|
What many people are saying here is simply and plainly, utterly wrong. The logically and mathematically correct approach, I would say it is almost self-evident, is to prioritize resources in the following manner:
- Add workers.
- Add worker-producers/drop-off points.
And there we have the quickest and fastest way to get 200. Never build anything but probes. In this model, you will be adding a nexus approximately per every fourth worker, seeing as four workers is exactly what it takes to support continous worker/zergling prodction, not counting supply. Curiously, building nexuses in this manners means you will never need any pylons anyway. The fact that you aren't actually racing to 200 psi, but playing vs a human player, complicates the model. Note, however, that extremely long term and not considering the psi-limit, the above mentioned is actually the one which will give you the best economy and thus the most units.
- Pause worker-producer production to add enough units to defend all your workers by following these steps, in the given priority.
- Add a unit from unit-producer.
- Add a unit-producer.
Complications to this model is potential drops in economy - a mineral line emptied, probes destroyed. Also, seeing as basic units aren't always the answer:
- Pause unit-production to add enough tech to spend your resources in the most efficient manner.
The pausing gateway-production to add gateways example is rather similar to 8drone->lord vs 9drone->lord. We all know which the koreans use. I assume they have put them to the test, and don't want to advantage of a slightly earlier overlord to scout.
Bottom line is, the question is at exactly which point you want to have the maximum amount of units. Three years from now? Probes only, until approximately a ten minutes remain. Note that every workers produced pays for itself in...well, less that a minute. The best way to muster a force in two minutes, for example, is some 7drone->pool build. The best way to muster a force in ten minutes involves some ten hatcheries before spawning pool.
The second bottom line is: never halt production in a unit-producer to add another of the same unit-producer. However, not being allowed to operate in decimals can produce results such as these:
- Zealot: 40 time, 100 minerals.
- Gateway: 80 time, 150 minerals.
- Income: 100 minerals per 20 TUs.
- 1 gateway in place
00: Idle.
20: Produce zealot in gateway.
50: 150 minerals available. Do NOT produce gateway, because:
60: 200 minerals available. Produce zealot. 1 Zealot done.
70: 150 minerals available. Produce gateway.
100: 150 minerals available. Produce zealot. 2 Zealot done.
140: 250 minerals available. Produce zealot. 3 Zealot done.
150: 200 minerals avaialable. Produce zealot in new gateway.
(310: 8 more zealots produced. Resources stabilized at 100 minerals. 11 Zealots total, +1 1/4 done).
00: Idle
30: 150 minerals available. Produce gateway.
50: 100 minerals available. Zealot.
90: 200 minerals available. Another zealot. 1Z done.
110: 200 minerals avaialable. Another zealot. 1Z done total.
130: 200.... New zealot. 2Z done.
(310: 9 more zealots produced. Total 11 + 2/4)
|
You lost me with that one ^_^
|
On May 03 2005 16:25 Babibo wrote:
What many people are saying here is simply and plainly, utterly wrong. The logically and mathematically correct approach, I would say it is almost self-evident, is to prioritize resources in the following manner:
- Add workers.
- Add worker-producers/drop-off points.
And there we have the quickest and fastest way to get 200. Never build anything but probes. In this model, you will be adding a nexus approximately per every fourth worker, seeing as four workers is exactly what it takes to support continous worker/zergling prodction, not counting supply. Curiously, building nexuses in this manners means you will never need any pylons anyway. The fact that you aren't actually racing to 200 psi, but playing vs a human player, complicates the model. Note, however, that extremely long term and not considering the psi-limit, the above mentioned is actually the one which will give you the best economy and thus the most units.
- Pause worker-producer production to add enough units to defend all your workers by following these steps, in the given priority.
- Add a unit from unit-producer.
- Add a unit-producer.
Complications to this model is potential drops in economy - a mineral line emptied, probes destroyed. Also, seeing as basic units aren't always the answer:
- Pause unit-production to add enough tech to spend your resources in the most efficient manner.
The pausing gateway-production to add gateways example is rather similar to 8drone->lord vs 9drone->lord. We all know which the koreans use. I assume they have put them to the test, and don't want to advantage of a slightly earlier overlord to scout.
Bottom line is, the question is at exactly which point you want to have the maximum amount of units. Three years from now? Probes only, until approximately a ten minutes remain. Note that every workers produced pays for itself in...well, less that a minute. The best way to muster a force in two minutes, for example, is some 7drone->pool build. The best way to muster a force in ten minutes involves some ten hatcheries before spawning pool.
