EPT[D] Chrono Boost, Economy, And You! - Page 2
| Forum Index > StarCraft 2 Strategy |
|
kbx4ever
United States4 Posts
| ||
|
michaelhasanalias
Korea (South)1231 Posts
On December 26 2010 05:30 wideye wrote: thanks. very interesting. how did you get all this information? I modeled most of it using build times and nexus energy regeneration, as well as parsing some data from old theorycrafting threads on mining minerals as sourced on the eponymous liquipedia page. There may well be some errors in this data, as with any spreadsheet models, so if you see anything questionable please don't hesitate to ask! On December 26 2010 09:51 FireBlast! wrote: Obvious platitudes about in-game variables aside this is good info. I have a bad habit of only trying to chrono boost my nexus in PvPs, this shows how damaging that line of thought could be Yeah, if there is a way to make it more applicable to specific builds, I'd be happy to try and work on it. I tried to eliminate as many variables as possible when doing this to create some pure data to use as a foundation to build upon. My goal here was just to crunch some numbers and see if I can't put to the test some of the uncontested conventional wisdom. Naturally, if you expand, the latter chronoboosts would end up netting a bit more income, but as I stated in the OP, I simply haven't had the time to incorporate 2-base functionality into any of the little tools I've been working on. And since we can't ever get to 2-base play without considering 1-base play, I figure this is a good starting point. On December 26 2010 09:55 ABCSFirebird wrote: I looked into the excel file and i am pretty sure that you got the income difference for each worker wrong. 0.7 for 17-20 is definately wrong - according to my measurements (on SoW) they should be about 0.4 and 21-24 0.1 (non on gas). Even if you consider that three probes are on one gas - the 20th shouldnt be the same as the 16th. But maybe i interpreted your table wrong. Under the operating assumptions at the top, I assume that the player takes gas optimally in whatever build he chooses. Note that this slightly impacts numbers, depending on when he takes the gas. The first 16 miners of mineral patches average ~42 minerals/minute, and the first 2 gas workers gather 42 gas/minute (which I lumped into resources). (so, 1-16 mining, 17,18 gas 1; 19,20 gas 2). The 2nd two (21-22) are each gathering gas at ~30gas/minute, and the last 8 workers mine at ~18minerals/minute. You're absolutely correct that the 24th worker is barely going to impact the mineral patch, but then it's very cumbersome to adapt a general model to specific maps to determine which workers will mine sub-optimally, which geysers require 4 workers, when exactly to take your gas, etc etc etc. This model assumes ideal situations, and places all patches an average distance apart. When beginning the modeling for the WEDAT tool I built last week, I began with a few threads that went pretty far in depth into mining theorycraft, and it just seemed easier to accept their numbers. Should you disagree, if you can create a better income rate per workers, I'd love to incorporate it. I'm less concerned about defending my own results than utilizing the most accurate and useful data. On December 26 2010 11:07 Ayrie wrote: wow, chrono'ing probes at the beginning of a game can exceed the income from a mule. amazing work. so now all we have to do is figure out how to chrono effectively for each opening build. i always figured that the first two chronos going to the 11th and 13 or 14th probes paid off immensely. so should you still spend the 2nd - 6th chronos on the warp gate research for a proper four gate rush, or spend the 2nd (and maybe 3rd) on the nexus for that extra 150 minerals? I think you may misunderstand the numbers I posted. For example, the 11 chrono boost (1st one) adds roughly 153.5 minerals to you over the time it takes to reach full saturation. One mule by itself early in the game should add 270 minerals over 90 seconds unless it's dropped on a non-near patch. In fact, Terran Income to saturation time (6:48) could theoretically mine ~1000 more minerals than a full-chrono'd Protoss on 1-base. But of course that doesn't factor in workers having to go off-patch for building, so it's not even a viable comparison. | ||
|
ProtossPenny
United States169 Posts
| ||
|
ABCSFirebird
Germany90 Posts
On December 26 2010 11:30 mlbrandow wrote: The first 16 miners of mineral patches average ~42 minerals/minute, and the first 2 gas workers gather 42 gas/minute (which I lumped into resources). (so, 1-16 mining, 17,18 gas 1; 19,20 gas 2). The 2nd two (21-22) are each gathering gas at ~30gas/minute, and the last 8 workers mine at ~18minerals/minute. Alright - now the numbers makes sense - the early assimilators with only two probes in them are a very unusual concept for me, pretty much never used in practice (atleast here in europe). As for the numbers of the last eight: these are my own results after examining the mining rate. They differ slightly from liquipedia and i only ran numeral measurements for the last four - since there is heavy noise - but as you can see it seems to make sense splitting the last eight in groups of atleast four. I don't say my numbers are absolutely correct - but an approximation which states that always about four mineral patches are close to the main and about four are further away would be a lot better in my opinion - you could even use the numbers from liquipedia (like 42 +-3 / minute depending on distance) for this. However as i entered my numbers in the table, the results aren't that different - so i probably should just thank you for your nice work at use the table as i like to  | ||
