|
On November 30 2010 01:59 Chill wrote: I've very happy this thread exists. I've felt helpless not knowing the ideal Zerg FE build. Chill, sarcasm is not allowed on the internets.
|
Lomilar claimed this was a variation of the 7RR without the roaches... Unfortunately the longer mining time at ten drones has put it fairly low in the resources mined category relative to all the other builds. Also the overlords are clearly timed for pumping roaches instead of drones at half the supply.
I tried some minor adjustments to the build but haven't come up with anything substantial. Hopefully someone will produce something soon, because I really like the idea of an economical and still extremely flexible build. Once someone can raise the resources mined count higher, I think the added drone and perhaps overlord would make up the remaining difference.
First, I'm not sure why that build was copied twice. My bad.
Secondly, if that's true why is it runner-up? I was really hoping for that to be as good as the OP was making it sound, but if what you say is true, that is a damn shame.
|
On November 30 2010 01:59 Chill wrote: I've very happy this thread exists. I've felt helpless not knowing the ideal Zerg FE build.
This thread is providing data to compare many builds to eachother in a meaningful way, and not only FE builds. I think this thread has put to rest many misconceptions and confusions being propagated in other threads where the only criteria being used was "first to 50 drones" or some such nonsense. Also, recognizing the potential power of an overpool build is a plus I think.
|
This is great :D Sometimes I want the most economic opening I possibly can. Some of my friends aren't all that great, and if I see them fast expand, I know I will be safe for at least 6 minutes. I might actually try to do one of these builds some day :b just as listed here. Just to see what happens in an actual game.
|
On November 30 2010 02:10 Xanbatou wrote:Show nested quote + Lomilar claimed this was a variation of the 7RR without the roaches... Unfortunately the longer mining time at ten drones has put it fairly low in the resources mined category relative to all the other builds. Also the overlords are clearly timed for pumping roaches instead of drones at half the supply.
I tried some minor adjustments to the build but haven't come up with anything substantial. Hopefully someone will produce something soon, because I really like the idea of an economical and still extremely flexible build. Once someone can raise the resources mined count higher, I think the added drone and perhaps overlord would make up the remaining difference.
First, I'm not sure why that build was copied twice. My bad. Secondly, if that's true why is it runner-up? I was really hoping for that to be as good as the OP was making it sound, but if what you say is true, that is a damn shame.
I placed this runner-up because it is the current leading build in both drone count and overlords, and I feel that some adjustments could be made to make it's resource count comparable to most other builds submitted, in the range of 4550 or so.
|
I'm trying to determine why 'my' build (sarcasm) mineral count is not nearly as high, and here are my current thoughts:
The 10 overpool seems fine. I kinda wish I could spend one more larva, but at that point we are mineral-capped, not larva capped.
The queen goes down right after the pool pops at 15. That seems fine as well. We have 50 minerals by the time the next larva comes up.
The hatchery timing I wish could be a little earlier, but we risk having 3 larva sitting around. Maybe change OL at 17 and hatch at 18 to an hatchover at 17. This will get the hatch up faster, and we still have one supply to wait for the OL. We sit on 3 larva for a bit here. Overhatch seems better, with maybe a drone inbetween.
At that point, spending all the larva seems to be the priority, but since our hatch went down earlier we need our second queen. I think a 20 timing should remain okay. If we do the earlier hatch, the first queen doesn't quite have enough energy (assuming we are perfect) to spit on the second hatch when it pops up. The 20 timing remains accurate.
At that point, I am not sure of how to time the overlords and drone and spawn larva pops... so... going to try this. Similar, but slightly worse results. I maynarded more drones over, but ended with 290 minerals/42 drones.
|
I have taken the liberty of adding a build order from another thread which is being claimed as more economical than any hatch-first build, adjusted of course by removing the extractor and creep. The test results have been posted, and here is a direct link to the OP.
http://www.teamliquid.net/forum/viewmessage.php?topic_id=172107¤tpage=8#142
On November 30 2010 01:51 Phrencys wrote:http://tinyurl.com/35o22a5 9 Overlord 14 Spawning Pool 16 Hatchery (send @200 mineral) > +5 (10 sec) 15 Queen > Spawn Larvae[auto] 17 Overlord 18 Extractor > +3 (2 sec) 20 Overlord 20 Queen > Creep Tumor 32 Overlord 40 Overlord 48 Overlord 56 Overlord 7:00 Checkpoint
Edit: This said I have to agree with the OP's conclusion: I've yet to find a hatch first BO that can match the economy of the above BO. This is all theorycraft assuming no block though, an assertion that can very often go wrong in the real world.
|
I find it strange that you are invoking the scientific method but haven't bothered to consider the repeatability of your results. There are, inevitably, a number of small timings that could have a slight impact on your results, but since the difference between a number of your "contenders" is a only couple percentage points, it may be worth it to re-run your experiments a few times. One data point is hardly conclusive.
