|
Macrorush is a software which uses genetic algorithms to calculate optimal build orders based on user inputted objectives.
The latest version covers all three races, and has no restrictions or limitations whatsoever.
To try it:
http://www.macrorush.com/
For the ones who know others build order generators, this one has some distinct functions like "macro score", "manual adjust", and so on.
For the ones who don't know what is a build order generator, it's a program that tries to calculate what is the best sequence to build in order to fullfil some goal, for example, what's the fastest way to get 2 blue flame hellions? The program tries to answer, and after some time (it takes some time to complete the analysis) shows the best solution it found.
|
Hmm... downloading now. Looks promising!
|
Been playing around with it a bit, here are the changes and additions I personally would like to see
-Ability to save/load build orders
-Change the layout so it's easier to access different values, it's a little annoying to have to scroll all the way down to change tech options
-Allow setting a goal for mineral/gas
-Allow different ways to display the build order so there's a yabot format and a detailed format that allows you to see when exactly units finish
-Allow adding/removing checkpoints
-Allow setting time limits in each checkpoints
I also really like the manual adjust feature, but it becomes quite bothersome to deal with it when you have huge build orders. I'd prefer it if you could just modify the supply at which the buildings/units are built instead of having to move them up/down.
Last thing, for some reason the program seems to be extremely slow. For example, it does 7000 simulations in about a minute, while SCFusion already did a few millions
Hope you keep updating this, I see potential in it.
|
Shintuku, thanks for the feedback. We'll consider some changes in future.
About the manual adjust, there is a reason for it be the way it is. That's the only way you can determine an exact build order. I agree that showing the supply count is better to easily understand the BO, but it's a good way to READ a build order, not the best way to WRITE one. Let's say you wanna set the BO to:
9 supply depot
10 barracks
11 marine
12 marine
13 marine
14 barracks
It's hard to know just looking at this list if at the time the second barracks starts we have 14 supply due 11 scv's and 3 marines, or due 12 scv's and 2 marines (maybe the 3rd marine is still on queue), so it's hard to know if I need 11 or 12 scv's. To set exactly what you want, you set in manual adjust the exact sequence. After processing, you can click on "update other tabs" to see the BO in "supply count" format.
About the performance, I recommend you to compare not the number of simulations per minute, but the number of minutes to reach the goal. I really can't say how other build order generators simulate each game, but there is clearly a huge difference in our software. We do not simulate millions of games, but with much fewer simulations we evolve to objective with similar speed.
Furthermore, you can compare the result, but not just the number of game seconds the program tells you some BO takes. You should enter the game, simulate the BO and see the real duration. All build order generators have some deviation in time from a real BO execution, including macrorush. But to know what is the best BO, the only way is to test each one inside the game. Macrorush can not tell you allways the best option, but we think it tell you a very good option in most cases, and it let you try the "perfection" with manual adjust.
|
cool! Cant wait to check it out! Thanks!
|
I would love to check this out! but does it work on mac osx? :/
|
I ran into a rather odd issue..... I installed the program yesterday and ran it once to tinker. I then loaded up my computer today... and when I ran the program it went into some verification process and then popped up a prompt....
"Cannot Start Application
Cannot continue. The application is improperly formatted. Contact the application vendor for assistance."
I can click for more details where it provides an error sheet if that would be helpful to you? I just don't want to paste it up here.
I'm on Windows 7.
PS: Slick looking app!
|
Is this a permanent error? It's still showing this message?
We'll try to find out what happened. Meanwhile you can try to install again and see if it works.
|
Originally I coulnd't reinstall (same error would popup).
Oddly enough it allowed me to install now..... will keep you posted if I see any errors pop up again!
:D
|
Shintuku,
I look back to your list of suggestions and noticed that you asked for "a detailed format that allows you to see when exactly units finish", but it's already there, the second tab shows you the detailed build order, with start and finish for each unit, plus minerals, gas and supply step by step.
