|
Ok, so here is all the input i used (Everything is >=):
Three to gas only
Starting build:
Build Drone
Build Drone
Extractor Trick Drone
Build Overlord
Build Hatchery
Build Drone
Build Drone
Build Drone
Build Extractor
Build Spawning Pool
Build Drone
Move Three Drones To Gas
Build Drone
Build Drone
Build Drone
Build Drone
Build Queen
Build Queen
Research Metabolic Boost
Move Three Drones To Minerals
Waypoint 1 max 3:15
8 Zerglings
Waypoint 2 max 4:00
16 Zerglings
Target:
300 Gas
3 Bases
1 Roach Warren
1 Evolution Chamber
1 Creep Tumor
12 Larva
57 Drones
9 Overlords
3 Queens
At the start i get a couple of successful solutions then after that the solutions get worse or even fail. I let it run over night and it even went up to around 70% for a failed solution.
|
On January 20 2016 21:10 zerge wrote:Ok, so here is all the input i used (Everything is >=): + Show Spoiler +Three to gas only
Starting build:
Build Drone
Build Drone
Extractor Trick Drone
Build Overlord
Build Hatchery
Build Drone
Build Drone
Build Drone
Build Extractor
Build Spawning Pool
Build Drone
Move Three Drones To Gas
Build Drone
Build Drone
Build Drone
Build Drone
Build Queen
Build Queen
Research Metabolic Boost
Move Three Drones To Minerals
Waypoint 1 max 3:15
8 Zerglings
Waypoint 2 max 4:00
16 Zerglings
Target:
300 Gas
3 Bases
1 Roach Warren
1 Evolution Chamber
1 Creep Tumor
12 Larva
57 Drones
9 Overlords
3 Queens At the start i get a couple of successful solutions then after that the solutions get worse or even fail. I let it run over night and it even went up to around 70% for a failed solution.
Thanks, I can reproduce the problem.
I think it's just due to the resource requirement in the final waypoint - I think the code that handles the display isn't seeing that the build is complete except for the gas, and hence keeps processing events hoping that it'll eventually work itself out. I'll look into it in more detail when I get the chance.
Incidentally, can you give me the Simple output for the build you have at 70% with a failed solution? So far at ~20% the build I've got show it as satisfied.
|
Sorry, i didn't save that build. I saved a successful solution that i got after a few seconds with very low percentage:
+ Show Spoiler + 14 Extractor Trick Drone
15 Build Overlord
15 Build Hatchery
17 Build Extractor
16 Build Spawning Pool
16 Move Three Drones To Gas
20 2 * Build Queen
24 Research Metabolic Boost
24 Move Three Drones To Minerals
27 Build Overlord
32 Build Hatchery
31 Move Three Drones To Gas
31 Queen Spawn Larvae
31 4 * Build Zergling
35 Build Hatchery
35 Build Overlord
35 Queen Spawn Larvae
36 Build Overlord
Waypoint 1 satisfied:
3:14.37: 59M 95G 1L 36/ 36S
Income: 1032M 114G
Buildings: 2 Hatchery 1 Extractor 1 Spawning Pool
Units: 27 Drone 3 Overlord 8 Zergling 2 Queen
Research:
36 Build Queen
39 Build Zergling
43 3 * Build Zergling
46 Queen Spawn Larvae
48 Build Queen
51 2 * Queen Spawn Larvae
Waypoint 2 satisfied:
3:51.94: 170M 194G 0L 51/ 58S
Income: 1233M 114G
Buildings: 3 Hatchery 1 Extractor 1 Spawning Pool
Units: 32 Drone 5 Overlord 16 Zergling 2 Queen
Research: Metabolic Boost
53 Build Overlord
53 Queen Spawn Creep Tumor
56 Build Overlord
56 Build Roach Warren
55 Build Creep Tumor From Creep Tumor
55 2 * Queen Spawn Larvae
64 2 * Queen Spawn Larvae
64 Move Three Drones To Minerals
66 Build Evolution Chamber
65 Build Overlord
66 Build Overlord
66 Move Three Drones To Gas
66 Move Three Drones To Minerals
66 Queen Spawn Creep Tumor
67 Move Three Drones To Gas
73 Queen Spawn Larvae
Waypoint 3 satisfied:
4:51.81: 573M 308G 12L 73/ 96S
Income: 2103M 114G
Buildings: 4 Hatchery 1 Extractor 1 Spawning Pool 1 Roach Warren 1 Evolution Chamber 3 Creep Tumor
Units: 57 Drone 9 Overlord 16 Zergling 4 Queen
Research: Metabolic Boost
14 Extractor Trick Drone
15 Build Overlord
15 Build Hatchery
17 Build Extractor
16 Build Spawning Pool
16 Move Three Drones To Gas
20 2 * Build Queen
24 Research Metabolic Boost
24 Move Three Drones To Minerals
28 Build Macro Hatchery
27 Build Overlord
28 Build Overlord
28 Queen Spawn Larvae
29 Build Zergling
31 2 * Build Zergling
33 Move Three Drones To Gas
33 Build Zergling
34 Queen Spawn Larvae
34 Build Hatchery
