EPT[D]9 Overlord or 10 Overlord? - Page 2
| Forum Index > StarCraft 2 Strategy |
|
threehundred
Canada911 Posts
| ||
|
Tipton
Sweden13 Posts
| ||
|
goszar
Belarus119 Posts
| ||
|
saritenite
Singapore1680 Posts
On April 07 2010 21:25 flothefreak wrote: Why was it 9ol in SCBW anyway? I wondered all the time why terran went 8or9/10 depot, protoss 8/9 pyl and only zerg made 9/9, waiting for the cap.. It's the Larvae system, you can produce 2 drones immediately after your OL pops while having a drone on every patch of minerals gathering for a 1) quick pool, or 2) 12 hatch build. | ||
|
Thamoo
Canada234 Posts
| ||
|
Klive5ive
United Kingdom6056 Posts
Here's my measurements from replays, feel free to check this: With 9 drones it takes 14 tics to mine the extra 100 minerals needed to get an overlord on 9. Drones mine at 5 minerals every 7 tics Extractor trick costs effectively 10 minerals (it's actually 6 but you can't spend the extra 4 unless you cancel another building). 10o This build wastes larvae time, setting all your future drones back. The result is less minerals in the long run. However at the 10 larvae stage this build has the most minerals. Therefore it will get a 10 pool down faster. If you plan a fast pool or something crazy like a 12hatch this would be the correct build. 10o+e Vs 9o Neither of these builds waste larvae time. 9o gets it's 10th drone 100 minerals later. Effectively that's about 14 tics. 1 drone would mine 10 minerals in that time. The 10o+e loses an effective 10 minerals on the extrick so it's immediately equal. Factor in the lost mining time on the extrick drone and the later 12th drone and you can see this build is worse. Conclusion: 9o is the strongest economic opening for Zerg. This should always be done before the popular 15pool 16hatch. | ||
|
goszar
Belarus119 Posts
On faster speed, drone mines 1 mineral per second on average. Drone build time is 12.75 seconds; overlord - 18.75 seconds. Builds start in equal conditions - 8 drones are mining, 9th is morphing, we have 50 minerals 1) You morph another drone now or wait 6.25 seconds to morph overlord, then 6.25 seconds more to morph drone, losing 12.5 minerals in mining time for 9o build. 2) in 10o, after overlord you wait for 75 minerals during 8.5 seconds (9 drones mining) to do extractor trick, losing 10 minerals + about 3 in mining time if you are quick (you can't use 4 minerals as Klive5ive pointed out). So far both builds are pretty even. 3) in 9o build, you get 11th drone 12.5-8.5 =4 seconds later than 10o+ExtTr, 12th drone 6.25 seconds faster (because of faster ovie). You don't have minerals for 13th drone for 5 seconds (10 drones mining), so it's 1.25 seconds faster. 4) In 10o+ExtTr, you have minerals for 12th and 13th right after ovie (well, sometimes it is 99), but they come later because for 12.5 seconds later ovie as described in #3. Conclusion: 9o build is ahead by about 3.5 minerals compared to 10o+Extractor trick, assuming player loses abut 3 seconds in mining time doing the trick. I think these builds were designed to be equal, but you lose those odd 4 minerals you can't spend. Table for explanations: ![[image loading]](http://img263.imageshack.us/img263/7553/9ovs10oe.jpg) | ||
|
Klive5ive
United Kingdom6056 Posts
On April 08 2010 23:17 goszar wrote: No one seems to actually have calculated the difference. Let's do this from pure theoretical standpoint, ignoring the scout. On faster speed, drone mines 1 mineral per second on average. Drone build time is 12.75 seconds; overlord - 18.75 seconds. Builds start in equal conditions - 8 drones are mining, 9th is morphing, we have 50 minerals 1) You morph another drone now or wait 6.25 seconds to morph overlord, then 6.25 seconds more to morph drone, losing 12.5 minerals in mining time 2) in 10o, after overlord you wait for 75 minerals during 8.34 seconds (9 drones mining) to do extractor trick, losing 10 minerals + about 3 in mining time if you are quick (you can't use 4 minerals as Klive5ive pointed out). So far both builds are pretty even. 3) in 9o build, you get 11th drone 12.5-8.34 ~ 4 seconds later than 10o+ExtTr, 12th drone 12.5 seconds faster (because of faster ovie). You don't have minerals for 13th drone for 5 seconds (10 drones mining), so it's 7.5 seconds faster. 