I'm also confused about Overlord timings in the builds you have listed. Some are very specific about them, and others less so. Variation in Overlord timing could certainly create small variations in results. It could certainly be the case that one of your sub-optimal builds, with variation in undefined Overlord timing, could pull ahead.
|
9 Overlord 12 Pool 15 Overlord 18 Queen 21 Hatch 22 Overlord 30 Queen
|
On November 30 2010 07:28 Dominator1370 wrote: I find it strange that you are invoking the scientific method but haven't bothered to consider the repeatability of your results. There are, inevitably, a number of small timings that could have a slight impact on your results, but since the difference between a number of your "contenders" is a only couple percentage points, it may be worth it to re-run your experiments a few times. One data point is hardly conclusive.
I'm also confused about Overlord timings in the builds you have listed. Some are very specific about them, and others less so. Variation in Overlord timing could certainly create small variations in results. It could certainly be the case that one of your sub-optimal builds, with variation in undefined Overlord timing, could pull ahead.
Your point about the overlord timing is absolutely correct. I have been asking for submissions that are very specific, but most have not been, and therefore I have to make assumptions that might be inefficient in the process of testing. If you think one of the builds previously tested could be refined, by all means submit your own. I can't really perfect every variation of every build order myself, which is why this needs to be a collaborative endeavor.
Regarding your other point: All of the builds have been tested at least twice with the leading result submitted. I myself have been testing the top two builds in order to limit error and also test variations.
|
I am pleased to announce after numerous tests and variations I have come up with an improved version of the Lomilar build. The exact order, with the results and a replay have been posted to the OP.
However, now I face a slight dilemma that I am seeking consensus on. Douillos' build is ahead in regards to total minerals mined, but Lomilar's build is ahead in regards to drone count. In order to decide the leading build, we must come up with a method of comparing the relative worth of drones to minerals.
I have been thinking of valuing a drone as it's original cost + the number of minerals it could mine within the 17 seconds it requires to produce a new one. However, this may be a flawed method. Please give me input regarding this issue so we can reach a conclusion.
|
I think in terms of economic analysis we've got a pretty good set of build statistics. Both of the winning builds provide a key situational advantage. The 10 overpool build allows for a lot of early game flexibility. I've been having a lot of success with it just because I'm not locked into either an economy or aggression build. The flexibility alone make this build worth the ~300 mineral you lose vs. the 16 hatch. However, the extra minerals mined in the 16 hatch build is just about enough to drop a third hatch. If you're opponent early expands, you could use that money to take a third, but my impulse would be to build an in-base third and use the money and fast base saturation to prepare for a particularly nasty timing attack. If you can get away with a 16 hatch and fight off early pressure, 3 hatcheries off two bases can make for some really scary aggression in the early-midgame.
|
Just playing around with the sc2calc build order calculator I came up with 14 Hatch 15 Pool being the best (beating out both later hatch firsts and all the pool firsts I tried. The calculator has a few flaws, like splitting drones evenly between hatcheries but not accounting for the travel time from where they're produced to when they go to the second hatchery. I think it still gives a strong outline though.
It turns out overlord timing can be pretty important, and can have a sizable impact on the end result. That's something that will have to be fine tuned with these builds. I played around with it and I feel like I have it set pretty good.
Another thing I liked about this build is that queens are synced up almost perfectly (3 seconds apart)
You should try testing this one in game and see how it works out.