The second bottom line is: never halt production in a unit-producer to add another of the same unit-producer. However, not being allowed to operate in decimals can produce results such as these:
- Zealot: 40 time, 100 minerals.
- Gateway: 80 time, 150 minerals.
- Income: 100 minerals per 20 TUs.
- 1 gateway in place
00: Idle.
20: Produce zealot in gateway.
50: 150 minerals available. Do NOT produce gateway, because:
60: 200 minerals available. Produce zealot. 1 Zealot done.
70: 150 minerals available. Produce gateway.
100: 150 minerals available. Produce zealot. 2 Zealot done.
140: 250 minerals available. Produce zealot. 3 Zealot done.
150: 200 minerals avaialable. Produce zealot in new gateway.
(310: 8 more zealots produced. Resources stabilized at 100 minerals. 11 Zealots total, +1 1/4 done).
00: Idle
30: 150 minerals available. Produce gateway.
50: 100 minerals available. Zealot.
90: 200 minerals available. Another zealot. 1Z done.
110: 200 minerals avaialable. Another zealot. 1Z done total.
130: 200.... New zealot. 2Z done.
(310: 9 more zealots produced. Total 11 + 2/4)
This is an amazing post. Take a look at nada's whirlwind push where he adds many factories at once - he still makes vultures out of his two factories. Take this build for example:
2fac vult -> expand -> nonstop tank -> reap benefits of expo -> 2 fac -> pump
OR
2 fac vult -> expand -> stop tanks -> 2fac -> reap benefits of expo -> pump
However, even when you make the two facs, you are still pumping vulture and nonstop scv from the two ccs, nonstop unit production. This post along with discussion with some players clarified that the first is the better build.
About the 8lord, 9lord comment, I think this is similar to cutting an scv to put down an expansion CC down quicker, which is certainly weird if you are going for a fast expo build, where each scv in that early game matters.
Vs 1 fac CC, my recommendation is going to DT tech with two base vs two base (with his being earlier and him having advantage), then taking min only or another main w/DT w/shuttle backup (after making sure he builds adequate turret protection in his main/natural against DT tech). Him having to build turrets + academy + wait for that will give you the time to secure 3rd base. Unfortunately, it becomes a lot harder once he expands shortly after becuase he sees you have two base. Terran vs Protoss is generally considered difficult for Terran, but a good T is just such a pain in the ass 
|
|
|
One thing that is slightly tangential to this topic but still important nonetheless is that I used to debate whether I should get the dragoon sooner (cut a probe for a few secs), or build the probe and delay the dragoon 2-3 seconds. Most of the time I ended up halting the probe production to get the thing a few seconds faster.
THIS IS WRONG.
In most cases (non rush), it is far superior to get the probe, then the goon shortly after. Obviously if you are doing something rushy then you must forgo economic advantage for the potential to hurt his economy, and you should cut as much that you feel is necessary.
|
Yeah I find that if i start cutting probes to get things slightly faster, then I am much poorer, have to wait for next probe the time after I get the unit too, then have to wait for minerals for pylon - when i have enough, i can either build the pylon and miss another probe or build a probe and get a late pylon. THen I don't have enough for another dragoon and this trend goes on. Cutting a few probes early game ca be very detrimental
|
I found a better way to put it.
1Build worker = increase your capability to build units
2Build drop-off = increase your capability to increase capability to build units AND make each worker more efficient
3Build unit-producer = increase your capability to build units within the capacity provided by workers
4Build unit = increase your power
If there are any math geniuses out there. We call our function, f(x), which describes actual power in terms of attacking and defending. Unit-producers increase the derived, units increase the function's value. This is provided our g(x) is sufficiently high, which describes income. Workers increase it, drop-offs increase the derived (and the value through better workerspread).
Who said maths suck?
|
On May 03 2005 23:47 Babibo wrote:
I found a better way to put it.
1Build worker = increase your capability to build units
2Build drop-off = increase your capability to increase capability to build units AND make each worker more efficient
3Build unit-producer = increase your capability to build units within the capacity provided by workers
4Build unit = increase your power
If there are any math geniuses out there. We call our function, f(x), which describes actual power in terms of attacking and defending. Unit-producers increase the derived, units increase the function's value. This is provided our g(x) is sufficiently high, which describes income. Workers increase it, drop-offs increase the derived (and the value through better workerspread).
