|
After some thread about map imbalance because of the minerals distribution (luna, r-point), I started to think about a symmetric distribution and take a look over the expos in Neo Forte, the ones with six mineral patches distributed in two opposed groups of three patches each one.
I realize something interesting about this distribution, usually the optimal number to gather minerals is three workers per patch with this every patch will have a worker mining all the time above this number you are wasting minerals in workers.
Turns out that the optimal number for these Neo Forte expos are exactly 12 (two workers per patch). Why is this? Because of the different distribution the problem of the wandering workers seeking for free patches to mine is eliminated, this problem causes that at some point a worker waste more time travelling to a free patch than actually mining.
Therefore the best places to expand in this map are these expos to get fast minerals with less workers, you could say the mains have more minerals (10 patches) but while you put more workers to gather the problem of the wandering workers will appear and you will need more or less 30 workers for optimal mining rather than 24 in two of these expos working 12 patches optimally. The expos at the top and bottom of this map don't have the same behavior but are close too.
How could this help a race with the capacity to make 12 workers fast? Make your experiments zergs and let us know.
This could be an interesting factor for making maps or maybe not, but there shouldn't be imbalances in mining because of the starting positions.
Enjoy your minerals.
|
arent u suppose to make unlimited scvs/drones/probes? including forte? and i always thought the mineral balance issues was on the main where one main u gather slower, i could be wrong though
|
Unlimited workers is a wrong idea, pros stop worker production with a "good number" not optimal but a good one to produce more army and make timing strikes.
|
intresting but I think the avg player just keeps trying to pump workers the whole game anyway. I know I'm far to slow a player to count workers and get everything I need to done.
|
On May 15 2006 23:23 antrax wrote:
Unlimited workers is a wrong idea, pros stop worker production with a "good number" not optimal but a good one to produce more army and make timing strikes.
i dont mean as "unlimited" but a lotta workers of course u dont get 200 workers outta one base
|
This seems like it might benefit toss in PvT, just whoring expands rather than probes (since the general rule of thumb is to have 2 expand to terrans 1, or at the very least stay 1 expo ahead). Rather than investing money in the 13th-20th probes at one of your exps, drop another nexus at another nat on forte, and then transfer 12 probes from main/nat/whatever.
Very interesting.
|
can you explain how to solve the "wandering problem"?
i tested a neo forte game in singleplayer and put 30 probes to mine the main, but they was walking back and forth alot, even after 5min
|
On May 15 2006 23:02 antrax wrote:
Turns out that the optimal number for these Neo Forte expos are exactly 12 (two workers per patch). Why is this? Because of the different distribution the problem of the wandering workers seeking for free patches to mine is eliminated, this problem causes that at some point a worker waste more time travelling to a free patch than actually mining.
Therefore the best places to expand in this map are these expos to get fast minerals with less workers, you could say the mains have more minerals (10 patches) but while you put more workers to gather the problem of the wandering workers will appear and you will need more or less 30 workers for optimal mining rather than 24 in two of these expos working 12 patches optimally. The expos at the top and bottom of this map don't have the same behavior but are close too.
Just interested to know.. How did you "experimentally come to this conclusion"? Not disbelieving you but did you hear it from somewhere, or you did an experiment on it to validate this. Also if you did the experiment, can you post the results of your experiment so we know how you come to that conclusion from your numbers? Thanks.
|
Effiency isnt always that siginificant, if it was people would just get 1 worker per patch then expand again; the maximum mineral flow from the mains is much higher even if you need more workers. Maybe a base just got mined out and you have spare workers, in this case youll be gutted if you only have 6 mineral blocks and 20 probes free.
The main minerals will also last longer.
|
thedeadhaji
39489 Posts
I would guess a brute force experiment would be to use 12 workers and mine for 10 minutes, repeat with 13, 14, 15, etc and gather data, and repeat a few times with each and average out the values.
|
On May 16 2006 03:02 thedeadhaji wrote:
I would guess a brute force experiment would be to use 12 workers and mine for 10 minutes, repeat with 13, 14, 15, etc and gather data, and repeat a few times with each and average out the values.
That's what I'm asking about. Was it done like this or was it just a theory he heard from somewhere or made up. =(
|
Ummm, it seems that this issue is not well known, I will try to describe the problem in a formal way. Stay tuned ...
|
and whats the dealwith the gas @ 13 ?!
|
basically by the time I take my 3rd expansion my main should be out of minerals, so those probes go there, 4 bases is plenty, and if you have like 80 or some probes you will be outmassed. Usually in late game im seriously working 4-5 expansions while the rest have less probes.
|
Norway28706 Posts
really interesting find and a good post
too bad most posters seem unable to appreciate it, first strategy post posted on this site I have ever learned anything from though
so do experiment more
and yeah, this basically means that if you are terran and fast expanding on forte, protoss and fast expanding on forte or zerg and fast expanding on forte, you can get maximal mining efficiency with 6 probes / scvs / drones less than what you would normally assume was the case.