4) In 10o+ExtTr, you have minerals for 12th and 13th right after ovie (well, sometimes it is 99), but they come later because for 12.5 seconds later ovie as described in #3. Conclusion: 9o build is ahead by about 15 minerals compared to 10o+Extractor trick, assuming player loses abut 3 seconds in mining time doing the trick. The trouble with your analysis is that "1mineral per second on average" an inaccurate way of measuring income. The mathematical modelling required to accurately calculate what happens is deceptively complicated. Before I made my post I started an excel sheet that would add up the contribution of each drone on different patches but realised it was an incredible amount of work and effort. However, you cleverly identified all the differences between the two builds and estimated the effect. The important thing is we both came to the same conclusion. Edit: I don't think it's close. 9o is better. As I said before the cost of the extrick alone is more than the lost mining time on drone10. | ||
|
goszar
Belarus119 Posts
"1mineral per second on average" comes from the article of SC2 theorist I fully respect (I can give you links if you understand Russian). Also my calculations were off, it's actually 3.5 minerals difference. So I assume the builds are equal if you ignore odd 4 minerals. That theorist promised to do the guide on Zerg builds, i'm waiting for it. He has already done Terran one, the biggest conclusions he made is that 9Rax -11 depot - orbital is equal to 10 depot -12 rax- orbital; and the best FE build is 9Rax-11depot-14CC. | ||
|
[F_]aths
Germany3947 Posts
| ||
|
TanukTanukTuk
United States40 Posts
In essence I made my experiment reproducible. Now, when I was testing 10 overlord + extractor trick, I always sent a drone at 65 minerals to create the extractor. I guesstimated this timing was most efficient because it makes the extractor right when you get 75 minerals, so you can make a drone and immediately cancel. Using this method and performing six trials for each build, I found that 9 overlord and 10 overlord+extractor produced similar results. 9 overlord seemed to always have 165 minerals at the 1:50 mark whereas 10 overlord+extractor had between 164 minerals and 169 minerals, which is to say the build are about the same. I will note that in every experiment (replay) 10 overlord+extractor hit the 200 (204 actually for 10o+e) mineral mark around 1:55 whereas 9 overlord hit it around 1:57. This is because at the I think I happened to choose a poor time using the 1:50 mark wherein a drone or 2 was about to return minerals but did not have the chance to. I really can't think of anything else to which to attribute this discrepancy. All in all, I don't think this is really worth nitpicking over as your split as well as the arrangement of the minerals at which you spawn, as well as making sure you're sending drones to the correct patches probably has a much greater influence on the speed and timing of your build. Hope this answers your questions. edit I realized I did not try doing 10 overlord no extractor (bad scientist!) assuming that people on the forums were correct (again bad assumption  ) and found that actually it was actually no worse than the other two builds having between 165-170 minerals. So I guess build however you feel like? | ||
|
Klive5ive
United Kingdom6056 Posts
On April 19 2010 17:04 TanukTanukTuk wrote: All in all, I don't think this is really worth nitpicking over as your split as well as the arrangement of the minerals at which you spawn, as well as making sure you're sending drones to the correct patches probably has a much greater influence on the speed and timing of your build. . I think this is a very important point. In BW when you watch the pros play you often see them microing 5 workers onto 4 minerals patches early game; maximising the closest mineral advtantage. This is even more important in SC2. Here's something I noticed about SC2. If early game you try to send a worker to a patch that is already being mined. If it doesn't hit that perfect timing where it creates a pair with the first drone it will always move to the closest free mineral patch EVEN if it has to move past all the