|
On September 25 2010 04:21 Emperor_Earth wrote:
I find it very interesting that new SC2 players completely ignore over a decade of thinking that went into SC:BW and immediately think that double rallying to nat min lines is preferable to maynarding. Even the semi-pros/amateurs in the GSLs do this and it boggles my mind to no end. SC2 is a different game from BW. While some of the core concepts remain valid, it's useful to re-evaluate long-held assumptions in light of the new game mechanics.
You can't rally workers to a mineral line in BW so there may be additional macro issues at stake that make maynarding more or less effective in SC2.
Thinking through some analysis is much better than just blindly accepting old dogma as still relevant.
|
Because of this, it is only a good idea to drop a base below 16 workers if you're maynarding to a gold, and even then only if you have to.
Or if you want your main base to take longer to mine out which is a very significant issue.
|
On September 25 2010 03:48 Cell.cell wrote: OK. I used an online stopwatch to find these numbers. At full saturation (16 workers on 8 mineral patches) your income is about 800 minerals per minute. It takes about 15 seconds to fully maynard scvs from your main to natural on LT. If you maynard 8 scvs immediately (half your workers) you lose 800minerals/2*15s/60s = 100minerals of mining time. If you rally both cc's to your natural, you lose 800/16*15/60 = 12.5minerals every 17 seconds (scv build time), until you have built 8 scvs, for a total of 100 minerals lost over the course of 136 seconds (2minutes 16 seconds).
16 workers at 8 patches is not full saturation on most maps. Full saturation requires 3 workers for the patches that are further away from the town hall. I know this is nit picking but its still worth mentioning.
|
whats the most effective way of counting your workers at a base for saturation? hotkey them?
|
On September 25 2010 05:10 CoFran wrote: whats the most effective way of counting your workers at a base for saturation? hotkey them?
Usually just boxing your workers at minerals and counting rows. If there are two full rows of drones, you have 16 workers mining. Any more than that is oversaturation.
This is an interesting look at this, because usually I only maynard the drones I have over 16 and change rallies without thinking about it. It seems like it's a much more interesting choice now. Not that I have been thoroughly convinced either way because there are arguments for both.
Also, I thought that FULL saturation was 3 workers per mineral patch. I think there was even a little tip for this in-game. Is it just the point that any more than 2 per see such insignificant gains that it's not worth doing?
|
Full saturation is 3 per patch but the third mines less efficiently. You never want to just stop making workers at 16 because the extra income is worth it, but if you have an expo you'll get more minerals per minute if you balance it out so they are mining as efficiently as possible.
|
Just to clarify for some people that might use the word maynard in other contexts than SC, it's a term describing the act of transferring your workers to a newly finished expansion. Foreign community calls it "Maynard"ing because it was generally started by a foreign gamer named Maynard in the early ages of Korean progaming. Don't use this word in this context in real-life or in any academic papers since the word has no meaning outside of starcraft (expect a male name).
Topic: Although this is nice to know, it doesn't much in-game uses because fast expand isn't viable, most people would have their expansions when they have at least over 25 workers anyway.
|
On September 25 2010 05:54 nitdkim wrote: [...]
Topic: Although this is nice to know, it doesn't much in-game uses because fast expand isn't viable, most people would have their expansions when they have at least over 25 workers anyway.
I guess you don't play zerg?
|
Who the fuck has an expansion complete and needs to maynard to it by the time they only have 16 workers?
I'm sorry, I agree with the mod. This thread is useless.
|
I ALWAYS maynard a majority of my workers to an expansion (and 3 per patch for gold). I hate getting mined out in my main, and by focusing my mules on my expansion I can keep up the income from my main base way longer than most of my opponents who just maynard 6 or so workers and then still have full saturation in their main.
|
Another thing is that if you rally both to the natural, you lose mining time in the travel distance of the scvs going from the main to the natural.
On September 25 2010 06:00 Vexx wrote: Who the fuck has an expansion complete and needs to maynard to it by the time they only have 16 workers?
I'm sorry, I agree with the mod. This thread is useless.
Have you never done 1gate FE, 1rax FE, or played Zerg?
|
The automine AI is bad enough that a split down the middle will be more efficient than all at one, even if your main has exactly 16 workers. You still loose money on the cc, but you gain an additional worker production queue. That said, if you have 24 workers, you will make that money back rather quickly. From my data, if you have 26 and split 13/13, you get about 260 extra mpm (pays itself back in under 2 minutes), plus you can add workers in half the time. If you are at exactly 16 and you split 8/8, you get about 20 extra mpm, so it would take 20 minutes of game time to pay for the investment, ignoring your ability to produce workers in half the time.
|
On September 25 2010 06:28 teamamerica wrote: Another thing is that if you rally both to the natural, you lose mining time in the travel distance of the scvs going from the main to the natural.
The OP covered that, did you read his post?
|
Prior to 2:1 saturation, maynarding is definitely less good in SC2 than it was in BW thanks to the smarter workers. There is some advantage to 8:8 versus 16:0, but it's not very large.
