Removed all instances of gBitTable[x]

[hedara90]

Description

Removed all instances of gBitTable[x] and replaced them with (1u << (x)).
Added a migration script for replacing gBitTable[x] with (1u << (x)).

Issue(s) that this PR fixes

#5007

Discord contact info

hedara

[tertu-m]

This is kind of pedantic, and other people could overrule me on this, but would it be possible to do at least some cleanup on the results? I'd like to get some more feedback from others as to whether that's necessary.
Another thought I had is maybe it would be better to have the template be something like (1u << (x))? This usually wouldn't matter but I worry about edge cases.

[hedara90]

I'm unsure about what cleanup you're referring to.
I agree with the 1u thing, what's the best format for it? (u32)1 or 1u?

[tertu-m]

I guess as far as cleanup I meant removing extra parens and stuff. At the very least there was a case where it was doing ((1<< (x)) << 16) and I think that should be rewritten into (1<< (x+16))

[hedara90]

Oh, like that. I did not parse that in my head yesterday.
I kinda went with excessive parentheses, tests failed when I didn't.

I don't know if it's possible to make it output clean conversions, I can clean the usages in Expansion, but I kinda want to be on the safe side for the script.

[tertu-m]

Oh, like that. I did not parse that in my head yesterday. I kinda went with excessive parentheses, tests failed when I didn't.

I don't know if it's possible to make it output clean conversions, I can clean the usages in Expansion, but I kinda want to be on the safe side for the script.

Oh yeah, in the general case, if you're trying to do a simple find and replace, the extra parentheses are necessary, and I don't see any issue with that. People can clean their code up later, and there won't be any performance penalty for it. I was just suggesting doing at least some cleanup for the expansion codebase.

[tertu-m]

I'm unsure about what cleanup you're referring to. I agree with the 1u thing, what's the best format for it? (u32)1 or 1u?

I'd do 1u myself?

[hedara90]

I was just suggesting doing at least some cleanup for the expansion codebase.

Yea, my brain was just locked in on somehow doing the cleanup in the script.

[hedara90]

Script output cleaned up

[tertu-m]

So far this looks good. I'll fix up the conflict and then merge it.

[AlexOn1ine]

not a review but what does 1u mean?

[AsparagusEduardo]

not a review but what does 1u mean?

Pretty sure it's unsigned 1

[AsparagusEduardo]

Now the question is, is it needed?

[hedara90]

It takes ~twice the time to make a gBitTable lookup compared to a bitshift.
And how did the table lookup work with the pre-fetch buffer?

[AlexOn1ine]

anything blocking this?

[AlexOn1ine]

@hedara90 can you update the description that a 1u is used instead 1?

[hedara90]

@hedara90 can you update the description that a 1u is used instead 1?

Done

[tertu-m]

Now the question is, is it needed?
I would say it’s not needed, but gBitTable’s existence in the first case is a bit of a misfeature.