Terran early game build order test results - surprising

I had some questions about early game build orders with Terran so I decided to do some testing with qxc's build order tester. I posted a long version on my blog, but I wanted to share my results here in the hopes that it will help someone in the community. This is simple stuff, and will mostly be applicable to beginner's.

Here's the three things I tested:
1. Difference between moving your SCV's first, then training one and training first, moving second (at the very beginning of games)
2. Does it matter if you rally your SCV's yourself, or should you just let the AI do the work?
3. Is it best (with Terran) to build a supply depot at 9 food or 10 food.

All pretty simple and straight-forward, but the results I got surprised me. I tested each of these scenarios quite a few times and found a pretty good 'average' result for each one.

First test - Which is more efficient: train then move or move then train.

I found that at about the :40 mark I was able to have 5-15 more minerals when I trained an SCV first, then told my guys to mine. Plus my SCV's were coming out a little earlier.

Second test - Manually rally my SCV's to the mineral patches, or just rally to the nearest and let the AI figure it out.

While the AI in Starcraft 2 is MUCH improved over the original, it still doesn't do as well as manually micro-managing your early game mining. I found that I was able to squeeze 10-20 more minerals out of my workers when I manually told them where to go versus letting the AI do it.

Third test - Supply depot on 9 or 10 food?

I had a good guess on this test, but I wanted to know for sure. And again this is only for Terran and doesn't apply if you're doing a 9 or 10 racks build obviously.

So I found out that again I was able to get 10-20 more minerals by building my depot at 10 food. A small difference but a decent result. Plus since I wasn't waiting on minerals my SCV's were coming out a few seconds earlier doing it this way. Building on 9 food just isn't that efficient.

Now these test results do seem small, but they add up! If you put these 3 little things together you're ahead by 25-55 minerals and a few seconds of SCV production, which can make a pretty big difference if you get rushed.

Hope this helped a bit!
Taerix

 

Nice work. The mention of AI being less effective than manual micro is totally true - as can be seen in Pro's perfection of the "split." BTW, if you didn't realize, the very beginning of the game sending your workers to mine is called the "split"

 

How the hell can you have been surprised by these results?
Everything here is effin' straight forward!

The moving SCV's first thing is totaly understandable since you only lose that mining time.
Not building SCV first is totaly understandable because you'll lose slowly over time and at... oh! the 40 mark, you'll have a loss.

The second test isn't necessary and has no surprising results as well. If you don't move the rally point and leave it at one mineral patch, that means the SCV is going to have to travel to another mineral. For example, the first SCV's mine at the first 6 patches, there are 2 left. Instead of rally to those that are left you let the AI wait before mining by sending it to the first one.. whatever you get the idea.

The third test is also a no-brainer. You're skipping SCV production time when going 9 supply.

 

I think if you've been playing the game for a long time then some things become clearly obvious. However there's a lot of new players now and much more to come when the game launches that just aren't quite sure about the optimal way to do things.

Obviously this post isn't for the pro players that already know all this stuff, just trying to help some newer Terrans out.

 

I had the idea of doing this, but I'm too lazy ^^ ty

 

I support these findings.. As I have always seen pros train their workers until they are max supply, zerg even goes 11/10 or even 12/10 using the extractor trick before even building up their food/supply count.

As for the split, i feel like it depends on your micro. If you can quick click your main and train a worker without losing half a second (quarter of a second is good), you should do it first. The split is the hardest.

when you are NOOB:
use the AI to split them up, which is sending all 6 workers to the same mineral patch

when you are a SCRUB:
send 4 workers to one mineral patch and 2 workers to another patch on the other side.

when you are GOOD:
send 3 workers to one patch of minerals that is between two mineral patches, same thing with the other 3 workers.

when you are PRO:
send 3 groups of 2 to separate groups of mineral patches.

i kind of struggle on this part due to my computer lagging every beginning of the game. So im a scrub, at the start but i catch up in APM

 

Im surprised that it took this long to get something like this out, it is always nice to have somebody confirm the universal terran start

 

all these findings apply to all races btw.

 

That's totally true for the first two things I found out. The last one is Terran specific since supply mechanics are different for each race. I don't think you'd want to wait as long to build a pylon as protoss since they only start with 10 supply.

Not sure though since I haven't tested it with toss yet.

 

Protoss builds up to 9 probes, then when they get 100 minerals, they throw a pylon down, and then a ten probe.
If they wall off against zerg, they send a probe out when they have 30-70 minerals.
There is some builds where you build a pylon at 8, and then chronoboost, or wait to you have ten, but nine is normal.

 

Im too lazy to select all my workers manually at the beginning haha

 

You forgot about:

when you are GOSU:
send 6 groups of 1 to separate groups of mineral patches


Korean pro gamers do that

 

I started out with a split of 2 groups, but when i heard that the AI-Split works quite good, i started to send all my Probes to 1 mineral and let the AI split them.

I always thought about building my Probe first and then doing the split, but never tried out. Nice to have someone who did (and shares his experience). Great work. Would it be possible to post a link to the Map you used or maybe posting the map itself? Heard about that map somewhere but didnt find it.

By the way:

What kind of a split have you done? 2groups of 3? 3 groups of 2? and how fast are you? I mean, it doesnt make sense for a noob like me to do a split of 3 groups of 2 since im too slow. 2 groups of 3 is the best i can do without taking to much time!

 

do you have a replay you can post showing that?

 

Yah I don't see how you could possibly click fast enough to do even 3 groups of 2, the AI is good enough that it seems like it would be faster. Doing 2 groups of 3 is totally possible though, and even pretty easy for someone slow like me =)

 

SC1

Go search Boxer FPVOD replays. Although they may only start with 4 workers, boxer probably can do it judging from what I see.

 

I'm not going to go searching through a lot of replays for an example. I looked one up on youtube and the first game I came across he only did 3. It looked like he did 4, but pausing-playing the replay it looked pretty clear the last two SCVs ended up both getting assigned to mineral patch 4 and the AI reassigned one to patch 3.

Even assuming that he had a bad start there, it seems pretty pressed to micro 4 much less 6.

This isn't even getting into what seems to me to be a rather major difference in worker starting placement between sc1 and sc2......

 

koreans don't do that, even if they did, it reaches a point, where they have lost too much time.

 

ya well done i saw thias post on teamliqiud too. well done. i