EPTI've been playing around with probe micro the last few days and decided to post my findings. As many already know, the distance of mineral patches from the main building will alter the amount of income received from workers, and will minor control in the early game, small advanatages can be aquired. Well, this post will describe exactly what number of probes is best to have on how many mineral patches available.
Notes;
-This analysis assumes that the base has 4 close and 4 far mineral patches.
-Numbers used for calculations are a mix of my own observations, and information posted here;
http://wiki.teamliquid.net/starcraft2/Resources
1 - Obvious Observation
The first thing that I managed to confirm is that having the first 6 workers mine close mineral patches provides a 12 mineral per in-game minute advantage over having only 4 workers mine close minerals. Since the first few minutes of the game are spent with a small amount of workers, this information confirms that my early game attention should first lie with what mineral patch my miners are focused on.
This observation remains constant until the 8th worker. At this point, having 8 workers focus on close mineral patches provides 24 minerals per in-game minute over having 4 on close and 4 on far patches, which is a very commonly seen event.
Up to this point, note that the only thing you have to pay attention to is your probes. The first pylon/overlord/supply depot doesn't need to be started until the 8th worker is actually out, and depending on the build/race may even be later than this point.
2 - The Magic Number
The second interesting piece of information that I've pulled out of my analysis is that 8 workers mining close mineral patches remains the ideal number until you have a total of 20 probes mining. The reason 20 is the upper limit here is that you cannot have more than 3 workers per far patch mining at the same time, so after 12 workers are dedicated to far patches, they need to start mining close ones.
I bet many of you are asking; Why 8? Well, the answer is actually quite simple. When 3 workers miner a mineral patch, you'll get an average of 102 minerals per minute regardless of whether the patch is close or far. This means that the advantage of close mineral patches only exists for 1 or 2 workers per close patch.
At 10 workers, having 8 on close and 2 on far will provide you with a mineral advantage of 18 minerals per in-game minute over an even 5-5 split.
At 16 workers, an 8-8 split is ideal and no advantage is generally experienced as the AI will automatically try to balance workers per patch.
At 20 workers, having 8 on close and 12 on far will provide you with a mineral advantage of 24 minerals per in-game minute over an even 10-10 split.
Using workers, excluding chrono and mules, as an example; having all workers up to 8 focus on close minerals and then trying to leave only 8 on close afterwards will generate a total mineral surplus of roughly 60 minerals by the 21st worker.
Chronoboost and Mules have a negative effect on this surplus, especially if you cannot use either but your opponent can. For example, a zerg player will remain with a mineral advantage until the first mule is cast. Also, a zerg player will remain with a mineral advantage until after the second chronoboost (because your worker count can match theirs until this point).
As an extreme example, if your opponent retains the worse possible worker split (would have to be intentional), you have the ability to gain a mineral advantage of roughly 320 minerals. (All 8 early workers on far minerals to 12 workers only mining either close or far to 19 workers split 12 close and 7 far...the later two being the most likely)
Conclusion
I've noticed mineral gains up to roughly 100 by the 20th worker, which is one or two additional units and would actually make a difference. Whether or not you feel your early game focus should be spent on mining, split between mining and early scouting, or spent APM spamming I will leave to you. My primary intention here is to disclose not-so-obvious information regarding close mineral patches (I've seen people force their first 12 workers to mine close as much as possible, which would be bad...you're better off letting the AI split them for you at this point.)