TheAlchemist89,
Good to know. It seems that some download file was corrupt, fresh reinstall is always a good test :D
|
Gonna take a look at it when I come home. I'll edit once I have something to add 
|
Firstly, it's great to see other projects working on build order optimisers. More projects encourages more innovation
On July 23 2011 21:21 MarcoBrei wrote:
About the performance, I recommend you to compare not the number of simulations per minute, but the number of minutes to reach the goal. I really can't say how other build order generators simulate each game, but there is clearly a huge difference in our software. We do not simulate millions of games, but with much fewer simulations we evolve to objective with similar speed.
While I agree that you can't necessarily compare the number of games simulated per second, I tried out your app on the traditional 7 roach rush (7 roaches with 1 queen) and I think it's struggling to find the optimal solution. I had it running for about 20 minutes (~40,000 games simulated) after which it said "Evolution process has stalled" and here are the results:
+ Show Spoiler +[Checkpoint:1 Generation: 1]
8 Spawning pool
9 Overlord
9 Queen
12 Extractor
12 Roach Warren
13 Roach
17 Overlord
17 Roach
19 Overlord
19 Roach
22 Roach
24 Roach
27 Roach
30 Roach
Time: 6:02
Number of Workers: 16
Macro Score: 15
Power score: 17
___________________________________
[Checkpoint:1 Generation: 2]
9 Overlord
9 Spawning pool
8 Queen
12 Roach Warren
12 Extractor
11 Roach
13 Overlord
13 Roach
15 Roach
17 Roach
19 Roach
21 Roach
23 Roach
Time: 5:28
Number of Workers: 9
Macro Score: 8
Power score: 17
___________________________________
[Checkpoint:1 Generation: 3]
9 Extractor trick (extractor, drone, cancel extractor)
10 Overlord
10 Spawning pool
9 Queen
13 Roach Warren
12 Extractor
11 Roach
13 Overlord
13 Roach
15 Roach
17 Roach
19 Roach
21 Roach
23 Roach
Time: 5:17
Number of Workers: 9
Macro Score: 8
Power score: 17
___________________________________
[Checkpoint:1 Generation: 4]
9 Extractor trick (extractor, drone, cancel extractor)
10 Overlord
10 Spawning pool
9 Queen
14 Extractor
13 Roach Warren
12 Roach
14 Overlord
14 Roach
16 Roach
18 Roach
20 Roach
22 Roach
24 Roach
Time: 5:15
Number of Workers: 10
Macro Score: 9
Power score: 17
___________________________________
[Checkpoint:1 Generation: 5]
9 Extractor trick (extractor, drone, cancel extractor)
10 Overlord
10 Spawning pool
9 Queen
14 Extractor
13 Overlord
13 Roach Warren
12 Roach
14 Roach
16 Roach
18 Roach
20 Roach
22 Roach
24 Roach
Time: 5:15
Number of Workers: 10
Macro Score: 9
Power score: 17
___________________________________
[Checkpoint:1 Generation: 6]
9 Extractor trick (extractor, drone, cancel extractor)
10 Overlord
10 Spawning pool
9 Queen
14 Extractor
13 Overlord
13 Roach Warren
12 Roach
14 Roach
16 Roach
18 Roach
20 Roach
22 Roach
24 Roach
Time: 5:15
Number of Workers: 10
Macro Score: 9
Power score: 17
___________________________________
[Checkpoint:1 Generation: 7]
9 Extractor trick (extractor, drone, cancel extractor)
10 Overlord
10 Spawning pool
9 Queen
14 Extractor
13 Overlord
13 Roach Warren
12 Roach
14 Roach
16 Roach
18 Roach
20 Roach
22 Roach
24 Roach
Time: 5:15
Number of Workers: 10
Macro Score: 9
Power score: 17
___________________________________
[Checkpoint:1 Generation: 8]
9 Extractor trick (extractor, drone, cancel extractor)
10 Overlord
10 Spawning pool
9 Queen
14 Extractor
13 Overlord
13 Roach Warren
12 Roach
14 Roach
16 Roach
18 Roach
20 Roach
22 Roach
24 Roach
Time: 5:15
Number of Workers: 10
Macro Score: 9
Power score: 17
___________________________________
[Checkpoint:1 Generation: 9]