Who said maths suck?
Can you go further into detail on the 1Build Dropoff and what you mean by each of those 4 categories? Increase your power? Do you mean, unit count currently or potential unit count in the future? Thanks.
|
Adding lots of gates at once is good if your spending your money on something else than units and plan to produce a lot of units in the near future, since you won't be able to use those gates anyway. Usually if you expand or tech, once it's finished you'll be left with a significant boost to cash for unit production.
Most of the time when I play toss I go for a pretty long time with only 2 gates producing non stop for my unit needs, working my way up to cyber core. (Depending on MU, usually vs toss or possibly zerg.)
You use the units you have optimally and after a while you secure an expo (possibly skipping a unit or so) which you have to defend, probably with 1 cannon for detection. Your still using your two gates 100 % because you do things while your income increases due to more probes.
Then your expo goes up and you will notice a definite economic boost since you still only have two gates.
Now since you probably want to both increase your gateway count and tech and add extra probes you have a choice. Either you can mass from gates while adding gates slowly, produce probes and can tech slowly. This will mean that once your done your going to have x ammount of units and x tech in some near future.
But personally I find that you can tech really hard (either to templar tech, robo tech or perhaps citadel of adun + 1 grades from your forge, it really doesn't matter) while adding probes and keep producing units from your 2 gates your better off. This will mean that your tech comes in much faster than in the first situation. Then since your in a pretty good situation defensivly you can stop most attacks anyway so rigth then you don't need those extra units. Then once you paid for the tech which is pretty expensive you'll have a good ammount of probes so you stop teching and stop producing probes/units (you'll be close to max or maxed on probes anyway) and add 6 gates for the money which is coming in and once that's done you can focus all of your income and speed on just getting as many units out of those gates as possible.
I gives you more units pretty soon and much faster tech.
It's almost never usefull to be moderate with production. A handfull of units extra and slow tech is only better if your not sure that your expansion will hold, but usually the superior defensive postion along with a cannon or so will mean that you don't need those few extra units anyway, but your still giving up map controll to the guy who just added gates and pumped pure units and your still going to have later tech and as many units as the guy who teched hard.
|
i prefer, when i play t, to add on as you macro. thats simple enough, and most people do it. what you have to be good enough to do, though, is to put ur depots down on time and keep your units coming out of your facs or raxs (most likely facs) at all times. if you completely stop your production to add facs on, then you dont have enough bases. you should definitely have two bases+ with 6+ facs constantly producing, given that its TvP, or w/e.
|
Pause worker-producer production to add enough units to defend all your workers by following these steps, in the given priority. Pause unit-production to add enough tech to spend your resources in the most efficient manner.
These are very complicated considerations...
- Add a unit from unit-producer.
- Add a unit-producer. you conclude that this is the best way to go about things. Then you go on to state that the bottom line is the time at which you have so and so. Seems right.
never halt production in a unit-producer to add another of the same unit-producer Agree, but often its negligible. Also, building more then one unit producer before utilizing the first one may be fine.
To sum up, about macro:
-get what you want
-get it when you want it
*both combine in some form to include building placement
Figure out the most efficient way...
Those things combine to give you your overall macro game. Then things like moving units around the map, etc. come into play.
Before I typed something about 9gates, now I use ~6. 
Exalted ill try that strat vs 1fact expo. If im pretty surehes going 1fact expo ill just drop a lot or do an elevator. If I adapt to the 1fact expo build ill just go for massing hard off of 3 nex.
Take a normal protoss dilemma 13core or 14 core. if you build pylon on 15 your going to get a 17 goon if you doing things correctly, but the 14core goon comes out ~10sec later. Thus, the next goon is out later and so on. This dilemma is a simple example of the 2most important parts that make up macro that i showed above. I think I'll only go 13core if i want to go range first, and thus offensive from now on.
1Build worker = increase your capability to build units
2Build drop-off = increase your capability to increase capability to build units AND make each worker more efficient
3Build unit-producer = increase your capability to build units within the capacity provided by workers
4Build unit = increase your power This seems good. This has me recall of what ive heard of the NTT style of play. Mass scv as terran vs p, if you get in trouble you can usually just use scvs to get out of it as long as its not too bad. So terran has an advantage this way?