36 Build Zergling
37 Queen Spawn Larvae
Waypoint 1 satisfied:
3:14.37: 37M 58G 1L 38/ 44S
Income: 942M 114G
Buildings: 2 Hatchery 1 Extractor 1 Spawning Pool
Units: 25 Drone 4 Overlord 8 Zergling 2 Queen
Research:
38 Build Overlord
38 Build Queen
40 Queen Spawn Larvae
46 3 * Build Zergling
50 Queen Spawn Larvae
53 Build Overlord
Waypoint 2 satisfied:
3:58.66: 20M 174G 0L 57/ 58S
Income: 1412M 114G
Buildings: 3 Hatchery 1 Extractor 1 Spawning Pool
Units: 37 Drone 5 Overlord 16 Zergling 2 Queen
Research: Metabolic Boost
58 Build Roach Warren
57 2 * Queen Spawn Larvae
65 2 * Queen Spawn Larvae
65 Build Overlord
65 Build Evolution Chamber
65 2 * Build Overlord
71 Queen Spawn Creep Tumor
Waypoint 3 satisfied:
4:46.84: 640M 300G 12L 71/ 96S
Income: 2084M 114G
Buildings: 4 Hatchery 1 Extractor 1 Spawning Pool 1 Roach Warren 1 Evolution Chamber 1 Creep Tumor
Units: 57 Drone 9 Overlord 16 Zergling 3 Queen
Research: Metabolic Boost
After that i watched the output for a few more minutes and i only saw slower or failed solutions.
I don't think it's the gas requirement, i tried a few different things like lair and +1 researching as target instead of gas and get the same results. Usually within the first minute there are a couple of good results that are overwritten quickly by worse results that stay a lot longer.
|
After that i watched the output for a few more minutes and i only saw slower or failed solutions.
The slower or failed solutions are actually what I'm looking for. But I think I might have enough to play around with.
I don't think it's the gas requirement, i tried a few different things like lair and +1 researching as target instead of gas and get the same results. Usually within the first minute there are a couple of good results that are overwritten quickly by worse results that stay a lot longer.
Yes, you're right, I just tested with the same details and found it went from a successful build to a failed one, seemingly because it waited until the lair and research completed before deciding the build was over. I believe it's still an issue where I was thinking (a slight difference between the calculation code and the output code), so I'll have to do some investigation with these builds. Thanks for the report.
|
Found this thread again because I forgot how to properly use the initial build order option, decided to make an account just to say thanks for such a great program. I'm an old man, but still a diamond level zerg, and I've used this program for a variety of reasons. One of my favorite successes with this program was a very simple 7 muta rush that I used in 3s with allies that were on board. Since void of the legacy's release, I've been messing with this program again, and thought I would share that last night I made 12 mutas hatch at 5:01 (in 2v2 with a zerg ally who knew and planned to keep me alive) off 2 bases and 3 gases. The inject finished the same time the spire finished, 36/60 supply, and almost exactly 1200 1200.
I'd post the exact build, but literally all I did was mutalisk = 12.
I studied this sort of stuff in college, and for me, the only thing sweeter than a timing that lines up like this, is watching the replay from the perspective of a master level terran who had nothing but reapers see the murder show up. Apm drops to 0, checks the ally's base, F10-N.
I've used this program for far safer builds, but had some annoying issues with using waypoints. I remember that a simple trick was to set the initial build order, but cutting and pasting wasn't working. I thought I would also share this issue I ran into and how I solved it. The first thing I did was put in a simple opening of 19 drones, 2 queens, 4 lings, 2 hatches, ling speed. It came out with a very easy to remember (old school 14 14) build that was 2:47. Now I wish "add waypoint" could extend onto a build, but in a new build order, I tried to put the parameters in identically with a waypoint 1 timer of 2:48, and 12 mutas at target, and it couldn't satisfy waypoint 1. After a while of tinkering, and figuring out the easy syntax that I had forgotten, using the initial BO option with the 14 14, I got the 2:47 timing and simple opening back.