|
michaelhasanalias
Korea (South)1231 Posts
On December 26 2010 13:00 ABCSFirebird wrote: Alright - now the numbers makes sense - the early assimilators with only two probes in them are a very unusual concept for me, pretty much never used in practice (atleast here in europe). As for the numbers of the last eight: these are my own results after examining the mining rate. They differ slightly from liquipedia and i only ran numeral measurements for the last four - since there is heavy noise - but as you can see it seems to make sense splitting the last eight in groups of atleast four. I don't say my numbers are absolutely correct - but an approximation which states that always about four mineral patches are close to the main and about four are further away would be a lot better in my opinion - you could even use the numbers from liquipedia (like 42 +-3 / minute depending on distance) for this. However as i entered my numbers in the table, the results aren't that different - so i probably should just thank you for your nice work at use the table as i like to Yeah, I use those numbers in 3 or 4 different ways, and I openly admit they are generalizations. Naturally, there is no map that exists where the patches are all uniform distance. If you can come up with some better numbers, I'd like to use them. Even very slightly more accurate for most cases would still be an upgrade. My question for you though is then, would it behoove me to split mining workers 1-8 and 9-16 into groups of 4 as well, since we're assuming four patches are near and four are far? And how diagonals impact this rate specifically? I believe that my method averages out to the correct amounts at saturation (which is principally what I'm looking at), but if there is a way to make the numbers more accurate for any time X pre-saturation it would certainly make the data more useful. Thanks for your input. edit: Referring again to this thread, I updated some of the mining numbers, which better adjust for mineral patch distances. The results are very, very similar (as the income differences were very small to begin with), but it does slightly tilt the value of the earliest chrono boost even more. 1st chrono - 157.7 minerals 2nd chrono - 81.9 minerals 3rd - 63.7 minerals 4th - 52.8 minerals 5th - 34.2 minerals 6th - 18.3 minerals 4wg loss: 242.6 minerals - Delays in worker production are unaffected by mining rate adjustments per worker (assuming the same saturated sum). Therefore, this number is unaffected, although the loss per unit of time does change slightly. | ||
|
ABCSFirebird
Germany90 Posts
On December 26 2010 21:13 mlbrandow wrote: My question for you though is then, would it behoove me to split mining workers 1-8 and 9-16 into groups of 4 as well, since we're assuming four patches are near and four are far? Only if a player is forcing the first eight probes to mine from the four nearest patches. At high level you see people doing this sometimes, but i don't think it is really worth it since forcing one probe to a mineral patch where already another probe is mining can be some lost mining time if the timing is bad. Well maybe to illustrate, lets assume a player forces probes and he is lucky with the timing, then there are 8 probes on 4 close mineral patches and each probe is mining with 45/min. The next 8 probes go to the far patches and there they are mining with 39/min. Obviously that results in an average mining rate of 42/min. If you don't force probes, which most people don't, than you end up with pretty much the same (2 probes on each patch at a total of 16 probes). The first 8 evantually spread equally on 8 patches, so some are mining with 45/min, some are with 39/min, for the next 8 it is the same. The thing is - nobody can use cb for getting some of the first 8 probes faster and if cb is used for 9-16 there is only an advantage in comparison to cb at >17 probes - so there is nothing to discuss, and therefore i wouldn't split 1-8 or 9-16 in groups of four. But now there are 4 mineral patches which are saturated with 90/102 - and four with 78/102. So probes 17-20 are likely to go on the 78/102 patches (since the probability that these are already in use isn't as high as for the 90/102 patches). So you gain 24/min for each probe. And for the last four only 12/min. Obviously these numbers are heavily overlayed by noise caused from probes running around and not staying on a mineral patch - but they are a better approximation, which is somwhat important for the cb examination, because earlier cb is more effective than the later one - and catching up by using cb after "wasting" some for warpgatetech is pretty much impossible. And how diagonals impact this rate specifically? What do you mean with diagonals? | ||
|
michaelhasanalias