10 Overlord
11 Overlord
11 Spawning pool
13 Queen
19 Extractor
18 Overlord
18 Extractor
17 Roach Warren
16 Roach
19 Roach
21 Roach
23 Roach
25 Roach
27 Roach
29 Roach
Time: 5:11
Number of Workers: 15
Macro Score: 14
Power score: 17
___________________________________
[Checkpoint:1 Generation: 10]
10 Spawning pool
10 Overlord
11 Queen
15 Extractor
14 Roach Warren
13 Roach
15 Overlord
15 Roach
17 Roach
19 Roach
21 Roach
23 Roach
25 Extractor
24 Roach
Time: 5:01
Number of Workers: 10
Macro Score: 8
Power score: 17
___________________________________
[Checkpoint:1 Generation: 11]
10 Extractor trick (extractor, drone, cancel extractor)
11 Overlord
12 Overlord
12 Spawning pool
14 Queen
19 Extractor
18 Overlord
18 Roach Warren
17 Extractor
16 Roach
18 Roach
20 Roach
22 Roach
24 Roach
26 Roach
28 Roach
Time: 4:58
Number of Workers: 14
Macro Score: 12
Power score: 17
___________________________________
[Checkpoint:1 Generation: 12]
10 Extractor trick (extractor, drone, cancel extractor)
11 Overlord
12 Overlord
12 Spawning pool
14 Queen
19 Extractor
18 Overlord
18 Roach Warren
17 Extractor
16 Roach
18 Roach
20 Roach
22 Roach
24 Roach
26 Roach
28 Roach
Time: 4:58
Number of Workers: 14
Macro Score: 12
Power score: 17
___________________________________
[Checkpoint:1 Generation: 13]
10 Extractor trick (extractor, drone, cancel extractor)
11 Overlord
12 Overlord
12 Spawning pool
14 Queen
19 Extractor
18 Overlord
18 Roach Warren
17 Extractor
16 Roach
18 Roach
20 Roach
22 Roach
24 Roach
26 Roach
28 Roach
Time: 4:58
Number of Workers: 14
Macro Score: 12
Power score: 17
___________________________________
[Checkpoint:1 Generation: 14]
10 Extractor trick (extractor, drone, cancel extractor)
11 Overlord
12 Overlord
12 Spawning pool
14 Queen
19 Extractor
18 Overlord
18 Roach Warren
17 Extractor
16 Roach
18 Roach
20 Roach
22 Roach
24 Roach
26 Roach
28 Roach
Time: 4:58
Number of Workers: 14
Macro Score: 12
Power score: 17
___________________________________
[Checkpoint:1 Generation: 15]
10 Extractor trick (extractor, drone, cancel extractor)
11 Overlord
12 Overlord
12 Spawning pool
14 Queen
19 Extractor
18 Overlord
18 Roach Warren
17 Extractor
16 Roach
18 Roach
20 Roach
22 Roach
24 Roach
26 Roach
28 Roach
Time: 4:58
Number of Workers: 14
Macro Score: 12
Power score: 17
___________________________________
[Checkpoint:1 Generation: 16]
10 Extractor trick (extractor, drone, cancel extractor)
11 Overlord
12 Overlord
12 Spawning pool
15 Queen
19 Extractor
18 Overlord
18 Roach Warren
17 Extractor
16 Roach
18 Roach
20 Roach
22 Roach
24 Roach
26 Roach
28 Roach
Time: 4:58
Number of Workers: 14
Macro Score: 13
Power score: 17
___________________________________
[Checkpoint:1 Generation: 17]
10 Extractor trick (extractor, drone, cancel extractor)
11 Overlord
12 Overlord
12 Spawning pool
15 Queen
19 Extractor
18 Overlord
18 Roach Warren
17 Extractor
16 Roach
18 Roach
20 Roach
22 Roach
24 Roach
26 Roach
28 Roach
Time: 4:58
Number of Workers: 14
Macro Score: 13
Power score: 17
___________________________________
[Checkpoint:3 Generation: 17]
10 Extractor trick (extractor, drone, cancel extractor)
11 Overlord
12 Overlord
12 Spawning pool
15 Queen
19 Extractor
18 Overlord
18 Roach Warren
17 Extractor
16 Roach
18 Roach
20 Roach
22 Roach
24 Roach
26 Roach
28 Roach
Time: 4:58
Number of Workers: 14
Macro Score: 13
Power score: 17
___________________________________
If I do the same simulation in SCFusion it gets this build order within about 5s:
+ Show Spoiler +10 Overlord
10 Extractor Trick
13 Spawning Pool
12 Extractor
11 Overlord
11 Move Drone To Gas
11 Move Drone To Gas