patches to get there, and those patches are faster mining. It wont even TRY to mine from the other patches it just ignores them and moves to the free patch. With that in mind knowing which patches are the fastest and setting up your "mining pairs" early game, is going to give you a decent advantage in SC2. As for "10 Overlord is better for SCoverpool". That is true BUT it might also be better for 13 pool too. If you're planning on waiting on 15 for your pool to finish so you can make 6 lings then the larvae advantage you gained from 9ol might not actually be worth it. I think we maybe need to look closer at 10ol since people made this assumption about 8ol in BW, assuming it was better. But then people realised since they were 12 hatching 11 pooling the larvae advantage was not significant. | ||
|
G4MR
United States371 Posts
| ||
|
HeyheyLBJ
Sweden160 Posts
Each drone up to 16 will mine with the same efficiency, namely very close to 0.69 minerals per (game) second. I therefore chose to let the builds converge at 16 workers, a Pool, and two Overlords (Pool built at 14 supply, OL at 16). I have the exact time for everything, but I chose to focus on these timings: * 16 workers * Second OL * Pool completed * First larva spawn after Pool Note that these results are not subject to any personal skills with splitting workers or anything of the sort. About the fractions mined, think of them as the positions of the drones. Ie. 1090.83 mined means that 1090 have been collected, and the position of the drones are valued at 0.83 extra. If they all hung out at the hatchery doing nothing, that value would be 0.00. The results: 9OL: 16 workers @ 154s. OL @ 152s. Pool @ 173s. Larva @ 182s. This build will have mined 1090.83 @ 154s. 1299.26 @ 173s. 10OL: 16 workers @ 156s. OL @ 154s. Pool @ 173s. Larva @ 184s. This build will have mined 1086.03 @ 154s. 1293.77 @ 173s. 10OL with extractor trick: 16 workers @ 154s. OL @ 152s. Pool @ 174s. Larva @ 182s. This build will have mined* 1085.51 @ 154s. 1293.94 @ 173s. *6 minerals removed from the mining total because of lost minerals due to cancelling extractor. 8OL: 16 workers @ 158s. OL @ 156s. Pool @ 156s. Larva @ 186s. This build will have mined 1070.94 @ 154s. 1277.32 @ 173s. Graph: http://pici.se/pictures/UFCSyZTIM.jpg (X-axis is game seconds, Y-axis is total mined). Conclusion: 9 OL is tied with 10 OL (ext trick) when it comes to the different timings (except the Pool which comes 1 second later for the 10 OL (ext trick) build). But it has mined more at all those times when we've accounted for that 6 minerals were lost due to extractor cancel. All in all, 9 OL is slightly superior in every way. | ||
|
goszar
Belarus119 Posts
I have recently seen the following build in Zotac cup played by some Oger guy: 10 Drones Extractor trick Overlord 3 Drones when overlord pops (one is usually 1 second late) Spawning pool (it is at 14). The other player did usual 10Ol and was 50 minerals behind and about 2 seconds behind on 14 drones. I have watched several times - they were equal up to 10 drones. My calculations show that this is superior even to 9o by small amount assuming you are going for 14 pool. | ||
|
Hammy
France828 Posts
Here is the test I conducted: I played three games, using three different build order "tactics". To standardize the test, I never split my drones at the beginning. I just clicked on the nearest mineral patch and let them do their work (this test is about comparing builds amongst each other, which requires as much normalization as possible). Also, the general build older was always: -13 pool -14 extractor -16 queen -18 zergling (just the one, since it's standard to chase the enemy worker if you want to expand, or just to scout) -Speed at exactly 100 gas -21 drone (check the time at which I made my last drone before queen is done) Perhaps the 13pool is a bad choice, and we might test this in our next installement, but this isn't the issue here (though I love this build). To get an idea of which BO "tactic" is best, I used the "build order" tab when a game ends, and wrote down the timings for the spawning pool, the queen, the zergling, zergling speed (which means 100 gas) and 21st drone (last unit built before queen pops). Many of