|
714 Posts
On March 19 2011 19:42 Shadowcaster wrote:
Griffith it seems that you have confused your self into believing a poorly constructed set of constructs is insightful.
But for other people here is what he is SAYING.
First he ASSUMES that any production structure that you create is constantly producing a UNIT.
And constant production of SAID unit over a MINUTE is the True cost of that unit.
The Claim that upgrades are cheap stems for the fact that the build time is LONG thus the per upgrade cost is averaged down over a period of time. Thus is the same for the thor and BC comment. No concern is given however to the amount of time it would take to save for this ? as if you were truly calculate in this fashion you would need to know that on 2 BASE your maximum GPM is about 440. And a BC takes 300 gas or 40 seconds of mining time. Similarly for these upgrades.
Perhaps this will someday be a useful line of thinking but upto this point, it is not quite there.
lol, yes there is "build up time", that should be obvious. but they are nothing more than spikes if you were to plot your minerals and gas over time. There is no long-term "build-up" if you follow the table, in which you may need to an additional 2 raxes to serve as a "buffer zone". Because immediately after you research upgrades, that costs you an additional 150/150, you will actually find yourself needing to consume minerals "faster").
As for the BC, yes occasionally you may float on 400 minerals and 300 gas to "build" up for a battlecruiser. But who the hell goes 2-base Battlecruisers?
|
Howdy all... Long time lurker, first time posting. I have some background and a great deal of interest in game theory concepts.
As several other people have written, the OP could have used more precise language, but addresses some important (and often neglected) concepts. Kcdc and company are correct to assess the real significance of this kind of data as relating to the production efficiency of structures. Clearly, this kind of table is no replacement for the online tools mentioned by several posters, but having some of the rough ratios here at your fingertips is quite useful.
And the data there is quite interesting. It is remarkable how dramatically a structure's efficiency can vary: constant marauder production burns minerals at roughly 3 times the rate of constant reaper production, which complicates any theoretical tech-switch from one to the other, despite the same production facility being used for both, because of the ugly shift in production capacity that would be necessary. (Note the use of “theoretical”... the actual usefulness of units is not the issue at the moment.)
However, for a more useful set of figures, I would recommend actually presenting them in terms of percentage of resource capacity per base, rather than in absolute figures. These figures are most useful as a rough guide for improvisation and concept work, and the step of dividing them by 800 or 440 seems an unnecessary hassle for this application.
Also, I respectfully disagree with the OP's assertion that this is “more useful for T” than Zerg or Protoss. I actually think this kind of table is most useful for Zerg, given the nature of the larva mechanic. A hatchery generates about 9 larva per minute, and each Zerg unit (counting ling pairs as a unit) costs ONE larva. As a result, a Zerg hatchery, macroed properly, will produce 9 units per minute. Build time is irrelevant to Zerg production capacity. This means that a table of production used per minute while building various units is actually a sensible tool for Zergs to have... even if most of us prefer to simply go “by feel.”
Note that it matters quite a bit whether these units are replacing fallen units or are part of the initial build-up; that is, it matters whether the Zerg must also produce overlords because overlords cost both minerals and larva. Replacement means the “9 units” rule is solid; build-up means that 1 supply units are built at 8/minute and 2 supply units at 7.2/minute.
For replacement, the math is extremely simple. Constant drone/zergling production from a saturated base requires two fully injected hatcheries, because they cost so much larvae/minerals. Roaches are fairly close to 1:1, or 1 hatch producing to 2 geysers working. If you know those benchmarks, the rest is obvious, because the ratios are simple inverses of the costs: A hydralisk is 2x the gas cost of a roach, and so a single hatchery can produce hydras to eat gas from 4 geysers. A muta is 2x a hydra, so one hatch can produce mutas to eat up 8 geysers, and so on.
For build-up, the numbers are trickier because of overlords, losing units to combat, structures, etc. However, without going into the numbers at this moment (maybe later...) the trend is for production to slow, and therefore for more capacity to be required to fully spend minerals. Note that the exception is the low cost units, such as the drone and zergling, where the overlord only increases the minerals burned per minute, and no gas is involved.