11 Roach Warren
10 Queen
12 Spawn Larvae
12 Roach
14 Roach
16 Roach
18 Move Drone To Gas
18 Roach
20 Roach
22 Roach
24 Roach
Waypoint 1 satisfied:
4:44.38: 134M 52G 3L 0L 0L 39E 26/ 26S
Income: 297M 114G
Buildings: 1 Hatchery 1 Extractor 1 Spawning Pool 1 Roach Warren
Units: 10 Drone 3 Overlord 1 Queen 7 Roach
Upgrades:
If you leave SCFusion running a bit longer it usually gets this BO within a minute:
+ Show Spoiler +10 Extractor Trick
11 Overlord
12 Spawning Pool
15 Extractor
16 Move Drone To Gas
16 Move Drone To Gas
16 Queen
18 Move Drone To Gas
18 Overlord
18 Overlord
18 Roach Warren
17 Spawn Larvae
17 Roach
19 Roach
21 Roach
23 Roach
25 Roach
27 Roach
29 Roach
Waypoint 1 satisfied:
4:38.03: 232M 54G 2L 0L 0L 39E 31/ 34S
Income: 516M 114G
Buildings: 1 Hatchery 1 Extractor 1 Spawning Pool 1 Roach Warren
Units: 15 Drone 4 Overlord 1 Queen 7 Roach
Upgrades:
I'm not sure why MacroRush is struggling with this, but perhaps something to look into?
Also, I found that sometimes I can't stop it in the middle of developing a build (Stop button is disabled), is that intentional?
I also noticed that it seems to use the queen's spawn larvae ability without showing it - any chance you could add that to the output so it's clear?
|
CarbonTwelve,
Have you tried to execute the BO in SC2? Some build order generators suggest valid build orders, but with (very) unreal execution time. If the time is not accurate, you must test the BO in game.
Macrorush may have performance improvements, but it's a hard task, and the developers have no much time to invest (unfortunately we have to work to pay the bills, and also we want some free time to play SC2 ).
The stop button is disabled intentionally when a strong mutation is happening with the cromossomes (build orders). If we let you to click in stop, you could wait a lot until the application really stops.
It's assumed that when you have a Queen, you are always injecting larva, the same way it's assumed when you have a extractor done you'll put 3 drones on it. We think this way the BO would be more simple to read.
|
On August 02 2011 06:36 MarcoBrei wrote:
CarbonTwelve,
Have you tried to execute the BO in SC2? Some build order generators suggest valid build orders, but with (very) unreal execution time. If the time is not accurate, you must test the BO in game.
I wrote SCBuildOrder & SCFusion, and have often tested the times in game for various builds. I can usually get within 1-2s of the suggested times from its output. Try it yourself if you like.
Macrorush may have performance improvements, but it's a hard task, and the developers have no much time to invest (unfortunately we have to work to pay the bills, and also we want some free time to play SC2 ).
Yes, I understand performance improvements are hard, and I suspect ultimately you'll find yourself limited by the language you're using (I'm guessing it's C#?). I also definitely understand the limitations of time to invest in the project as I often struggle with it myself.
The stop button is disabled intentionally when a strong mutation is happening with the cromossomes (build orders). If we let you to click in stop, you could wait a lot until the application really stops.
I don't think you should limit what the user can do though. If the user wants to stop processing (ie, they want to free up the CPU usage), they should be able to do so without killing it via the task manager.
It's assumed that when you have a Queen, you are always injecting larva, the same way it's assumed when you have a extractor done you'll put 3 drones on it. We think this way the BO would be more simple to read.