CuddlyCuteKitten makes some good, valid, commonsense, points
Babibo, much appreciated if you would explicate yourself and in commonsense ways.
|
KnickKnack, what were you trying to point out in this post?
" never halt production in a unit-producer to add another of the same unit-producer
Agree, but often its negligible. Also, building more then one unit producer before utilizing the first one may be fine."
Uhh... No. Babido has already covered the only time you would want to do this, and that is during when you stop them to afford your Tech and then build units once the Tech is complete. This discussion obviously does not apply to Zerg, who saves money and larva until a tech is complete to get many/9 of the unit quickly.
You also seem to have qualms about perfect timing. I didn't feel progamers were on such a higher level than top europeans, but eventually you'll realize the level of play that they have is very close to that level of play, especially with the top tier players (not like assem).
Mass SCV is not good vs P, the idea of good macro wasn't fully developed during NTT's time, is not recommended at all - I've lost so many games due to too many SCVs, it also gives you a misinterpretation of your power - many times I move out and in the replay I have the same control 140/140, but its mostly scvs when he has a huge amount of units.
Babido, don't dumb down your material for us - much of Starcraft involves complex mathematics and the reason for this post. We don't need commonsense things that we can see for ourselves.
|
I think an important thing to consider from a Terran point of view is the fact that scvs in the process of building a factory lose a lot of mining time. This could be another reason why you see Terrans in replays simultaneously build multiple factories at once after going 1 fact / CC. Instead of cutting mining time of 1 scv at a time when gradually adding factories, they instead have the scvs keep mining so that more resources can be mined for a set period of time allowing them to add all the factories at once. This will get you a set amount of factories faster than it would by gradually addng them. I know that it's also effective to just add 1 factory after opening w/ 1 fact / CC and slowly push out with tanks and turrets while gradually adding factories, but the two different ways you can add factories is a matter of when the Toss gets their expansion in if you survive the early game.
For Toss, I think it's more beneficial to gradually add gateways since probes lose very little time mining when warping in buildings, so cutting unit production to add more gateways doesn't seem very sensible.
Also, under no circumstances should a unit-producing building be added in place of actually building a unit.
|
well it depends in the distance, map, position and oponents BO. In an ideal situation I once maxed at 13:45 with 3 expos +main and cannons.
|
A critic on MPXMX's model. He is right I think in that you need to stop producing units at some point to build gateways if you want to avoid the "gap" in unit production after an expansion comes up.
However, usually this is purely theoretical. If you can feed 5.5 gates for example, you still need to get 6. Which means over the long run you have the chance to pump away your money you amassed when your expo comes in and gates are not yet built. Same goes for cannons. If you expand in PvZ, it usually never hurts to build a cannon or two to make up for your temporarily inadequate production facilities. Also getting upgrades can be timed to "eat extra money"
Alltogether this makes the "add gates when money stacks" strategy superior to "build gateways in advance". You are much more flexible in when and if to get tech. It is also MUCH easier to pull off perfectly timed.
MPXMX analysis might be valid in really tight scenarios like two toss players boring each other to death with 3gate goons vs 3gate goons. Expanding is dangerous in this situation. If the other player sees you expanding and does full probe stop + 4th gate, you're up to be overrun. The gap will basically kill you. Doing a perfect unit stop build should allow you to expand slightly faster in that case.
Any competent Protoss out there to confirm or counter this ? -.-
|
Say i have 250mins. I decide i want a gateway and a zealot. I build gateway -> i train zeal. By doing this Im not producing in one gate needlessly yet stick up another one. But the gate and the zeal will both start in less then a second, so the time diffrence is negligable.
As for why building more then one unit producer before utilizing the first one may be fine: building 2robos before building a shuttle or reaver or ob. Building 2gates before making a zealot, etc.
qualms about perfect timing?
I have not typed anyting about the diffrence in skill level between progamers and top foreginers. I doubt you know what i realize and/or will realize on the subject.
The point i was getting at for scv's is that scvs are the strongest worker, thus if terran builds too many(for whatever reason) he can use them more effectively to keep himself alive then zerg or toss can use their workers. So terrans have that advantage. I agree that a mass scv style is not the way to go.
Explicating more should certainly not be a dumbing down of the content. Using so-called common sense statements need not be dumbing down either.
Grass.nS), good.