9 Overlord 14 Hatchery (send @200 mineral) > +2 (10 sec) 15 Spawning Pool 17 Overlord 19 Queen > Spawn Larvae[auto] 21 Queen > Spawn Larvae[auto] 26 Overlord 30 Overlord 42 Overlord 52 Overlord 60 Overlord 6:00 Checkpoint 7:00 Checkpoint http://tinyurl.com/37jhkhw
|
On November 30 2010 09:37 Fenam wrote:Just playing around with the sc2calc build order calculator I came up with 14 Hatch 15 Pool being the best (beating out both later hatch firsts and all the pool firsts I tried. The calculator has a few flaws, like splitting drones evenly between hatcheries but not accounting for the travel time from where they're produced to when they go to the second hatchery. I think it still gives a strong outline though. It turns out overlord timing can be pretty important, and can have a sizable impact on the end result. That's something that will have to be fine tuned with these builds. I played around with it and I feel like I have it set pretty good. Another thing I liked about this build is that queens are synced up almost perfectly (3 seconds apart) You should try testing this one in game and see how it works out. Show nested quote +9 Overlord 14 Hatchery (send @200 mineral) > +2 (10 sec) 15 Spawning Pool 17 Overlord 19 Queen > Spawn Larvae[auto] 21 Queen > Spawn Larvae[auto] 26 Overlord 30 Overlord 42 Overlord 52 Overlord 60 Overlord 6:00 Checkpoint 7:00 Checkpoint http://tinyurl.com/37jhkhw
Ok thanks... I've been meaning to test one of these builds recommended by these optimizers/calculators/testers...
One other question I have though: Do these programs value each drone equally in regard to mining, or do they adjust for the diminishing marginal value relative to the increasing marginal saturation of a base? This seems like a very complicated calculation and I suspect none of them take this into account. This difference though has been the difference between resources mined on many of the builds I have tested that are pool-first.
|
I think they have slightly different mineral calculations, but they do account for diminishing returns when you put 3 drones on a mineral patch. I'm not sure how the other apps work, but this one values each drone at .7 minerals per second and the third as .3 minerals per second on each patch. It spreads the drones out evenly across each base to maximize the income.
|
On November 30 2010 09:44 jdseemoreglass wrote:Show nested quote +On November 30 2010 09:37 Fenam wrote:Just playing around with the sc2calc build order calculator I came up with 14 Hatch 15 Pool being the best (beating out both later hatch firsts and all the pool firsts I tried. The calculator has a few flaws, like splitting drones evenly between hatcheries but not accounting for the travel time from where they're produced to when they go to the second hatchery. I think it still gives a strong outline though. It turns out overlord timing can be pretty important, and can have a sizable impact on the end result. That's something that will have to be fine tuned with these builds. I played around with it and I feel like I have it set pretty good. Another thing I liked about this build is that queens are synced up almost perfectly (3 seconds apart) You should try testing this one in game and see how it works out. 9 Overlord 14 Hatchery (send @200 mineral) > +2 (10 sec) 15 Spawning Pool 17 Overlord 19 Queen > Spawn Larvae[auto] 21 Queen > Spawn Larvae[auto] 26 Overlord 30 Overlord 42 Overlord 52 Overlord 60 Overlord 6:00 Checkpoint 7:00 Checkpoint http://tinyurl.com/37jhkhw Ok thanks... I've been meaning to test one of these builds recommended by these optimizers/calculators/testers... One other question I have though: Do these programs value each drone equally in regard to mining, or do they adjust for the diminishing marginal value relative to the increasing marginal saturation of a base? This seems like a very complicated calculation and I suspect none of them take this into account. This difference though has been the difference between resources mined on many of the builds I have tested that are pool-first.
Yes they do account for it, at least EvolutionChamber does. 2 drones per patch is optimal (nearly), after that, there are diminishing returns, and you're better off having your drones at another base.
|
two ov's in a row for Lomilar's build?
|
My attempts at a 10 pool 18 hatch in a ladder game.
My feelings during the game is that, its good.. I love it, I had so much money.. I just felt like I could continuosly produce units.. I like it.. it feels smooth.
|
On November 30 2010 10:08 BnK wrote: two ov's in a row for Lomilar's build?
Yes... If you notice, the first overlord is built at 17/18 supply. There is a window of opportunity while supply capped to build a second overlord with exactly 1 larva and 100 minerals. The supply from this additional overlord gets used up quickly, since the simultaneous creation of 4 drones + a queen uses 6 supply instantly. Without the addition of this second overlord you would find yourself quickly supply capped again.
|
I think it should be noted that the pool first builds generally provide more larvae, while hatch first builds provide more minerals mined. Knowing this, it seems logical that if you are doing ling based play (high larvae) then a pool first build would be optimal as larvae is the limiting factor in this strategy, while a roach based strategy may benefit more from fewer larvae and more minerals mined, thus making hatch first play more optimal.
|
|
|
|