I am considering doing a more complete write-up with full tables for all races in the future, though I know they are bound to be theoretical rather than iron rules.
|
I'd like to slightly revise my Terran two base vs Protoss composition. I find that, being Vikings are more gas intensive than Banshee per time, you need to save gas somewhere. Hence Tank/Viking (common in TvT) could be viable against toss and it would be equivalent in terms of gas to Banshee/Thor. Take the following compositions (note that I am not accounting for supply cost):
1st Composition
Banshees (one port): 150/100 per minute
Thors (Two Fact): 600/400 per minute
Reactored Fact: 400 per minute
Two reactored Rax: 480 per minute
Ratio: 1630/500 (pretty decent)
2nds composition
Tanks (two factory) 400/334 per minute
Vikings (one port no reactor) 222/148 per minute
( remainder from above): 880 per minute
Ratio: 1502/482 (Excellent ratio)
Note that the compositions present hard counters to virtually anything in the protoss mid game, and gives you the flexibility to transition back and forth between Thors/Banshees and Tank/Viking with the click of a button.
I feel composition 1 works with any phoenix/Immortal builds and 2 works with heavy colossi/void builds, so vary your comp. as such. Upgrades might be a little light but two armories and one bay for early upgrades should give you enough balance for the early game. Never deviate from the ratios from above on two base.
|
Hi, I am surprised about the cost of the upgrades in your table and why they are so very cheap.
i.e.
L1 Infantry 38 38
I am looking into the viability of a very early Eng Bay +"a critical upgrade" (matchup/BO responsive). Per Guide to critical upgrades
I am confused that another assessment of upgrades highlighted that they don't pay-off until like 20+ marines Forum link. However, based on real cost, does the viability of a very early Eng Bay pay off?
Ebay 214 0
L1 Infantry 38 38
Total 252, 38 for first.
if 1x Marine = 150 0 real cost, then that makes early Ebay (i.e. build @80g for immediate L1 upgrade) very attractive. Maybe I'm looking at this wrong.
|
On March 19 2011 00:58 Griffith` wrote:
Why is this useful? Because income per base is normalized to per minute, ie. 2 base Terran averages an income of of ~1800 minerals and 456 gas per minute. Suppose there was a unit that costs 1500 minerals and 1500 gas but took an hour to build, is it "expensive"? nah. Same analogy goes for upgrades, they're actually really cheap, but just have a high "up-front" cost. Honest question here: what's the benefit of this table over something like the Terran Unit Production calculator?
In particular I think a table like this makes it hard to think about opportunity cost incurred by banking resources for expensive things like BCs or +3 upgrades, as well while you wait for those sort of things to build. Upgrades cost very little in terms of income, but they take ages to pay off. Plus, you need to have a certain number of units that benefit from the upgrade on the field for it to be a better investment than just getting more units. For example in terms of DPS, in a pure marine army, Attack+1 isn't a good investment until you have at least 28 marines, if not more. You'd be better off just making more marines.
|
I think the data should be normalized to cost per food ratio, at least adding the food consumed per minute consumed. Even so, adding a table normalized with cost per minute per food, meaning we would get fractions of minutes relative to the "food consumed" in that time would actually show what you mean to say.
So a Marine is 2.4/min * 1 food * 50:0 minerals per minute, not accounting for supply depots. Then we just add in the cost of the Supply Depot. We can start comparing this to Marauders at 1.2/min * 2 food * 100:25 normalized for supply and then we can more accurately consider a unit's value.
|
An interesting way of looking at the cost of things without a doubt, although you may be overthinking things a tad.
|
My favorite part of this list?
With one Barracks, one Factory, one Starport, and one Reactor (on the Starport), you can spend almost the entire production of a saturated base.
|
Edit: Actually, I think I'll just work out the rest of the numbers and save it for a post. I'm finding a lot of really interesting things and don't want to keep editing this every 5 minutes.
|
On March 22 2011 23:46 Blamajama wrote:
I'd like to slightly revise my Terran two base vs Protoss composition. I find that, being Vikings are more gas intensive than Banshee per time, you need to save gas somewhere. Hence Tank/Viking (common in TvT) could be viable against toss and it would be equivalent in terms of gas to Banshee/Thor. Take the following compositions (note that I am not accounting for supply cost):
1st Composition
Banshees (one port): 150/100 per minute
Thors (Two Fact): 600/400 per minute
Reactored Fact: 400 per minute
Two reactored Rax: 480 per minute
Ratio: 1630/500 (pretty decent)
2nds composition
Tanks (two factory) 400/334 per minute
Vikings (one port no reactor) 222/148 per minute
( remainder from above): 880 per minute
Ratio: 1502/482 (Excellent ratio)
Note that the compositions present hard counters to virtually anything in the protoss mid game, and gives you the flexibility to transition back and forth between Thors/Banshees and Tank/Viking with the click of a button.
