EPT[D] harvester optimization - Page 4
| Forum Index > StarCraft 2 Strategy |
Liquid`Sheth
United States2095 Posts
| ||
|
LuDwig-
Italy1143 Posts
My experiment is made at the natural expansion @5 o clock of Kulas Ravine. I have made it with 24 workers and I start mining in the same way for both experiments (the 24 workers start mien together) Here are my results. I take note of mineral count every 30 seconds. ![[image loading]](http://img504.imageshack.us/img504/6474/graphn.jpg) As you can see the "+1 expansions" not only mine at the same time of the "normal expansion" but in 8 minutes it takes an advantage of 240 minerals | ||
|
Arcalious
United States213 Posts
| ||
|
CruS
Sweden218 Posts
On April 21 2010 04:25 Arcalious wrote: I tested this method using MULEs and it had a negative effect. Not a good method if you plan to use MULES. I also feel the small gain in mineral collection is cancelled by the small loss in gas collection. If anything, this test shows that it won't really hurt if for some reason you missplace your building. Yes it has a dramatic effect on MULEs, however the gas issue was nothing to be bothered with. barely 1 per minute? I know my hatch will be going down on +1 more often than not thanks to this guy. | ||
|
kawoq
Guatemala357 Posts
On April 21 2010 04:25 Arcalious wrote: I tested this method using MULEs and it had a negative effect. Not a good method if you plan to use MULES. I also feel the small gain in mineral collection is cancelled by the small loss in gas collection. If anything, this test shows that it won't really hurt if for some reason you missplace your building. I was thinking just that while I read the 4 pages, since terrans will for sure use mules I think it wont be good to misplace the CC since it will make them have to travel more and they don't have to wait for a free patch so no need for the misplaced CC. I think Zerg and Protoss can make more of this. Great find and congratz for the key you win for this post, you deserve it! PD: just curious, wouldnt this logic apply to Broodwar too? have anyone tryed? | ||
|
Silver777
United States347 Posts
| ||
|
spinesheath
Germany8679 Posts
On April 21 2010 04:40 kawoq wrote: PD: just curious, wouldnt this logic apply to Broodwar too? have anyone tryed? Nope, it doesn't work because harvesters in BW don't wait at all, while in SC2 a worker will wait at a mineral patch if the worker that is currently mining is at least 50% (or some value similar to that) done. SC2 has a buffer for slightly varying distances because of that. | ||
|
SichuanPanda
Canada1542 Posts
| ||
|
Amazn
United States83 Posts
| ||
|
McCain
United States187 Posts
On April 21 2010 04:56 Silver777 wrote: This seems like a great idea for toss, terran it seems inefficient due to mules mining over SCV's, as for zerg I am curious if the slight speed increase still allows for the square back mining to be efficient or will they reach their a split second to soon? Drones don't get a speed bonus on creep, so it should be fine. | ||
|
folke123
Sweden133 Posts
the optimal amount of workers on minerals in your main is NOT 24? 3 at each patch? but like 21? 3 at the 5 patches farther away and 2 at the 3 closer patches? Do I get a negative income of building 3 at each or just no more income of them? and then on an expansion if I put it down 1 step back, I would get the maximum minerals with 3 on each patch? | ||
|
Aether
Canada123 Posts
On April 21 2010 03:26 Full wrote: I just tested 15 drones with further distance vs 13 drones at closer distance This was on high yield. They both finished at the same time. Not sure if i should have done more drones, but i can link replay if u want. Yeah, you need more workers to make use of that difference. What's important to focus on is not the efficiency of the workers. This is relatively unimportant. It's the efficiency of the mineral node, ie: is it being mined every possible second. Full saturation on a normal placement should be 15/14 workers vs 18 on the farther end. If you do 18 vs 15 the difference is more like a minute and a half. | ||
|
Rufio52
United States15 Posts
 I've heard lots of people agreeing that this can help for Protoss and Zerg on gold expos, but I think Terran is the one race that can really use this to their advantage on a large percentage of their games. For Terran, it seems like it would best be used almost the exact opposite as P or Z would use it. It would be bad to do this on a gold expansion because that's where all the MULEs should be going, and if they make one less trip then it's a big loss. But because terran have the unique ability to move buildings, they are the only ones who can take advantage of this on any and every base each game, as long as they have enough SCVs and leave a single base at the normal distance for MULEs. For this hypothetical game example, I'm going to completely ignore attacking/defending and just focus on mining efficiency: So let's say it's mid-game and we are playing Terran with whatever strategy (doesn't matter). We have our main and natural at normal distances from the minerals. We have our eye on a gold expansion, so we are going to be creating extra SCVs to get ready to transfer them to the gold expo. With this extreme saturation and an extra base to dump all the MULEs, this player can lift and land his natural CC one space away to increase effeciency and income until the gold base gets up and running. Why move the natural, you ask? Because there are more minerals there at this point in the game, and we are about to have the gold base to spend the MULEs on. We can save the MULEs temporarily for the gold base. Remember, we wouldn't be moving the CC back if we weren't just about to take the gold. After the gold base is up and running at the normal distance for MULEs (and a Planetary Fortress since this base won't be moving), a bit later on in this same game, we are under pressure and need a little boost in econemy but can't expand. So instead we just build more SCVs at our main and move the CC one space further back so we can get the mineral boost. Let's pretend that our opponent gets some excellent harrasment at our natural and kills almost all of our SCVs there. We already have extra SCVs and simply move some of them from the main to the natural and move both CCs back to their original positions. This way, even if we lost a huge number of SCVs it doesn't hurt our economy too much (or at least as much as it normally would). A bit later and we get a hidden expantion at a different base. Because we can't