It has come to my attention that current C compilers compile C to assembly and then assembles and links the assembly code. What about a new type of compiler, which just develops an intuition on how to generate the raw binary. I'm thinking of creating the next generation C compiler, it's a trained neural network which compiles by intuition rather than preset assembly instructions. It should probably compile code much faster than current C compilers and we can market it as organic and natural since it's just like a C compiling animal as opposed to the current compiler robots.
Most C compilers translate the source to an internal, intermediary format that is easier for the optimizers to transform. The result is then translated into the target assembly language.
Name:
Anonymous2011-03-13 19:50
Tie it in with AVX and you're good!
oH wAiT
Name:
Anonymous2011-03-14 2:48
we have clang already.
It should probably compile code much faster than current
Yet another weaboo thinks that Neural Network are cool and silver. IHBT,
Compilers are good at what they do. Do you think a human (huge biological neural network) is faster than a compiler? No.
You can omit the "compile to assembly text" part, but you'll still need to generate some sort of symbolic assembly. Some Lisp compilers actually omit the "compile to assembly text" and just generate internal in-memory assembly which is assembled using its own symbolic assembler. If you actually knew enough about compilers, you'll realize that the assembler is actually one of the least resource intensive parts of a compiler.
Neural networks are nice for certain things, but I don't think this would be the best solution for what you want, although I wouldn't rule it out completly as a strategy for certain optimization tasks.
Name:
Anonymous2011-03-15 14:11
test
Name:
Anonymous2011-03-15 14:13
I love compilers
Name:
Anonymous2011-03-15 14:27
anii
Name:
Anonymous2011-03-15 14:27
anii
Name:
Anonymous2011-03-15 15:53
>>12
Human brain has billions of neuron cells. Each cell functions as a separate processing element. Human can do millions of parallel searches each second. Each fucking second mind searches giant database of complicated patterns. You can't simulate this process with your toyish desktop PC.
Name:
Anonymous2011-03-15 15:55
>>17
Its like a raytracer, that runs 1,000,000 frames per second on a complex scene.
>>17
I know, although it's more like retrieving and storing cached thoughts. What I was claiming is that humans are too SLOW to be compilers (or neural networks of that architecture) and computers are too SLOW to be humans. It sounds paradoxical, but it's not, CPUs are designed to run a sequential stream of instructions fast, they do this by implementing the CPU in extremly fast parallel circuits/cells, human brains are a physical architecture implemented in neurons which are rather slow compared to normal electronics, but the sheer size and actual path lengths (maximum depth is some 20 neurons or so for 1 second (that is, it takes 1 second for information to pass from one neuron to another, and this path is 20 neurons wide; in practice the whole network is activated, information is passed through the entire network, not just a single path), but I'll need to find the exact citations on that).
tl;dr: Humans are good at performing certain types of tasks requirng general intelligence. CPUs are good at performing preprogrammed tasks much faster than any human could. Compilation is one such task. A human is not a good CPU (way too slow). A CPU is most certainly not a good human (way too slow). The architectures have different "purposes" (humans don't have a purpose, but they are an adaptation which resulted in general intelligence, which aids them in surviving and much more).
Name:
Anonymous2011-03-15 16:38
>>19
>more like retrieving and storing cached thoughts
You can't cache pattern matching.
Open http://www.google.com/images?q=random
Each picture there would require matching against all pictures already in your brain. You simply can't cache this.
Name:
Anonymous2011-03-15 16:39
Apparently the feeble minds of /prog/ are not ready for the compiler revolution.
Also I think it's funny that nobody mentioned how it would be completely unreliable as a C compiler since it would only sometimes produce expected behavior even if trained correctly.
>>3
This type of compiler would already perform every optimization possible, please leave /prog/ and perform harakiri on yourself, this is the only way for your family to save face.
We here at the SUSSIX dev team (Me) plan on implementing this idea into SCC - SCC COMPILES C. As you may have noticed we (actually Me) are very glad in recursive acronyms, this is because every SUSSIX program has infinite stack space as they only run on Turing Machines.
THIS HAS BEEN A MESSAGE FROM THE SUSSIX DEV TEAM (ME)
Name:
Anonymous2011-03-15 16:51
LoL if it's not broke don't fix it. Compiler construction is actually one of the most well-researched fields in Computer Science.
>>20
It's a bit different than that. Statistical pattern recognition as well as pattern storage is what we could call the high-level process performed by our neurons (which perform their task at a much lower level). It's not the same caching your web server performs to store visited pages or the same caches your memoized function has. It's a rather ambiguous cache which remembers patterns and patterns of patterns and patterns of patterns of patterns (and so on), it does this in a messy biological way. This can be seen as a form of caching, even if it's not what you expect from traditional digital caches. >>24
Nature gave us adaptations which lead to us surviving and reproducing. You either survive or your children survive, that's just the cold fact which produced everything that is alive in our environment. It is not a purpose. It's merely a fact. Nature has no intentions, it just is. If you survive, you continue to exist. If your children survive, some of your genetic code will continue to exist. The process has no purpose, it just is.
I never said that nature or the process intended anything, the universe obviously has no intents. We, the humans however have a purpose, which is to procreate.
I'm not going to read that as I'm actually not interested in this discussion or your opinions, I realize that it was foolish to presume we had the same definition of purpose.
And in case you were wondering, I do not claim superiority in this discussion, I just claim neutrality.
>>25
If it was ambiguous, you would've easily confused cat with dog. But it's precise enough to quickly select matching image from thousands of complicated alternatives.