant i dont want to have 7 workers at 3 minutes when i say "build drones constantly" ~.~ | ||
|
lugaidster
Chile30 Posts
Cheers! | ||
|
Popof
Brazil1 Post
 Would like to also suggest you to add an "Pause" button aside with the "Stop" one. If I pressed pause, the program would temporarily stop using my cpu. Pressing "Resume" would make the algorithm continue from where is has stopped. That would be usefull because when the program is actually running, it sucks almost 100% of my CPU, so I can't test a build that is not totally completed yet (by actually playing it on the game); It takes a good amount of time till the completion likelihood reachs 100%. To test the build I have to stop the program, test it on the game and then start the algorithm from zero again. I know you could say "ah, just wait the full completion time", but I think that addon would be really handy. | ||
|
shadogi
United States194 Posts
Let's say that I want a total of 3 hatcheries at the end of my build (ie. starting hatchery and 2 expansions), but I want one of them built before 4:00 and the other at 6:00. Does that mean under waypoint 1 (4:00) and 2 (6:00) I would put in one hatchery each because I want one built during each interval or do I put 1 under way point 1 and 2 under waypoint 2 as I will have 2 hatcheries at the end of way point 2? I hope I'm being clear. Thanks! | ||
|
SpaceYeti
United States723 Posts
| ||
|
Schnullerbacke13
Germany1199 Posts
 | ||
|
CarbonTwelve
Australia525 Posts
So I've released a new version. There have been quite a lot of changes included in this, and to cover some of the important ones: 1. The engine has been updated to the latest patch changes (v1.4.3+) 2. Various improvements to efficiency, however I've also stopped it from rejecting invalid build orders and instead it will now insert the appropriate commands to make it valid. This means that the game rate will likely decrease for most builds, but it should find the optimal build order a lot faster. You can see a good example of this with the 7 roach rush (7 roaches & 1 queen); in v0.9 it can easily get stuck with just 10 drones, but with v0.95 it now finds the optimal build almost straight away (within 1-2s in most cases). 3. The 'completion likelihood' no longer displays past 90%. The reason for this is that it's clear from the comments in the thread that many people were running the build until it gets to 100%, but this is not how this figure should be interpreted - the system will never be 100% certain of a build, and this 'likelihood' is a simple formula that can be used as a guide to whether it's likely to find a better build. So now it counts up to 90% and then after that just shows "> 90%". As always (well, as it always used to be) let me know if you find any issues, and thanks if you're still interested in the app As a teaser, more is on the way, and I'm hoping that the next version will be 1.0, and along with it will have the system load the building/unit/research requirements, costs, etc. loaded from XML. This is a big change though, so it may take a little while... | ||
|
NeverContent
Sweden2 Posts
Oh and one other thing... will the source code ever be public again? | ||
|
CtrlAltDefeat
Israel106 Posts
| ||
|
iHirO
United Kingdom1381 Posts
| ||
|
CarbonTwelve
Australia525 Posts
On December 01 2012 08:41 NeverContent wrote: Yay! It's great that you're back. I have a question though. It says one of the new things is "Extra supply limits"... What does this mean? I initially thought that maybe you could go beyond 200/200 now, but I can't seem to get that to work. Sorry, it's a reference to the Terran ability to call down extra supplies - there's an 'Extra Supplies' entry that allows you to prevent the algorithm from using that ability (so that it would only use mules. Oh and one other thing... will the source code ever be public again? Eventually, yes, I need to redo the folder structure in the google project page... On December 01 2012 09:03 iHirO wrote: Nice! Any chance of a mac compatible version? Possibly, but it's not a high priority. There are still some fundamental things that AFAIK aren't supported in MacOS that would require a significant redesign to handle (ie, thread local storage). | ||
|
NeverContent
Sweden2 Posts
On December 01 2012 09:35 CarbonTwelve wrote: Sorry, it's a reference to the Terran ability to call down extra supplies - there's an 'Extra Supplies' entry that allows you to prevent the algorithm from using that ability (so that it would only use mules. Ah... should've been obvious. Eventually, yes, I need to redo the folder structure in the google project page... Awesome... looking forward to that a lot. Thnx. | ||
|
Schnullerbacke13
Germany1199 Posts
| ||
|
ksyrog
Poland15 Posts
Flat 40/20/0 minerals per probe would be nice. BTW, I found something interesting - your algorithm have tendency to build unnecessary assimilators. | ||
|
CarbonTwelve
Australia525 Posts
On December 09 2012 21:59 ksyrog wrote: BTW, I found something interesting - your algorithm have tendency to build unnecessary assimilators. It does this because it values gas more than minerals. The build order is still the fastest possible, so it's trying to get as much gas in that time as it can. | ||
|
Aelfric
Turkey1496 Posts
| ||
| ||