get enough SCVs right away, we should position the CC at the normal spot. Then maybe our main starts running out of minerals (because we've been mining faster than normal) or we just get enough SCVs produced there that it gets over saturated at the normal distance, so we pick up and land one space away. All the while sending all our MULEs to the gold base so we get as much income as efficient as possible. This can go on in a pattern; As long as we have one base at the normal distance to spend MULEs on (preferably a gold base), we can make all other bases one space out, provided sufficient SCVs. So far in this game we are mining/have mined three bases faster than normal which can add up to a significant increase in income. Not only that, but we weren't forced to have less than optimal mining at any point in this game, with or without worker harass, like Protoss or Zerg would have. Now this optimization may have helped, but then we get into some grey areas I'm still wondering about. Part of me feels like this may be over-analysis, but another part says this may save my ass if it comes up and I know what to do. Back to our example game: Once again a little later in this same imaginary game, we lose the gold base but we aren't out of the game yet (maybe some excellent harasment, base trade, whatever). We are able to save most of the SCVs that were there, and they move to our main & natural to keep the hidden base secret. Now what? Move one base closer for MULEs? Move both closer since we are going to have ridiculous over-saturation and SCVs are going to be dancing regardless? Move neither for the same reason (over-saturation)? This is of course assuming we have the time and presence of mind to do anything, and the enough pressure that we can't expand. In essence, once you get over 3 workers per patch, does it cease to matter what position the HQ is in? Super-saturation (more than 3/patch) may seem extremely unlikely to happen, but if this is a sensible strategy then a lot of time we are going to have more workers compared to the normal now, so it would be more likely to have super-saturation. And one final question outside of any hypotheticals: Does the gas mining not change even if the minerals make a corner and the gas is in the middle? So the HQ is one space vertical and one space horizontal off of the normal position, and the gas is as far away as possible? On April 21 2010 06:41 folke123 wrote: Just want to see if I get this right. the optimal amount of workers on minerals in your main is NOT 24? 3 at each patch? but like 21? 3 at the 5 patches farther away and 2 at the 3 closer patches? Do I get a negative income of building 3 at each or just no more income of them? and then on an expansion if I put it down 1 step back, I would get the maximum minerals with 3 on each patch? To answer what I can: The point of the OP is that you can get more efficient use out of 24 workers if the HQ is placed one space further back than normal. If you can't move the base (Z & P & Planetary Fortress), then you rapidly start to loose worker efficiency once you go over 2 at the close mineral patches and 3 at the farther ones. The problem with 3 workers at the close patches is that one of the workers will get to the patch too soon and switch patches and spend too much time trying to find an open patch when really it would have been quicker to just wait it out at the first patch. Even if a base is one space back then with 3 workers there is near max patch efficiency. I can't test these myself as I don't have a Beta key, but I have been following as much of what is happening as I have time for. And if you can't tell, I love the super technical. :D | ||
|
Aether
Canada123 Posts
On April 21 2010 04:25 Arcalious wrote: I tested this method using MULEs and it had a negative effect. Not a good method if you plan to use MULES. I also feel the small gain in mineral collection is cancelled by the small loss in gas collection. If anything, this test shows that it won't really hurt if for some reason you missplace your building. Yeah, I suspected this might be the case. One possible way terrans can make use of this though is to use normal placement on high-yields, but on a normal expo, lift off and land 1 space farther once a normal mineral field is fully saturated, only using their MULEs on the normally placed CCs. So if your natural is saturated, you have a high-yield, you would use all of your MULEs on the high-yield, land the CC at your natural one space farther, and you should see gain as a result, since the MULEs are acting independently of the misplaced CC. I don't believe there is a loss in gas, the reason being that with normal placement the workers have to wait a split second to enter the guyser, so by moving it only one square farther, the travel time eats into the waiting time BEFORE it eats into the mining time. | ||
|
Aether
Canada123 Posts
Good observation Rufio. | ||
|
MementoMori
Canada419 Posts
| ||
|
fantomex
United States313 Posts
| ||
|
MementoMori
Canada419 Posts
On April 21 2010 09:44 fantomex wrote: This newb question but: way do I see players maynard 4 drones to expansion when they have less than 16 mining at their main (in many cases way under 16... like 12). Is it so they can just mindlessly spam drones at both bases or is there some advantage that makes up for all the minerals lost while they're moving from base to base? It's because (at least from my understanding) even though in theory you could have 2 on each mineral patch in the main, it's not so easy to make them actually mine like that. They'll probably be dancing around and losing time when it would be better to just send em over to the expo. | ||
|
Aether
Canada123 Posts
As for the point about the liftoff time, I did consider that, Not sure how it would impact the overall time. I can try testing that as well. | ||
|
Rufio52
United States15 Posts
On April 21 2010 10:19 Aether wrote: I'm going to do some more thorough tests on whether gas is impacted or not. My game is minimized right now running tests. As for the point about the liftoff time, I did consider that, Not sure how it would impact the overall time. I can try testing that as well. I can't test it, but I would think that it's not more than 3 ~ 4 seconds of lost mining time. Surely if you end up moving a lot back and forth then there could be some significant lost time, but as long as you switch positions at a very deliperate pace and don't rush things too much, it would seem to be worth it in the long run. | ||
| ||