That's not always the best thing though - sometimes you'll want to use queens for creep tumors, and sometimes you don't need 3 workers on gas. One improvement you might want to look at for that is gas micro options (whether to always put 3 on gas, put 1 at a time, or move back and forth as required) - check SCFusion if you want to see an example of this.
|
Its downloading a 0Kb file foe me 
|
CarbonTwelve,
I'll test the build orders again, but when I did it in the past with SCBuildOrder it suggested a build order 30 seconds faster than the real time (4:30 -> 5:00). I'm not a pro gamer to execute a build order with perfection (I'm a programmer, not a pro gamer  ), but I tried my best and still got an unreal time.
The gas micro options is something great to see in a program, but it's not pratical at all to do in game. Most people (90% or more) won't be able to properly execute BO's with micro gas management, so we decide do simplify.
Anyways, thanks for your concern about our software.
|
Finally I had some time to test build orders. The 7 roach suggested by SCFusion I was able to do in 4:53 (best time), never in 4:38. I think that it can be done better, not sure if in 4:38. And I recognize that this BO is still better than macrorush 7 roach.
In another test, I set SCFusion and Macrorush to get 1 marauder only. Result:
SCFusion
6 Supply Depot
6 Barracks (Naked)
6 Refinery
7 Move SCV To Gas
7 Move SCV To Gas
7 Move SCV To Gas
7 Barracks Tech Lab
7 Marauder
Waypoint 1 satisfied: 2:58,15
Macrorush
7 Supply Depot
7 Barracks
7 Refinery
7 Tech Lab - Barracks
8 Marauder
Time: 3:06
But when I tried to simulate the build orders in game, both could be done in exactly 3:07 (my best try, for both). In this case macrorush BO is a little better in macro perspective (1 additional worker). And the time suggested by SCFusion (2:58) seems to be impossible to achieve.
As I said before, macrorush will not always show the best solution, so I recommend people to try build orders in game, adapt with manual adjust to see new scenarios. Use it as an alternative software to try build orders. Also, macrorush seems to be accurate in estimated time, at least for terran build orders (other races were little less tested in matter of time accuracy).
|
Here's a replay of me doing the 7RR in ~4:41 with the above build order (I delayed my spawning pool a little, hence the extra couple of seconds): http://www.sc2replayed.com/replay-videos/11285
I had to upload it as a manual upload (normally for vid files) as the standard uploads aren't working atm on sc2replayed. You should still be able to download & watch it though, lemme know if there's issues.
Anyway, this at least shows what's possible. I'm sure I could do the marauder one too if you wanted proof on that. SCFusion has been tested multiple times on its accuracy and most build orders can be performed within a couple of seconds of the suggested times. You can also put the output into http://haploid.nl/sc2/build_order/ to confirm the times through another site. Eg, here's the 7RR:
+ Show Spoiler +10 Extractor Trick
11 Overlord
12 Spawning Pool
15 Extractor
16 Put 1 on Gas
16 Put 1 on Gas
16 Queen
18 Put 1 on Gas
18 Overlord
18 Overlord
18 Roach Warren
17 Spawn Larvae
17 Roach
19 Roach
21 Roach
23 Roach
25 Roach
27 Roach
29 Roach
Haploid gives a guide time of 4:36.
|
I got this with SCFusion in like 10 seconds as CarbonTwelve has stated, interesting that it skips the Queen and claims 2 seconds faster than the above build.
+ Show Spoiler +10 Spawning Pool
10 Overlord
10 Extractor Trick
12 Extractor
11 Roach Warren
11 Move Drone To Gas
12 Move Drone To Gas
12 Overlord
12 Move Drone To Gas
12 Roach
14 Roach
16 Roach
18 Roach
20 Roach
22 Roach
24 Roach
Waypoint 1 satisfied:
4:34.11: 171M 53G 1L 0L 0L 0E 26/ 26S
Income: 381M 114G
Buildings: 1 Hatchery 1 Extractor 1 Spawning Pool 1 Roach Warren
Units: 12 Drone 3 Overlord 7 Roach
Upgrades:
This is what MacroRush gave me after ~30 mins
+ Show Spoiler +[Checkpoint:3 Generation: 21]
9 Spawning pool
10 Extractor trick (extractor, drone, cancel extractor)
11 Overlord
11 Roach Warren
10 Extractor
10 Overlord
10 Roach
12 Roach
14 Roach
16 Roach
18 Roach
20 Roach
22 Roach
Time: 4:49
Number of Workers: 11
Macro Score: 9
Power score: 14
You both seem to be using a genetic solution here, maybe SCFusion has optimized the solution space or something. It seems to get to the answer much faster.
|
Have never tried these programs!