|
It's definitely a good post, but like any other new discoveries, it needs at least some validation or else anyone can come up with anything that looks new but aren't real.
|
It makes sense, i have been thinking about this aswell, since there is 3 on each side they cant travel so its only needed 2 on each patch to be one on the patch at all times
seems correct to me
|
Ah I remember the wgtour dossier on this. It really makes sense however and I will have to keep it in mind. I suppose for protoss and terran it will be more of a waste as zerg might not need as many drones in one area.
|
On May 16 2006 11:19 Liquid`Drone wrote:
really interesting find and a good post
too bad most posters seem unable to appreciate it, first strategy post posted on this site I have ever learned anything from though
so do experiment more
and yeah, this basically means that if you are terran and fast expanding on forte, protoss and fast expanding on forte or zerg and fast expanding on forte, you can get maximal mining efficiency with 6 probes / scvs / drones less than what you would normally assume was the case.
Thx for the comment ...
Years ago i was trying to solve the problem of the wandering worker but nothing really worked. For me this is more a defect of the gathering AI, don't know if programmers didn't notice or if there were computational costs involved (you seek once the closest free patch and go, calculating it all the time could be really costly and multiply it for every worker ...).
Now with this kind of mineral distribution this problem is almost zero and if map makers adopt it for new maps the game-play can change in a significant way. Moreover the same symmetric distribution for every starting point represents fairness for every player.
Some people could argue that an important aspect of the game is simplified, the classic non-stop workers now would be a fixed number (2 per patch) but still you have to expand, you can harass eco, etc., etc., i guess the time will tell us if this is huge or not.
Finally i am afraid that protoss can take advantage of this because of his building technology. Or that terran take the worst part.
Here is a map with 2 possible distributions with a total of 8 patches per starting point, still the wandering worker reappears but is minimal i think you could considered 2 worker per patch optimal.
http://rapidshare.de/files/20631932/antrax_distribution.scm.html
1 vs 1 ? 
Enjoy it and let's see if some mapmaker decide to use it.
|
Even given that zerg can actually mine his natural to the capacity that terran's and protosses can, I don't think zerg is going to be able to compete very well with terrans and tosses on this map. Being able to fast expand by walling off and since zerg's are forced to build a 2nd hatch before expoing to defend both main and nat, thus putting them in a position VERY weak to fast wraith or tank drop, more than compensates for this, unfortunately. But useful information still, thanks for sharing~~
|
ok i just tested out neo fortes mineral efficiencies.
here are the results:
Test One-
Main base with 20 miners (2 on each patch):
@5 minutes: aprox. 5000 minerals.
@10 minutes: aprox. 9730 minerals.
mined out at 16 mins 53 secs.
Natural Expo with 12 miners (2 on each patch):
@5 minutes: aprox. 3776 minerals.
@10 minutes: aprox. 7616 minerals.
mined out at 12 mins 5 secs.
Test Two-
Main base with 30 miners (3 on each patch):
@5 minutes: aprox. 6100 minerals.
@10 minutes: aprox. 12920 minerals.
mined out at 12 mins 20 secs.
Natural Expo with 18 miners (3 on each patch):
@5 minutes: aprox. 4200 minerals.
@10 minutes: aprox. 8500 minerals.
mined out at 10 mins 44 secs.
conclusions:
As you can see having more miners on 6 patches speeds up intake of money. putting more miners on 10 patches speeds up intake of money. ill let someone else find the line where the bonus begins and stops...
adding 10 miners to your 10 patches increases your intake of money slightly at the 5 minute mark and significantly at the 10 minute mark.
adding 6 miners to your 6 patches slightly increases your intake of money at the 5 minute mark and slightly increases your intake of money at the 10 minute mark.
adding 6 miners to your expo makes it mine out roughly 2 minutes faster than the original 12.
adding 10 miners to your main makes it mine out roughly 5 minutes faster than the original 20, well worth the 500 minerals put into the extra miners.
here are the average minerals mined per minute for both bases with the miner number changes.
(MPM=minerals per minute)
-=5 minute mark=-
Main-20 miners: 1000 MPM
Main-30 miners: 1220 MPM
Expo-12 miners: 755 MPM
Expo-18 miners: 840 MPM
-=10 minute mark=-
Main-20 miners: 973 MPM
Main-30 miners: 1292 MPM
Expo-12 miners: 762 MPM
Expo-18 miners: 850 MPM
Having more miners per patch with a max of 3 [per patch] can give a decent boost in economy if thought out correctly during a game.
the major bosst appears around 10 minutes of mining.
someone else can do a test to see if having 4 miners per patch is worth the extra 500 minerals.
I hope this helped somebody...
BTW: this only applies to Neo Forte for those who did not read the other posts.
if someone REALLY wants to get accurate results they can retime the mineral agtherings in the Replay. i went AFk for like 20 minutes in the replay also.
http://www.yousendit.com/transfer.php?action=download&ufid=93F17BBA25BA57AB
|
we should have a map contest to see the most effiecient mineral layout ^___^.
|
I didn't know that 3 miners on each patch is the msot effecient. For me its always been 2 miners per patch is very close to optimum.