FreeZEternal, ive shown a example rep of one of my older ways of macroing, i had +1done, 3 cannons i think, 9gates, maxed, 4nex at 13:30, stargate warping in. And i was not takign advantage of many very early expos or probes. Expo at 5:30, expo at 8:30, expo at 12:00. all reasonable times in game. Im prety sure i could get maxed by 13:00 with 6gates and a similar build.
He is right I think in that you need to stop producing units at some point to build gateways if you want to avoid the "gap" in unit production after an expansion comes up. Well that was an older model. The ultimate goal is to be able to pump almost constantly.
Alltogether this makes the "add gates when money stacks" strategy superior to "build gateways in advance". You are much more flexible in when and if to get tech. It is also MUCH easier to pull off perfectly timed. "the add gates when money stacks" way is superior for overall macro, but if you want to get a certain ammount of units or tech by a certain time you may build them "in advance" as it were.
Im not sure about the 3gate goon vs 3gate goon. I dont usuallydo 3gate goon, for one. Two, their are things were not taking into account, like parts of efficency, who micros better, etc. But, basically if one guy masses harder then the guy who expanded and manages to kill most of the other guys units(or contains him) and nexus while getting one of his own hes in great shape to go on to win. While if one of them is able to expand and hold it while taking reatively small losses hes in great shape to go on to win.
|
My model didn't account for teching,
Now that I look at it, teching probably fills the gaps that are created in production when you gradually add facts or gateways. And teching is important... That speaks in favour of gradual build-up. Also, gradual addition results in more units until that special time when your many extra-early factories or gateways catch up and overtake it. That means gradual = safer.
So basically, the only time it seems worthwhile to add many producing facilities at once, and perhaps pause production for it is
-right before taking an expo and increasing your economy
-when you don't intend to tech in near future
-when pausing production is not likely to result in death
All those must be true to make adding multiple production facilities at once beneficial for macro, in a real game. The most common scenario for this is when terran is taking his natural or 2nd natural in tvp, when the terran already has turrets, siege mode and mines (most of the significan tech at this point in the game), and when cilffed tanks provide excellent defense that allows a pause in production without death.
For a protoss, a similar non-dangerous period for decision makign occurs when the terran goes fact-cc, but this time, I think teching plays a bigger role
|
On May 06 2005 12:47 Knickknack wrote:
Say i have 250mins. I decide i want a gateway and a zealot. I build gateway -> i train zeal. By doing this Im not producing in one gate needlessly yet stick up another one. But the gate and the zeal will both start in less then a second, so the time diffrence is negligable.
If you have 250 mins, there is no problem. The question is about what happens when you have 100 mins, do you wait to 150 to build a second gateway and then spend 100 on a zealot? Or spend the 100 on zealot then wait to 150 for the gateway. The answer is obvious.
On May 06 2005 12:47 Knickknack wrote:
Im not sure about the 3gate goon vs 3gate goon. I dont usuallydo 3gate goon, for one. Two, their are things were not taking into account, like parts of efficency, who micros better, etc. But, basically if one guy masses harder then the guy who expanded and manages to kill most of the other guys units(or contains him) and nexus while getting one of his own hes in great shape to go on to win. While if one of them is able to expand and hold it while taking reatively small losses hes in great shape to go on to win.
a) We don't care what builds you do, he used this example to talk about when both players are macroing equally
b) Micro and other stuff you can do (go reaver lolz) are NOT part of this discussion - this discussion is about getting units as fast as possible.
c) If he manages to hold his expand and kill the attack forces he wins the game, period. 1 base vs 2 base is not pretty.
Interesting point by grass talking about how scv mining time does get taken into account when building them slowly, but I think that having 2 scvs building, then 2 building is better for the economy then 4 at once - especially this should not be done at the cost of stopping unit production. However, if you need to go for a quick attack, by all means, stop SCV PRODUCTION to get the 6 facs down, but don't stop units at the factory. That is what we all should have learned from this discussion.
|
you guys have too way much time on your hands...
|
I dont care if you care what builds I do. Seriously, lay off. You seem to like to try and be aggressive in your posting. And typing that "we" dont care about your builds is extremely presumptuous, as I posted in a medium were all tl persons could view my comments. Either explain who "we" is or dont speak for anyone who may read my comments.
I was simply showing that he was not considering important parts of the game. Which, if he wanted the most complete view, he/she should include.
c), not necessarily, though, highly probable.
|
|
|
|
|
|
EPT