This is probably user error somewhere, and/or a very simple question for you to answer, but thought I would share my experience with your program. ggwp. thx for helping an old man pwn some n00bs.
|
On January 22 2016 06:07 StigPiranha wrote:
The inject finished the same time the spire finished, 36/60 supply, and almost exactly 1200 1200.
I'd post the exact build, but literally all I did was mutalisk = 12.
I studied this sort of stuff in college, and for me, the only thing sweeter than a timing that lines up like this, is watching the replay from the perspective of a master level terran who had nothing but reapers see the murder show up. Apm drops to 0, checks the ally's base, F10-N.
Awesome, I love hearing about things like this 
I've used this program for far safer builds, but had some annoying issues with using waypoints. I remember that a simple trick was to set the initial build order, but cutting and pasting wasn't working. I thought I would also share this issue I ran into and how I solved it. The first thing I did was put in a simple opening of 19 drones, 2 queens, 4 lings, 2 hatches, ling speed. It came out with a very easy to remember (old school 14 14) build that was 2:47. Now I wish "add waypoint" could extend onto a build, but in a new build order, I tried to put the parameters in identically with a waypoint 1 timer of 2:48, and 12 mutas at target, and it couldn't satisfy waypoint 1. After a while of tinkering, and figuring out the easy syntax that I had forgotten, using the initial BO option with the 14 14, I got the 2:47 timing and simple opening back.
This is probably user error somewhere, and/or a very simple question for you to answer, but thought I would share my experience with your program. ggwp. thx for helping an old man pwn some n00bs.
Yes, I can replicate this. It's a little strange, but I think I've got an idea of what is going on, which is that when there's a maximum time it looks at what you've got by time X and calculates the value at that time, rather than continuing on and calculating the full completion of the waypoint and then trying to improve the time. I think it should be a fairly simple thing to change, even if only to test the theory.
|
I tried the SALT export option and when importing into SALT, the build order starts at 6 supply instead of 12. Not sure if it is SALT or SCFusion's fault.
I can enter my first build order at 12 supply 0:00 time manually in the SALT editor without a problem.
|
On January 28 2016 04:02 drKobalt wrote:
I tried the SALT export option and when importing into SALT, the build order starts at 6 supply instead of 12. Not sure if it is SALT or SCFusion's fault.
This is a known issue to be fixed in the next version. It was caused by a change in the specs for the SALT format.
|
SCFusion v1.27 has been released: https://sourceforge.net/projects/scfusion/files/
* SALT supply count fixed (hopefully, untested atm)
* Waypoints now have a better chance of succeeding
* Lurker Den requirement fixed
* Minor other bugs fixed
FWIW, I have also uploaded the latest code to GitHub.
|
hey i am really excited about this software you made man! I am trying to use it to design a all in roach attack...
What i basicly want is the following end result
7 roaches at least and move out
with burrow research under construction and missle attack to level 1 (i prefer if it can calcualte it by the time the 7 roaches arraive at the opponents base to complete the 2 researches)
can the software decide how much drones is optimal to get to this point to constantly spawn roaches afterwards
also i want the software to calculate what would be the best time to build the queen so the mineral/larvae production are spent on time..
if how do you do it?
I have tried with several waypoints targets settings i cant make it work...
also how do you exactly use the initial order option? (i tried to write down something like "14 build extractor, 14 build spawning pool"
thanks
edit: how do i clear the output if i change something and i want to recalcualte some new settings? Why none of my builds the likehood of completion never reached above 90% (i don't quite understand what does that says)
edit2: how should i use the min/max time? I.e. i want to start spine craweler between i.e. 2:10 and 2:15- for that i create new way point i set min time 2:10 max 2:15 and i set 1 under maximum under construction.. altough in the output all the waypoints are satisfied the spine crawler dose not shows anywhere in the build order
edit3: i set a scounting worker to start from 0:21 and returns at 1:55 - after the simulation in the output i see that at 1:23 the scouting worker returns.. is this a bug?
|
On February 22 2016 03:15 masky007 wrote:
can the software decide how much drones is optimal to get to this point to constantly spawn roaches afterwards
No, it can't decide that for you, however if you calculate what income rate you want (based on your larvae spawn rate I guess) then you can set that income rate as a target.
also i want the software to calculate what would be the best time to build the queen so the mineral/larvae production are spent on time..
if how do you do it?
I have tried with several waypoints targets settings i cant make it work...