I would stick to roughly 2 or a bit more miners per patch because the extra scvs are not worth it because of their costs compared to the increased economy.
|
Nice find, wonder if the pros know...
|
Thanks for testing Hipnotize, it's true that you can add miners and this will improve a LITTLE your income but what about this factors:
The cost of the extra miners, a difference of 8 miners means 400 minerals
The supply that these miners consume, one pylon means 100 minerals
The time that takes to produce these extra miners.
My point is how fast can we achieve an optimal rate of gathering and then concentrate resources in army 500 extra minerals sounds interesting in early stages of the game.
skyglow 2 miners per patch should be true that are close to optimum but because of current mineral distributions the problem that I explained degrades the final income.
|
On May 16 2006 16:23 Hypnotize wrote:
we should have a map contest to see the most effiecient mineral layout ^___^.
fastest possible map?
|
On May 16 2006 18:28 YoUr_KiLLeR wrote:Show nested quote +On May 16 2006 16:23 Hypnotize wrote:
we should have a map contest to see the most effiecient mineral layout ^___^. fastest possible map?
that is about as efficient as it gets 
|
On May 16 2006 16:20 Hypnotize wrote:
Test One-
Main base with 20 miners (2 on each patch):
@5 minutes: aprox. 5000 minerals.
@10 minutes: aprox. 9730 minerals.
mined out at 16 mins 53 secs.
Natural Expo with 12 miners (2 on each patch):
@5 minutes: aprox. 3776 minerals.
@10 minutes: aprox. 7616 minerals.
mined out at 12 mins 5 secs.
Test Two-
Main base with 30 miners (3 on each patch):
@5 minutes: aprox. 6100 minerals.
@10 minutes: aprox. 12920 minerals.
mined out at 12 mins 20 secs.
Natural Expo with 18 miners (3 on each patch):
@5 minutes: aprox. 4200 minerals.
@10 minutes: aprox. 8500 minerals.
mined out at 10 mins 44 secs.
-=5 minute mark=-
Main-20 miners: 1000 MPM
Main-30 miners: 1220 MPM
Expo-12 miners: 755 MPM
Expo-18 miners: 840 MPM
-=10 minute mark=-
Main-20 miners: 973 MPM
Main-30 miners: 1292 MPM
Expo-12 miners: 762 MPM
Expo-18 miners: 850 MPM
Now that's a valid experiment. No offense antrax, cos I think it's only fair that numbers do the talking and 100 minerals per minute boost isn't little (Equivalent to an peon going 12 trips). Good post though, everyone learns from this thread.
|
On May 15 2006 23:02 antrax wrote:
The ones with 6 mineral patches distributed in two opposed groups of three patches each one.
I realize something interesting about this distribution. Usually the optimal number to gather minerals is 3 workers per patch with this every patch will have a worker mining all the time above this number you are wasting minerals in workers.
Turns out that the optimal number for these Neo Forte expos are exactly 12 (two workers per patch). Why is this? Because of the different distribution the problem of the wandering workers seeking for free patches to mine is eliminated, this problem causes that at some point a worker waste more time travelling to a free patch than actually mining.
Therefore the best places to expand in this map are these expos to get fast minerals with less workers, you could say the mains have more minerals (10 patches) but while you put more workers to gather the problem of the wandering workers will appear and you will need more or less 30 workers for optimal mining rather than 24 in two of these expos working 12 patches optimally. The expos at the top and bottom of this map don't have the same behavior but are close too.
How could this help a race with the capacity to make 12 workers fast? Make your experiments zergs and let us know.
This could be an interesting factor for making maps or maybe not, but there shouldn't be imbalances in mining because of the starting positions.
Enjoy your minerals.
Can't believe you guys actually think hes right with no proof or how he came to this conclusion. I've actually tested optimal number of workers for 3 patches and its 8. My friend was saying 3 workers per patch is the rule of thumb also, its actually 2.5. Knowing that you can use this formula; 3x2.5 is 7.5 = 8 . Since there is 2 sets of 3 that means 16 workers is the most optimal.
The other thing that doesn't make sense about his post is, Couldn't you just as easily take a base with 10 crystals and put 12 workers on that and get just as much money, if not more from less waiting time?
|
Oh, Hypnotize sorta already beat me to it. But still its 2.5 Workers per crystal not 3. That is the most efficient way.
|
Generally common sense doesn't need proofs, but it seems that this is not the case.
I made experiments and guess what it's true.
I have to analize the results to give a better explanation or interpretation, but my experiment is simple put 12 miners in the new distribution, put 12 in the mineral expansion both have 6 patches, who do you think end first?
And of course continue with another experiment, put 24 miners in a main (10 patches) and take two expos (12 patches) and 12 miners per expo, start mining at the same time.
Who will end first?
Explain ... if you can.
|
Hm I did not discover a big difference between close (to each other) and loose mineral placement. No real stats though, my measurements were not accurate enough 
|
|
|
|
|
|
EPT
