Optimize microinstructionData

[lilweege]

Addresses #11.

A profile of the code revealed a couple inefficiencies. Most of the time in update was spent copying and iterating over vector. The following are a few of the culprits.

Copying a vector:

Astro8-Computer/Astro8-Emulator/main.cpp

Line 435 in 20e25a7

vector<bool> mcode = microinstructionData.at(clamp(microcodeLocation, 0, 2047));

Copying again and iterating:

Astro8-Computer/Astro8-Emulator/main.cpp

Line 448 in 20e25a7

int readInstr = BinaryVecRangeToInt(mcode, 9, 11);

Astro8-Computer/Astro8-Emulator/main.cpp

Lines 935 to 947 in 20e25a7

	int BinaryVecRangeToInt(vector<bool> vec, int min, int max)
	{
	int result = 0;
	int base = 1;
	for (unsigned int i = max; i >= min; --i)
	{
	result += vec[i] * base;
	base *= 2;
	}
	return result;
	}

mcode is always 14 bits long, which means replacing each vector<bool> with uint16_t is low hanging fruit in terms of optimization. Extracting the int from the bits becomes as simple as a few shifts and a bitmask.

This change is enough to make a noticable improvement to most programs. For instance, on my machine the time to run rainbow_gradient.armstrong drops from ~3.5s to <0.5s (in the debug build).

[sam-astro]

Wow, that is a huge improvement, thank you!