What I would do is calculate the best time for the 7 roaches (or at least, the time you're comfortable with) then set that as a waypoint with that time, and then the target can be the income rate with a queen or two as needed/desired.
also how do you exactly use the initial order option? (i tried to write down something like "14 build extractor, 14 build spawning pool"
You need to specify the actions as you see in the simple output, without the supply counts. Ie:
Build Drone
Build Drone
Build Overlord
Build Drone
Build Extractor
Move Three Drones to Gas
edit: how do i clear the output if i change something and i want to recalcualte some new settings?
The output intentionally keeps the previous build to save recalculating what it's already spent time on.
Why none of my builds the likehood of completion never reached above 90% (i don't quite understand what does that says)
The likelihood completion is just a crude estimate. This is an imperfect algorithm that relies on random chance to find new builds, so it can never guarantee that it is complete, it can only give an estimate.
how should i use the min/max time? I.e. i want to start spine craweler between i.e. 2:10 and 2:15- for that i create new way point i set min time 2:10 max 2:15 and i set 1 under maximum under construction.. altough in the output all the waypoints are satisfied the spine crawler dose not shows anywhere in the build order
You'd need to set a waypoint at 2:10 with 0 spine crawlers under construction, and one at 2:15 with 1 spine crawler under construction.
i set a scounting worker to start from 0:21 and returns at 1:55 - after the simulation in the output i see that at 1:23 the scouting worker returns.. is this a bug?
I might not have adjusted the scouting worker times for LotV time vs HotS time. I'll need to check it out.
|
Regarding the failed solutions, i tested it with the new version and it still happens. I think it's a problem with custom start build orders as i don't get failed solutions when i switch the starting build for a couple of additional waypoints. Also had not problems with setting ressources as goal as long as i don't use custom builds.
|
Is this a open source project? Can you run things from the command line? It would be nice if you can build from source on Linux.
|
On February 22 2016 08:36 Loccstana wrote:
Is this a open source project?
Yes, the source code is on GitHub.
Can you run things from the command line?
No, there is no command line support. What would you like to achieve with command line support? Just running it from the shell?
It would be nice if you can build from source on Linux.
Sorry, even though it uses wxWidgets there's quite a bit of Windows specific code, in particular thread local storage. I believe Linux and Mac have equivalents, but I haven't investigated them or looked for an open library with efficient cross platform TLS support.
|
I'm running it for the first time. Trying to get a build for a quick medivac drop with decent infrastructure, so I am asking for a medivac with 8 marines, as well as 2 orbitals, 2 reactors, 25 scvs and that kind of things.
It's been running for a good 15 minutes or so, last 5 minutes the likelihood climbed up towards 20% or so with a reasonable build (not really what I wanted, but probably a good solution to what I asked for). Then suddenly it found a new great build and the likelihood dropped down to 0% again. The new build was the same as before (I think), except that there is a "lift barracks" immediately followed by a "land barracks". 
It showed a few other builds with the same lifting and landing of buildings.
Just figured you want to know, and maybe hard-code something little that disallows the lift-land combo, making life easier for the terran villagers.
also: how long does it usually take?
|
|
|
I have been using the tool for a while and have been trying to do some comparison using this. I was looking at Nerchio vs. MarineLorD http://ggtracker.com/matches/6730717, and I was trying to see how fast the tool could get to MarineLorD's 5 minute mark of 17 marines, 2 medivacs, and 2 bases. The program is putting out two starports with tech labs attached to it. Any particular reason why it would be doing this, I could understand two reactors, but not tech labs. The program is currently at over 90% completion and has run 120 million games. I have it set at constant SCV, >= 18 marines, 1 reaper, 2 widow mines, and 4 medivacs, and stimpack is required. How many games does it take to truly get an accurate build order?
|
On July 22 2016 05:53 navenitram wrote:I have been using the tool for a while and have been trying to do some comparison using this. I was looking at Nerchio vs. MarineLorD http://ggtracker.com/matches/6730717, and I was trying to see how fast the tool could get to MarineLorD's 5 minute mark of 17 marines, 2 medivacs, and 2 bases. The program is putting out two starports with tech labs attached to it. Any particular reason why it would be doing this, I could understand two reactors, but not tech labs. The program is currently at over 90% completion and has run 120 million games. I have it set at constant SCV, >= 18 marines, 1 reaper, 2 widow mines, and 4 medivacs, and stimpack is required. How many games does it take to truly get an accurate build order?
I would need to see the build order to know why, although sometimes it will build things unnecessarily to assist with timing other actions to make improvements to the build order time. You can usually alleviate this by using the 'allow wait commands' option.
|
How can I start SCFusion with custom terran build orders ?
|
Is SCFusion applies latest LotV changes? I'm asking because there are no "Mag-Field Accelerator" research for Cyclone.
|
|
|
|
|
|
EPT