these timings are redundant, but I took all of them to weed out human mistakes. Game1: The 9 Overlord build! -13 pool: 1:49 -16 queen: 2:54 -18 zergling: 3:07 -zergling speed: 3:34 -21st drone: 3:37 Game2: Regular 10 Overlord build! -13 pool: 1:48 -16 queen: 2:54 -18 zergling: 3:07 -zergling speed: 3:34 -21st drone: 3:37 Game3: The 10 Overlord with extractor trick! -13 pool: 1:49 -16 queen: 2:55 -18 zergling: 3:07 -zergling speed: 3:34 (amazing how accurate this is, considering the variability in building extractor, beginning to mine gas etc...) -21st drone: 3:39 (I slacked off here, pretty sure it should have been at around 3:37. In my defense it's been a long day^^ I checked the replay, and yeah the larvae poped a bit before I began to build). Results: Well the timings are all the same (no this isn't rigged, I'm really suprised by how close these number are). I don't think a 1s difference is a relevant result since we're talking about fractions of seconds here. I was spamming with hotkey and mouse click to try to make the results accurate, but there's still AI and human response to take into consideration  .Conclusions: => The three builds are identical! Do whichever you like. You should note that the extractor trick makes you lose 2x "drone returning minerals", and that in this test I got lucky: My drone started building the extractor and my 10th(11th) drone started morphing at exactly 75minerals. Also, when I canceled the extractor, the closest mineral patch was cleared to avoid losing time in mining. I think I executed this well, but it's easy to get an extractor trick a few seconds off when massing games (which means this isn't my prefered choice). => Based on what I've got here, just like in SC:BW, your larvae production doesn't seem to stop at 3 larvae. Of course the 4th larvae won't pop, but the respawn cooldown will begin. Perhaps there's a minimum respawn time, but in the 10overlord build, that minimum respawn time (if it exists) doesn't apply. Basically, when using a 10 overlord build, you're not wasting larvae spawn time, since none of my later units were delayed. Too long:didn't read: 9 overlord = 10 overlord = 10 overlord w/ extractor trick WITH THE 13POOL BUILD If going for an overpool, obviously you should 10OL. | ||
|
Klive5ive
United Kingdom6056 Posts
On April 19 2010 18:22 HeyheyLBJ wrote: All in all, 9 OL is slightly superior in every way. This is the 3rd person to come to the same conclusion then. I think we can put that argument to bed now. 9ol is the best build if you're going to drone to 16. However, in your test LBJ you didn't give the mineral count at the time you would want to 13 pool etc... | ||
|
Klive5ive
United Kingdom6056 Posts
On April 19 2010 20:18 Hammy wrote: Too long:didn't read: 9 overlord = 10 overlord = 10 overlord w/ extractor trick WITH THE 13POOL BUILD If going for an overpool, obviously you should 10OL. This is confirmation of what I was trying to explain. With a 13 pool instant mineral count is important since you can't rebuild that 13th drone until you've placed the spawning pool. Your analysis wasn't as precise as people above so we still can't be 100% sure what is best for 13pool (but I expect all 3 are very close as you suggested.) | ||
|
HeyheyLBJ
Sweden160 Posts
On April 19 2010 20:00 goszar wrote: HeyheyLBJ Thanks a lot. I have recently seen the following build in Zotac cup played by some Oger guy: 10 Drones Extractor trick Overlord 3 Drones when overlord pops (one is usually 1 second late) Spawning pool (it is at 14). The other player did usual 10Ol and was 50 minerals behind and about 2 seconds behind on 14 drones. I have watched several times - they were equal up to 10 drones. My calculations show that this is superior even to 9o by small amount assuming you are going for 14 pool. This is the build I tested with. The 14th drone will come much faster with the extractor trick build since it is one of the drones in the first batch after the OL pops. So I'm not sure what you mean with 2 seconds and 50 minerals behind. | ||
|
cyclone25
Romania3344 Posts
On April 08 2010 18:45 Tipton wrote: How about a double extractor trick (at around 100min) for a 10/12 ovie ? I saw Check doing it once. I'm curious about it too. | ||
| ||