Reference Table;
+ Show Spoiler +
Total probes # close # far Close income Far Income Total Income Difference from best
6 6 0 270 0 270 0
6 5 1 225 39 264 6
6 4 2 180 78 258 12
6 3 3 135 117 252 18
6 2 4 90 156 246 24
6 1 5 45 195 240 30
6 0 6 0 234 234 36
7 7 0 315 0 315 0
7 6 1 270 39 309 6
7 5 2 225 78 303 12
7 4 3 180 117 297 18
7 3 4 135 156 291 24
7 2 5 90 195 285 30
7 1 6 45 234 279 36
7 0 7 0 273 273 42
8 8 0 360 0 360 0
8 7 1 315 39 354 6
8 6 2 270 78 348 12
8 5 3 225 117 342 18
8 4 4 180 156 336 24
8 3 5 135 195 330 30
8 2 6 90 234 324 36
8 1 7 45 273 318 42
8 0 8 0 312 312 48
9 9 0 372 0 372 27
9 8 1 360 39 399 0
9 7 2 315 78 393 6
9 6 3 270 117 387 12
9 5 4 225 156 381 18
9 4 5 180 195 375 24
9 3 6 135 234 369 30
9 2 7 90 273 363 36
9 1 8 45 312 357 42
9 0 9 0 336 336 63
10 10 0 384 0 384 54
10 9 1 372 39 411 27
10 8 2 360 78 438 0
10 7 3 315 117 432 6
10 6 4 270 156 426 12
10 5 5 225 195 420 18
10 4 6 180 234 414 24
10 3 7 135 273 408 30
10 2 8 90 312 402 36
10 1 9 45 336 381 57
10 0 10 0 360 360 78
11 11 0 396 0 396 81
11 10 1 384 39 423 54
11 9 2 372 78 450 27
11 8 3 360 117 477 0
11 7 4 315 156 471 6
11 6 5 270 195 465 12
11 5 6 225 234 459 18
11 4 7 180 273 453 24
11 3 8 135 312 447 30
11 2 9 90 336 426 51
11 1 10 45 360 405 72
11 0 11 0 384 384 93
12 12 0 408 0 408 108
12 11 1 396 39 435 81
12 10 2 384 78 462 54
12 9 3 372 117 489 27
12 8 4 360 156 516 0
12 7 5 315 195 510 6
12 6 6 270 234 504 12
12 5 7 225 273 498 18
12 4 8 180 312 492 24
12 3 9 135 336 471 45
12 2 10 90 360 450 66
12 1 11 45 384 429 87
12 0 12 0 408 408 108
13 12 1 408 39 447 108
13 11 2 396 78 474 81
13 10 3 384 117 501 54
13 9 4 372 156 528 27
13 8 5 360 195 555 0
13 7 6 315 234 549 6
13 6 7 270 273 543 12
13 5 8 225 312 537 18
13 4 9 180 336 516 39
13 3 10 135 360 495 60
13 2 11 90 384 474 81
13 1 12 45 408 453 102
14 12 2 408 78 486 108
14 11 3 396 117 513 81
14 10 4 384 156 540 54
14 9 5 372 195 567 27
14 8 6 360 234 594 0
14 7 7 315 273 588 6
14 6 8 270 312 582 12
14 5 9 225 336 561 33
14 4 10 180 360 540 54
14 3 11 135 384 519 75
14 2 12 90 408 498 96
15 12 3 408 117 525 108
15 11 4 396 156 552 81
15 10 5 384 195 579 54
15 9 6 372 234 606 27
15 8 7 360 273 633 0
15 7 8 315 312 627 6
15 6 9 270 336 606 27
15 5 10 225 360 585 48
15 4 11 180 384 564 69
15 3 12 135 408 543 90
16 12 4 408 156 564 108
16 11 5 396 195 591 81
16 10 6 384 234 618 54
16 9 7 372 273 645 27
16 8 8 360 312 672 0
16 7 9 315 336 651 21
16 6 10 270 360 630 42
16 5 11 225 384 609 63
16 4 12 180 408 588 84
17 12 5 408 195 603 93
17 11 6 396 234 630 66
17 10 7 384 273 657 39
17 9 8 372 312 684 12
17 8 9 360 336 696 0
17 7 10 315 360 675 21
17 6 11 270 384 654 42
17 5 12 225 408 633 63
18 12 6 408 234 642 78
18 11 7 396 273 669 51
18 10 8 384 312 696 24
18 9 9 372 336 708 12
18 8 10 360 360 720 0
18 7 11 315 384 699 21
18 6 12 270 408 678 42
19 12 7 408 273 681 63
19 11 8 396 312 708 36
19 10 9 384 336 720 24
19 9 10 372 360 732 12
19 8 11 360 384 744 0
19 7 12 315 408 723 21
20 12 8 408 312 720 48
20 11 9 396 336 732 36
20 10 10 384 360 744 24
20 9 11 372 384 756 12
20 8 12 360 408 768 0
21 12 9 408 336 744
21 11 10 396 360 756
21 10 11 384 384 768
21 9 12 372 408 780
22 12 10 408 360 768
22 11 11 396 384 780
22 10 12 384 408 792
23 12 11 408 384 792
23 11 12 396 408 804
24 12 12 408 408 816
Questions answered in the thread;
+ Show Spoiler +
On March 03 2011 19:47 sixzeros wrote:
I dont think it means that 20 is saturation, just that above 16 it doesnt matter where those workers settle because they all earn the same so rather than upset a happy 2 worker patch, send them to the far away ones until you hit 20, but you'll still have the opportunity to attain gains after 20.
