Optimize Rubyboy::Bus

[byroot]

Uses the same style as: #24

main:

Ruby: 3.4.1
YJIT: true
1: 3.882817 sec
2: 3.83399 sec
3: 3.835251 sec
FPS: 389.5409804902295

Ruby: 3.4.1
YJIT: true
1: 3.981531 sec
2: 3.804446 sec
3: 3.823725 sec
FPS: 387.606848134431

This branch:

Ruby: 3.4.1
YJIT: true
1: 3.503513 sec
2: 3.437195 sec
3: 3.52252 sec
FPS: 430.07760129092094

Ruby: 3.4.1
YJIT: true
1: 3.620516 sec
2: 3.582359 sec
3: 3.484475 sec
FPS: 421.05854117250766

[byroot]

Redoing benchmarks after rebasing now that #24 was merged:

Before:

Ruby: 3.4.1
YJIT: true
1: 3.476809 sec
2: 3.597895 sec
3: 3.617027 sec
FPS: 420.8860099454429

Ruby: 3.4.1
YJIT: true
1: 3.606878 sec
2: 3.63947 sec
3: 3.637668 sec
FPS: 413.45032936372013

After:

Ruby: 3.4.1
YJIT: true
1: 3.362835 sec
2: 3.319436 sec
3: 3.426609 sec
FPS: 445.15317226042845

Ruby: 3.4.1
YJIT: true
1: 3.394805 sec
2: 3.419763 sec
3: 3.438782 sec
FPS: 438.8809511037856

[sacckey]

I apologize for the delay in the review.

I've left some minor comments, so please make the necessary changes. Thank you.

[sacckey]

Would you mind changing it to be written as follows, without using an ampersand:

when 0xf
  case addr >> 8
  when 0xfe
    ~~
  when 0xff
    ~~

[sacckey]

Just a minor detail, but since it was originally written as when 0xff10..0xff26, it might be more readable to write it starting with the smaller value, like if last_byte >= 0x10 && last_byte <= 0x26.

[byroot]

It was done this way for performance reasons. Almost all values are >= 0x10 few <= 0x26, so it avoid a comparison in most cases.

If Ruby had a very advanced optimizing compiler that wouldn't be needed, but it's not the case yet.

But if you fell very strongly about it, I can change that.

[sacckey]

I see, got it!

I don't have a strong opinion, so it's fine as is.

[sacckey]

Thank you for the optimization!
It runs much faster now thanks to you.

I'm planning to release it as version 1.5.1