I understand that there are many performance improvements to be reaped from using low level languages.

While this can sometimes be true, it may not be as straight-forward as you think. Lower level technologies offer you finer control over your software, the potential performance benefits can be gained from using a solution that is more specific to your particular problem rather than something more general that has been created to handle various different problems. These potential advantages are far from automatic however, and it's entirely possible (even probable for a non-expert) to instead create a less optimal solution.

In the case where you've done a particularly good job of choosing your higher level technology and it's a good match for your specific problem, you would even find that you would simply have to replicate exactly the same solution if you decided to work at a lower level.

Choosing lower level technology (whether that be language, library, engine or whatever) hoping for performance reasons is not a good idea unless you have the experience to be able to take advantage of that. You should also consider whether or not the potential benefits will outweigh a potentially significant increase in development time.



As for learning a bit more about how things work at a lower level, I'd recommend taking a class or checking out some reading material on algorithms and data structures. You might also consider learning some basics of how simpler hardware works, and learn a little bit about assembly.

Also I'm guessing that the lower level I go, the more probably I actually break something right?

Absolutely -- with great power comes great responsibility. Languages like C++ assume the programmer knows what they are doing, and if that isn't true you don't have some of the safety nets that other languages can provide. In the hands of a skilled programmer this means they have increased control and can potentially gain some performance improvements by tailoring the programs behaviour to exactly suit their own needs and not waste time trying to handle more general cases that don't actually apply. Unfortunately, in the hands of a beginner it means there are more potential mistakes to be made, and more you need to be aware of to avoid said mistakes.

It sounds like you're probably on the right track, and if you're planning to study computer science at college there should be some classes available that will teach you about some of those lower-level topics you're interested in.

Yeah, so just "make things from scratch" for knowledge's sake, but when working on a deadline embrace pre-existing code right?

Absolutely, that's an excellent approach to take. It's great to learn new things, and knowing more about how things work at a lower level can often improve your skills even when working at a higher level, but if you want to produce good quality software without wasting a lot of time there's absolutely no reason you shouldn't take advantage of existing technologies that meet your needs and have already been bug-tested, etc.

Java is a good language for someone who is just a beginner of learning program, but it dose not suit to making game because its high level and low performance.

The creators of Minecraft, RuneScape, Puzzle Pirates and Wurm Online (amongst others) would probably disagree... Java is just fine for games, and whilst it isn't for everyone there's absolutely no reason to spread nonsense about it being "not suited to making games". The fact that it's higher level is potentially a good thing (as has already been covered at length in this topic), and while there can be some additional performance concerns the idea that it's "too slow" for games is largely just out-dated and no-longer relevant.

Well somebody has to write a game engine from scratch. If you look at the previous examples of idTech3, Unity, UDK, etc - yes, they did get a lot of subsequent reuse, but they all began as engines written from scratch. If nobody wrote an engine from scratch these wouldn't exist to be subsequently used by those who don't want to or need to write one.

So you need to sit down and think about what exactly it is you want to achieve. Where is the payback for you? If you want to make a game then reusing an existing engine is likely to be a good option. If you want to learn about some specific aspect of technology, then modifying an existing engine might be more appropriate. If you want to get deep involvement in all aspects of what an engine does, then writing your own may be what you want. But it's firmly a horses-for-courses decision, and only the person making the decision really knows which one is right for them.