Have to get them asap!
- Gj on making the program btw!
|
Why the fuck do all the awesome things come out when i dont have internet at home... good job tho dude, Seems very cool!
|
|
|
On August 05 2011 16:29 jarf1337 wrote:
I got this with SCFusion in like 10 seconds as CarbonTwelve has stated, interesting that it skips the Queen and claims 2 seconds faster than the above build.
Yep, it is a little faster (which means it's doing what it's supposed to), but having the queen is the better build so I add that to the requirements.
You both seem to be using a genetic solution here, maybe SCFusion has optimized the solution space or something. It seems to get to the answer much faster.
I think it's mostly due to the sheer number of games that SCFusion can analyse. It's extremely optimised which is why I wrote it in C++.
On August 05 2011 19:03 Inori wrote:Show nested quote +On August 03 2011 04:34 MarcoBrei wrote:CarbonTwelve, I'll test the build orders again, but when I did it in the past with SCBuildOrder it suggested a build order 30 seconds faster than the real time (4:30 -> 5:00). I'm not a pro gamer to execute a build order with perfection (I'm a programmer, not a pro gamer ), but I tried my best and still got an unreal time. I fully agree with this. Maybe it's just my luck, but any BO I've ever tested with SCBuildOrder was always behind by about 30 sec - 1min. I'm not a pro, but my execution level should be enough (mid masters).
Have you got an example? If so send it through to me and I'll look into it (don't want to spam this thread too much with SCBuildOrder/SCFusion stuff).
|
On August 05 2011 13:03 CarbonTwelve wrote:
I'm sure I could do the marauder one too if you wanted proof on that.
I think it's impossible to get one marauder in 2:58, with this BO. In fact, I feel like it's impossible to get one marauder in less than 3:00 with any build order.
I'm not here to start a fight between the two programs, as I said before macrorush will not have the best solution in all situations, also will not be the faster program, but I did think it was necessary to point out some common mistakes related to "number of games processed" (this number has no such relevance since each program has a complete distinct methods to evolve generations) and "real time execution" (I recommend people to test build orders in game, since any suggestion of any build order generator may have some deviation).
|
On August 05 2011 11:05 MarcoBrei wrote:
But when I tried to simulate the build orders in game, both could be done in exactly 3:07 (my best try, for both). In this case macrorush BO is a little better in macro perspective (1 additional worker). And the time suggested by SCFusion (2:58) seems to be impossible to achieve.
I tried the build orders out too, but I managed to get my Marauder out a full 7 seconds faster than you were able to. It popped out at 3:00. I used the build order from SCFusion.
Replay: link
|
I tried it once and managed a 2:59 marauder. After a few more tries, I decided to try it in slower game speed. Managed a 2:57 marauder. I think the build order can be improved by initially sending 4 scvs on gas, because waiting for the gas for the tech lab is the bottle neck in this build, and the scv that built the refinery doesn't come out facing the CC thus the refinery isn't mined efficiently.
2:57 marauder
|
On August 06 2011 00:14 MarcoBrei wrote:Show nested quote +On August 05 2011 13:03 CarbonTwelve wrote:
I'm sure I could do the marauder one too if you wanted proof on that. I think it's impossible to get one marauder in 2:58, with this BO. In fact, I feel like it's impossible to get one marauder in less than 3:00 with any build order. I'm not here to start a fight between the two programs
I don't want to start a fight either, I only started showing replays once you were saying the times from SCFusion were impossible to achieve, which has been shown isn't the case.
