BitFieldView: Video Changes

[Minty-Meeo]

This PR is part 2 of splitting the monolithic PR #10610. It is dependent on PR #10846.

[Minty-Meeo]

This pull request increments GX_PIPELINE_UID_VERSION (GXPipelineTypes.h) and STATE_VERSION (State.cpp).

[Minty-Meeo]

What a headache that rebase was.

[dolphin-emu-bot]

FifoCI detected that this change impacts graphical rendering. Here are the behavior differences detected by the system:

• burnout2-vehicletextures on ogl-lin-radeon: diff
• custom-brawl-char on ogl-lin-radeon: diff
• fishing-resort-map on ogl-lin-radeon: diff
• fortune-street on ogl-lin-radeon: diff
• fortune-street-fog on ogl-lin-radeon: diff
• f-zero-rain on ogl-lin-radeon: diff
• mini-ninjas on ogl-lin-radeon: diff
• mkdd-efb on ogl-lin-radeon: diff
• nddemo-lighting on ogl-lin-radeon: diff
• pbr-sfx on ogl-lin-radeon: diff
• rs2-skybox on ogl-lin-radeon: diff
• sadx-ui on ogl-lin-radeon: diff
• sfa-shadows on ogl-lin-radeon: diff
• sf-assault-flashing on ogl-lin-radeon: diff
• soa-black on ogl-lin-radeon: diff
• sonic-riders-blur on ogl-lin-radeon: diff
• sonic-riders-zg-4p on ogl-lin-radeon: diff
• sonicriderszg-gb on ogl-lin-radeon: diff
• spyro-bloom on ogl-lin-radeon: diff
• spyro-depth on ogl-lin-radeon: diff
• ssbb-mod-lloyd on ogl-lin-radeon: diff
• sw3-dt on ogl-lin-radeon: diff
• tla-menu on ogl-lin-radeon: diff
• tp-skin on ogl-lin-radeon: diff
• tsp3-pinkgrass on ogl-lin-radeon: diff
• vegas-party-depth on ogl-lin-radeon: diff
• xenoblade-menu on ogl-lin-radeon: diff
• ztp-grass on ogl-lin-radeon: diff
• zww-armos on ogl-lin-radeon: diff
• zww-water on ogl-lin-radeon: diff
• zww-waves on ogl-lin-radeon: diff

_{^{automated-fifoci-reporter}}

[Minty-Meeo]

And that is why I made a backup

[Minty-Meeo]

Do not be alarmed by the force push. I decided to remake this goliath of a PR to both further split it into commits and also because rebasing proved difficult. I have a backup of the original changes on branch bfview-video-copy which I will be referencing throughout the process.

[Minty-Meeo]

I found conflicting info on whether a TexInputForm field is supposed to be two bits or one bit wide. I think the info in XFMemory.h strongly suggests the latter.

[Minty-Meeo]

I am thinking the cc and ac 24-bit values in the PixelShader UID's stagehashes might be better off as C bit fields. I really want to make them into C23's BitInts or some kind of native int24_t, and a C bit field is the next best thing.

[Minty-Meeo]

I've wanted to make the usage of the component field cleaner, but I think the code might end up less efficient. Before, it would be right shift by constant (since it is zero, this is a no-op), mask by constant, mask by constant, compare against 0 (two operations, after the two masks get fused). Now it is right shift by constant, mask by constant, compare against 0 (three operations).

[Minty-Meeo]

Well I thought I was clever in optimizing Common::ExtractBit to use the same idiom as seen in VertexShaderGen, only to see the compiler do this: https://godbolt.org/z/evjr8hrrx.