Best Use & Engineering Application Guide

Engineering Category: SWAR Pixel Processing & Vision Kernels

Best Use in Production: Branchless in-register pixel averaging, color-space conversion, and edge gradient filtering for zero-copy framebuffers.

When to Select This Capsule: Sole canonical implementation for its 64-vector functional equivalence class. Low-latency L1-resident kernel (1766211/100000 cyc/op); well-suited for high-frequency pipeline stages.

Snap-Circuit Wiring Role: Specialized Multi-Pin Stage `(u16,u8,u8)->u16` — wire matching input pins via `snap_dag` or `snap_pipeline`.

Hardware _rdtsc: 1766211/100000 cycles/op
Latency: 3364483/400000 ns/op
Throughput: 400000000/3364483 ops/us
Compiled ELF64 .text: 218 B (73 x86_64 insns)
C64 64KB RAM Budget: 709/65536 (709 B source)
Ternary Vote: + (score 1) | Salience: 5083/56
Pure-Rust Capsule Source Code (709 B)
pub fn hsv_to_rgb565(hue: u16, sat: u8, val: u8) -> u16 {
    let hue = hue % 360;
    let region = (hue / 60) as u8;
    let remainder = ((hue - (region as u16 * 60)) * 255 / 60) as u8;

    let p = ((val as u16 * (255 - sat as u16)) >> 8) as u8;
    let q = ((val as u16 * (255 - ((sat as u16 * remainder as u16) >> 8))) >> 8) as u8;
    let t = ((val as u16 * (255 - ((sat as u16 * (255 - remainder as u16)) >> 8))) >> 8) as u8;

    let (r, g, b) = match region {
        0 => (val, t, p),
        1 => (q, val, p),
        2 => (p, val, t),
        3 => (p, q, val),
        4 => (t, p, val),
        _ => (val, p, q),
    };

    ((r as u16 & 0xF8) << 8) | ((g as u16 & 0xFC) << 3) | ((b as u16) >> 3)
}
rustc -C opt-level=3 x86_64 Disassembly (218 B .text / 73 insns)
hsv_to_rgb565:
    push    rbx
    movzx    eax, di
    imul    eax, eax, 11651
    shr    eax, 22
    imul    eax, eax, 360
    sub    edi, eax
    movzx    eax, di
    imul    r10d, eax, 1093
    shr    r10d, 16
    imul    eax, r10d, -60
    add    eax, edi
    mov    edi, eax
    shl    edi, 8
    sub    edi, eax
    movzx    eax, di
    imul    r9d, eax, 34953
    shr    r9d, 21
    movzx    edi, dl
    movzx    r8d, sil
    mov    eax, r8d
    xor    eax, 255
    imul    eax, edi
    shr    eax, 8
    movzx    ecx, r9b
    mov    ebx, ecx
    imul    ebx, r8d
    not    ebx
    movzx    esi, bh
    imul    esi, edi
    shr    esi, 8
    cmp    r10w, 4
    ja    .LBB7_1
    xor    ecx, 255
    imul    ecx, r8d
    not    ecx
    movzx    ebx, ch
    imul    ebx, edi
    shr    ebx, 8
    lea    rcx, [rip + .LJTI7_0]
    movsxd    rdi, dword ptr [rcx + 4*r10]
    add    rdi, rcx
    jmp    rdi
.LBB7_3:
    mov    ecx, ebx
    mov    edi, edx
    jmp    .LBB7_9
.LBB7_7:
    mov    ecx, eax
    mov    edi, ebx
    jmp    .LBB7_8
.LBB7_5:
    mov    ecx, edx
    mov    edi, eax
    mov    eax, ebx
    jmp    .LBB7_9
.LBB7_6:
    mov    ecx, esi
    mov    edi, eax
.LBB7_8:
    mov    eax, edx
    jmp    .LBB7_9
.LBB7_4:
    mov    ecx, edx
    mov    edi, esi
    jmp    .LBB7_9
.LBB7_1:
    mov    ecx, eax
    mov    edi, edx
    mov    eax, esi
.LBB7_9:
    and    dil, -8
    movzx    edx, dil
    shl    edx, 8
    and    cl, -4
    movzx    ecx, cl
    lea    ecx, [rdx + 8*rcx]
    shr    al, 3
    movzx    eax, al
    or    eax, ecx
    pop    rbx
    ret

Head-to-Head Functional Equivalence Class Peers (1 Implementation(s) Matching 64 Test Vectors)

Global RankCapsule & DomainHardware _rdtscLatencyThroughputELF .textTernary VoteAuthor Aye
#1672hsv_to_rgb565 (THIS CAPSULE)
domain: FLUG-OS-x86_64 | id: c02ca0dbff297a1977024bff3260754e
1766211/100000 cyc/op3364483/400000 ns/op400000000/3364483 ops/us218 B .text (73 insns)+ (1)aye::Hue+Gemini(0+1)::x86_64 @ 432Hz