Korea (South)1231 Posts
I was looking at a post on MULE efficiency from a while back (can't seem to find it now), and it showed RGY-colored patches for every mining position of every 1v1 map. Green meaning the MULE would return 270 minerals (near), Yellow meaning they would return 240 minerals (far), and Red meaning they would mine 270 minerals but only return 240 (middle). These middle usually were at diagonal positions relative to the CC. I'll look again to see if I can dig it up. It was an awesome post from several months ago. | ||
|
Kammalleri
Canada613 Posts
| ||
|
Supah
708 Posts
Maybe it's just me, but compared to Mules and Inject, CB seems a little underwhelming? | ||
|
RLam
United States6 Posts
| ||
|
Hurkyl
304 Posts
On December 27 2010 03:19 RLam wrote: great post, but are you saying we should focus more on cbing our nexus instead of getting warpgate tech? if so...wouldnt that hurt us unit count wise?....sorry for the nooby question He wasn't saying -- he was giving facts so that people can make more informed decisions. There is always a trade-off between investing in economy and unit count -- so yes, there is window where CB'ing warpgate tech will have more units than CB'ing the nexus. Eventually that window will close, and the build that CB's the nexus will have a permanent unit count advantage. These facts tell you how much you are spending to make that window appear -- and how much you have thrown away if you don't do anything in that window. (e.g. rush, or defend a rush) | ||
|
hatstarcraft
United States4 Posts
| ||
|
Dromar
United States2145 Posts
When I play protoss, I crono boost probes and upgrades primarily, although I have at times crono boosted immortal or observer from a robo before I moved out. | ||
|
Deleted User 3420
24492 Posts
(im not saying it's reasonable to be asking for those figures.. clearly it's borderline impossible. just that the numbers could be a little misleading in the context of a real game) | ||
|
Hurkyl
304 Posts
On December 27 2010 04:50 Dromar wrote: Personally I think crono boosting units is a bad idea unless you're going for a timing attack or desperately need to defend. Crono boost basically gives you 10 seconds of time. If you crono a zealot or something, just to have it sit at your ramp, it's wasteful IMO. Right, it's not something to do on a whim (unless you were going to waste the energy anyways). The most obvious use is to boost out units for a rush (or if you desperately need units to defend). A more subtle use is to shuffle money around. It lets you build one Gateway, but still produce units as if you had one and a half, letting you free up the money you would have spent on a second Gateway for other purposes, such as an earlier Cybernetics Core, or early Nexus... or even on just continuous Probe production. Should you think you need the units I speculate it's better to Chrono Boost units than it is to pause Probes to get a second early Gateway and then resume Probes with Chrono Boosts... but I haven't tried testing it. | ||
|
Deleted User 3420
24492 Posts
On December 27 2010 06:21 Hurkyl wrote: Right, it's not something to do on a whim (unless you were going to waste the energy anyways). The most obvious use is to boost out units for a rush (or if you desperately need units to defend). A more subtle use is to shuffle money around. It lets you build one Gateway, but still produce units as if you had one and a half, letting you free up the money you would have spent on a second Gateway for other purposes, such as an earlier Cybernetics Core, or early Nexus... or even on just continuous Probe production. Should you think you need the units I speculate it's better to Chrono Boost units than it is to pause Probes to get a second early Gateway and then resume Probes with Chrono Boosts... but I haven't tried testing it. chronoboosting probes also requires u to spend more money faster, delaying your ability to build production facilities or units in the short term. | ||
|
AirbladeOrange
United States2571 Posts
| ||
|
michaelhasanalias
Korea (South)1231 Posts
On December 27 2010 04:55 travis wrote: This doesn't take into consideration the cost of creating production facilities to catch up on production you lose out on by not chronoboosting units or warpgate. Just a small nitpick. (im not saying it's reasonable to be asking for those figures.. clearly it's borderline impossible. just that the numbers could be a little misleading in the context of a real game) I was thinking about this earlier today before I left, and I actually don't think it would be that difficult to map out. While you technically lose out on the probe mining bonus of a chrono boost if you use that energy to boost unit production from a warpgate, you would also gain a mineral "savings bonus" if you used it on producing a unit, since you could delay (or avoid) construction of an additional facility. Thanks for the observation though. You're absolutely correct in that the probe mining bonus taken in a vacuum can be misleading (versus the production gain of a unit-producing structure). I'm going to try and factor it into the OP when I can run some numbers. | ||
|
Deleted User 3420
24492 Posts
| ||
|
Gecko
United States519 Posts
| ||
| ||