In fact in my tests I found that the workers above 20 actually mined more efficiently because they seem to stop the aimless wandering around. 21,22,23 and 24 were much more efficient.
So for example if you had 40 workers to assign and were stuck on 2 bases, you'd be better off to group and send 24 to 1 base, and then 16 to the other. You're then guaranteed to have all your 3rd drones mining at their most efficient and its much easier to micro in a hurry.
** edit.. of course at that stage, nothing beats a 3rd. mind you every drone you add after 48 is only about 30% efficient even if you have a 3rd.
I dont think it means that 20 is saturation, just that above 16 it doesnt matter where those workers settle because they all earn the same so rather than upset a happy 2 worker patch, send them to the far away ones until you hit 20, but you'll still have the opportunity to attain gains after 20.
In fact in my tests I found that the workers above 20 actually mined more efficiently because they seem to stop the aimless wandering around. 21,22,23 and 24 were much more efficient.
So for example if you had 40 workers to assign and were stuck on 2 bases, you'd be better off to group and send 24 to 1 base, and then 16 to the other. You're then guaranteed to have all your 3rd drones mining at their most efficient and its much easier to micro in a hurry.
** edit.. of course at that stage, nothing beats a 3rd. mind you every drone you add after 48 is only about 30% efficient even if you have a 3rd.
Saturation occurs at 24 workers, after this number of workers there aren't enough mineral patches to ensure that a probe is almost always mining. Out of curiosity, I calculated the average mineral gain per additional worker;
7th worker adds 16.67% average minerals per minute.
8th worker adds 14.29% average minerals per minute.
9th worker adds 9.26% average minerals per minute.
10th worker adds 8.46% average minerals per minute.
11th worker adds 7.79% average minerals per minute.
12th worker adds 7.21% average minerals per minute.
13th worker adds 8.56% average minerals per minute.
14th worker adds 7.88% average minerals per minute.
15th worker adds 7.30% average minerals per minute.
16th worker adds 6.79% average minerals per minute.
17th worker adds 5.24% average minerals per minute.
18th worker adds 4.96% average minerals per minute.
19th worker adds 4.70% average minerals per minute.
20th worker adds 4.48% average minerals per minute.
21st worker adds 2.45% average minerals per minute.
22nd worker adds 2.38% average minerals per minute.
23rd worker adds 2.32% average minerals per minute.
24th worker adds 2.26% average minerals per minute.
After the 16th worker (8-8 ideal split), the gains start becoming less and less important.
With 48 workers the ideal income would come from 3 bases, each having 16 workers (minerals only, of course). This would provide roughly 2016 minerals per minute over the two bases with 24/24 which would provide 1632 minerals per minute.
On March 03 2011 04:08 whatthefat wrote:
Very interesting, but how many of these potential mineral gains are lost by the time required to force workers to mine from a close patch (which typically involves them sitting still for a moment while you spam click on the patch)? My guess is not a lot, but it would be interesting to actually measure.
Very interesting, but how many of these potential mineral gains are lost by the time required to force workers to mine from a close patch (which typically involves them sitting still for a moment while you spam click on the patch)? My guess is not a lot, but it would be interesting to actually measure.
A worker on a far patch mines on average 39 minerals per in-game minute. A worker on a close patch mines on average 45 minerals per in-game minute.
The in-game second equivalent is 0.65 and 0.75 respectfully.
If it takes 5 in-game seconds to set the worker, then you lose 0.65*5 = 3.25 minerals and gain 45-39 = 6 so your net gain is 2.75 minerals per minute. Still worth the effort. Keep in mind that the better you get at microing workers, the least time is wasted.
(meaning it's not hard and I just suck at it).
.