I feel composition 1 works with any phoenix/Immortal builds and 2 works with heavy colossi/void builds, so vary your comp. as such. Upgrades might be a little light but two armories and one bay for early upgrades should give you enough balance for the early game. Never deviate from the ratios from above on two base.
I think that your numbers are going to be too high for reality of two base play, at first u arent fully saturated, and you need to take into account all the depots you need to maintain that production and hopefully slip an upgrade or two in there.
May I suggest,
4xdepot
Thorx1
Banshee x2
marine x5 (avg12 marines/min)
total 1600/400
Transitions nicely out of a techlab/reactor opening - get a raven as the first unit out of the starport and do a big push between 15-20mins when appropriate, micro marines to stay behind thors if there are colossus
|
Well to add some criticism
-> you explained why the costs are normalized and that is right, but I don't see any of your calculations i just see the results so it leaves me a little skeptical about your findings
-> your normalization is impractical because you base it around constant production in one minute of that unit (e.g. siege tank constant production, scv constant production ( 4x50=212?)
-The cost relative to you IPM (income per minute) of a terran unit actually decreases the more orbital commands you have on the field in the late game as you can call down mules, exchange locations on command centres, and save food and money on keeping worker counts minimal
User was warned for this post
|
This is cool but pretty useless.
|
Why did you necro this thread from over a year ago?
|
im just facepalming at how people are not seeing that this is simply a thread discussing and giving examples of mining income rate vs production spending rate.
|
Hello everyone i think the Ghost needs an update:
Min/Gas/Time/Supply
200/100/40s /2
Min/Gas/supply
300/150/3 per minute without supplydepotcosts
338/150/3 per minute with supplydepotcosts
hopefully i didnt mess up the math
Greeting everyone
|
If it works for you its great but I think this is really bad, as an economist. Okay I won't be that hard to say its
really bad, I'll just say its very one-dimensional.
If you want to deal with the actual prize, just take the units for what they actually cost.
If you want to predict value as you are trying to do, you need to take into account more variables like for
example opportunity costs and budget constraints.
If you have 3 factories and those are occupied and committed to constant Thor production, what is
the opportunity cost of not producing Siege Tanks as you can't produce any Siege tanks if your
factories only make Thors? What is the opportunity cost of a Zerg committing all his larvae to unit X
instead of unit Y?
Or, even better, you're assuming infinity money. In reality, your budget is constrained
in early/mid game when you're only on 1-2 bases. A siege tank feels more expensive to you, when you
produce it at the 4 minute mark, but if you produce it at the 15 minute mark, a Siege Tank does not feel expensive
since you have a much larger bank then. Thats why Zergs have like 30+ Infestors (which are damn expensive) lategame
because the budget constraint loosens up as the game progresses and you get more bases, more income thus
units feel "cheaper" to you because of increased income + production capacity.
The budget constraint would be calculated in some way of the amount of income you get in from all your different expansions. What economists do is to put max consumption subjected to restricted income (utility maximization problem in microeconomics). What you do is to maximize consumption subjected to nothing. And even if you want to be so fancy to put up a utility-max problem of this, its still not enough because of another factor not considered which I will explain below:
A third factor you have not factored in is probably the most important one: units have values that are not reflected in their price.
Say a Sentry for example, it ridicoleusly expensive. But NOT having a few Sentries early game versus certain build orders particulary baneling busts from Zerg or 4-gates from Protoss, gives Sentry infinity value. They are so valuable that you loose the game if you don't have it. Conclusion: The price of a Sentry does not reflect its value, as it has infinity value if your opponent does a certain build order. Other examples: If you are a Zerg and have a DT in your base, Overseers have infinity value. If you don't have a DT in your base, you have map vision and play against an opponent who never does invis units, Overseer has value = 0. But the price of the Overseer is still the same, how come?
If you play P vs a Zerg who does Roach Hydra, then Colossi have really high value, worth more than they cost. If you play P vs a Zerg who goes Mutalisk, Colossi have value = 0. Yet Colossi always cost 300/200 nomatter what, so the price does not factor in the value of the unit in combat!
Short summary: What youre trying to do (imo) is to narrow down SCII build orders into a very one-dimensional framework, and what I'm trying to show is different examples of many variables that you have to consider, which your calculation ignores.
|
|
|
|
|
|
EPT