I still maynard workers to balance saturation, as I want to mine out my main as late as possible. There's not much advantage to be had here, but there isn't much cost either.
|
There are multiple threads like this from BW. Look them up, you'll find your answer.
|
On September 25 2010 03:50 Adaptation wrote: Can you tell me how much more money i make in 3 minutes with 10 workers at main and 10 workers at expo, vs 20 at one base over 4 minutes?
Im quite certain the 10-10 wins this, despite the 20-30 seconds of no mining when transfering workers.
Until you get past 16 workers, mining speed is based purely on how far away the patch is. It's not until you start getting 3 workers on a patch that you start losing efficiency.
So 10 on each of two bases will mine slightly faster than 20 on one base, but the difference is pretty small.
There are some misconceptions in the original post, though, which seems to be claiming that you lose minerals because of the travel time.
Travel time is mostly irrelevant, because you should be doing one of two things: maynarding workers to the expansion, and building new workers at each base, sending them to the nearest patches, or leaving workers at the main, and rallying workers from both bases to the expansion. So if you don't maynard, you still have to pay the travel time cost, as workers go from the main to the expansion.
That cost does come a little later when not maynarding, but that's just another argument for not transferring.
The real cost, and I did extensive tests on this during beta, is that workers need to "settle in" to a routine, where they're just going back and forth between a single patch and the base. Any workers who are shuffling around looking for places to mine are obviously not actually mining, and are not contributing anything at all to your mineral input. It's not until they settle in that they start adding to your income.
Every time you add a new worker, there's some amount of time between the time that it's built, and the time that all the workers are settled and fully mining. Every time you add a worker, the routine gets broken. The new scv will pick somewhere to mine, but will likely interrupt another worker who had already settled into that spot, and that worker then has to go find somewhere, which likely interrupts another worker, etc.
The more workers you have, the longer that settle-in time is. And it can get *really* long. For example, on lost temple, where 22 workers is the maximum that will ever be actually working at once, I've seen it take as long as two full minutes for everybody to settle in after adding the 21st and 22nd worker. That is two full minutes before the worker you built even BEGINS contributing to your economy. Average time was less than that (I don't remember the timings, it's been so long, but somewhere around 60 seconds seems to be sticking out), but it absolutely shocked me to see that the last worker could take that long before it even began to contribute.
If you have a saturated main, with all the workers settled in, and not bouncing around, you then take half of those workers to the expansion, and start building workers at both bases, you're messing up the routines in both bases, rather than just one, which has a penalty associated with it.
In the tests I ran, I started with 20 workers at a main (22 is the max on many maps, but those last two workers can take a really long time to settle in, so it may not even be worthwhile to hit max saturation) fully settled into a routine, and compared the difference between taking half of them to the expansion, and leaving them there and just rallying new workers to the expansion, until both bases had 20 workers each.
Transferring showed a very consistent cost of around 150 minerals, due to the "settle-in" effect.
That's where the real cost of transferring comes from.
|
On September 25 2010 03:57 Cell.cell wrote: If you are at or below saturation, each worker adds linearly to your income. During maynarding, the remaining half scvs at your main have an income of 800/2 = 400 minerals per minute. In that 15 seconds of maynarding, your workers at your main mine 100 instead of 200 minerals. Once the half of your scvs get to your natural, your income returns to 800.
This is only true up to 16 workers (2 per patch) the third worker on a patch decreases the efficiency. Not nearly enough to want to stop at 16, but it does have an effect.
|
On September 25 2010 04:09 disco wrote: I did some very easy tests for you, here's some numbers.
16 probes, split evenly among 2 bases will generate around 640 to 680 minerals 16 probes, all on one base will get you 580-640 minerals.
Having done very extensive tests on mining rates and the effects of transfers, I have to assume that you didn't account for the "settle in" time in those tests. Probably you just sent 16 workers, measured minerals after some amount of time, and then did the same thing for 8 workers on each base.
16 workers will take quite a bit longer to settle in than 8, and you'll probably have multiple bouncing workers initially, while they try to establish a routine, while 8 will just immediately go one to a patch.
If you build up to 16 one at a time, you only ever have one worker bouncing at a time, so that effect will be minimized somewhat.
Also, as I'm sure you found out, it's EXTREMELY difficult to measure mining rates by looking at the mineral count, which is why your tests show such a massive range. When I was testing mining rates, I focused on the trip time instead, and can say with complete confidence that 2 workers on one patch mine at exactly double the rate of one worker on the same patch.
|
Most of the time your getting an expo when you have over 16 workers anyways so I don't see this being a huge deal but I guess it makes you think more about how much you maynard
|
On September 25 2010 04:20 Cell.cell wrote: Yes I did not take into account the position of the patches themselves. Is it true in general that the natural has more close patches, or was this the particular map main/natural combination you chose to do your test?
It's not globally true, but is very map specific. For example, I believe the Metalopolis naturals have one more "close" patch than the mains.
|
|
|
|