On August 06 2011 04:44 Not_That wrote:I tried it once and managed a 2:59 marauder. After a few more tries, I decided to try it in slower game speed. Managed a 2:57 marauder. I think the build order can be improved by initially sending 4 scvs on gas, because waiting for the gas for the tech lab is the bottle neck in this build, and the scv that built the refinery doesn't come out facing the CC thus the refinery isn't mined efficiently. 2:57 marauder
This can also depend on the map and spawn location as to how close the gas geysers are. I used to assume there was 1 close geyser and 1 far away (usually a diagonal from your base), but a lot of people were reporting bugs that for gas intensive builds (eg, DTs) it would put 7 workers on gas (3 for the close one, 4 for the one further away), so I switched it to just assume 2 close geysers (so it only uses 3 each).
Either way, the build order is certainly valid and the times are fairly accurate.
|
apparently all 3 of my firewalls say that it is risky.... (becuase it wants to download some component..) Not gonna take the chance.
|
CarbonTwelve,
Ok, it's possible to do one marauder very close to 2:58, but I noticed that it's required to choose carefully the mineral patches, also build constructions very close to minerals, so it seems the income rate for each SCV is set to very high and travel time for constructions are very low. In more complex build orders you cannot assume the same income rate (because you will have to use all mineral patches) and you cannot assume either so low travel time for workers (space around comand center is limited). *Maybe* it's something to look into (I had no time to test more build orders), it could explain the feeling about some build orders been impossible to do in the suggested time. If SCFusion already consider all these variables then nevermind.
In general, looks like macrorush is not extremely optimistic about how build order will be executed as SCFusion, maybe it's just a different approach.
About performance, we change some few things to allow paralel processing and the results now are much better. In my tests the speed was improved about 2 to 4 times. Thanks for pointing that something could be improved.
Dragom, the software may try to download:
1. Dot.Net Framework, from microsoft.
2. Updates for the application itself, from my host.
It's safe, no risky at all.
[edit] New version contains also images for zerg and protoss, not only terran!
|
On August 08 2011 00:15 MarcoBrei wrote:
If SCFusion already consider all these variables then nevermind.
It does. It assumes you use the close mineral patches before using the ones further away, but once your SCV count gets higher then obviously you'll be using all the far away ones as well. As for travel times, those too are modelled relatively accurately, depending on what building is being built. most buildings have a travel time of 10s, gas buildings have 2s, CCs/Nexi/Hatches have travel time of 30s. Zerg travel times are about half that as they only have to travel there, not there and back.
|
Independente dos comentários contra, eu aprovei muito esse programa. É praticamente uma ferramenta profissional. Agora para dar certo só precisa da habilidade do jogador. Parabéns para os desenvolvedores e criadores! IoI
|
Please guys, take it to the PM's if you wanna have an argument about whose program is better. It hurts this thread if 2 developers do this to each other.
My POV: I like the appearance of Macrorush, it's way easier to understand than other Buildordergenerators I've seen.
Something with the code seems to be funny, though. I search for Waypoint 1: 3 sentries,
Waypoint 2: 1 Colossus and what I get is this:
Makro BO:
6 Pylon
6 Gateway
6 Assimilator
7 Chrono boost - Nexus
7 Cybernetics Core
9 Sentry
11 Assimilator
11 Sentry
14 Sentry
16 Chrono boost - Gateway
18 Robotics facility
18 Pylon
19 Robotics bay
25 Pylon
25 Colossus
Time: 9:49
Number of Workers: 19
Macro Score: 27
Power score: 19
And why are there different checkpoints If I cant have a target time?
Chekpoint 1: 3 sentries, 1 Colossus would have been the same, right?
|
AvengerMind, thanks. I appreciate you wrote in portuguese, but let's try to keep the conversation in english, because that's a rule in this forum.
Babyschwein, you can use checkpoint to determine the desired order. If you want to build 1 colossus and after that 3 sentries, you will have to use checkpoints, because if you put all in checkpoint 1 the program will probably give you a BO where the sentries are built before the colossus.
New version (released couple days ago) has a little extra performance improvement (I think about 25% better).
|
|
|
|
|
|
EPT
