Best Use & Engineering Application Guide

Engineering Category: Exact Integer Number Theory & Transcendentals

Best Use in Production: Hardware-verified modular arithmetic, integer roots, and combinatorial invariants for exact rational math engines and cryptographic primitives.

When to Select This Capsule: Sole canonical implementation for its 64-vector functional equivalence class. Low-latency L1-resident kernel (6471/500 cyc/op); well-suited for high-frequency pipeline stages.

Snap-Circuit Wiring Role: 1-to-1 Unary Transform Stage `(i64)->i64` — snaps directly into linear pipelines or fan-out DAG branches.

Hardware _rdtsc: 6471/500 cycles/op
Latency: 404479/100000 ns/op
Throughput: 100000000/404479 ops/us
Compiled ELF64 .text: 478 B (154 x86_64 insns)
C64 64KB RAM Budget: 117/16384 (468 B source)
Ternary Vote: + (score 1) | Salience: 30130/351
Pure-Rust Capsule Source Code (468 B)
pub fn fx_sqrt_q16(value: i64) -> i64 {
    // sqrt of a Q16.16 raw value, staying in Q16.16: sqrt(v * 2^16) * 2^16.
    // Guarded so any i64 lands in range: negatives and the shift-overflow
    // headroom both clamp, and the result is exact integer isqrt -- no float.
    const FRAC: u32 = 16;
    if value <= 0 {
        return 0;
    }
    let headroom = i64::MAX >> FRAC;
    let raw = if value > headroom { headroom } else { value };
    (raw << FRAC).isqrt()
}
rustc -C opt-level=3 x86_64 Disassembly (478 B .text / 154 insns)
fx_sqrt_q16:
    test    rdi, rdi
    jle    .LBB7_1
    movabs    r8, 140737488355327
    cmp    rdi, r8
    cmovb    r8, rdi
    shl    r8, 16
    cmp    rdi, 65536
    jae    .LBB7_3
    bsr    esi, r8d
    xor    esi, 31
    mov    ecx, esi
    and    cl, 30
    shl    r8d, cl
    mov    eax, r8d
    shr    eax, 24
    lea    rdx, [rip + .Lanon.a270e05f789020653a98a620bea8a67e.3]
    movzx    ecx, byte ptr [rdx + 2*rax]
    movzx    edx, byte ptr [rdx + 2*rax + 1]
    mov    edi, ecx
    add    edi, ecx
    shl    edx, 4
    mov    eax, r8d
    shr    eax, 20
    and    eax, 15
    or    eax, edx
    xor    edx, edx
    div    di
    mov    edi, r8d
    shr    edi, 16
    xor    r9d, r9d
    shl    cl, 4
    movzx    ecx, cl
    add    ecx, eax
    shl    edx, 4
    and    edi, 15
    or    edi, edx
    imul    eax, eax
    sub    di, ax
    lea    eax, [rcx + rcx - 1]
    cmovae    eax, r9d
    sbb    cx, 0
    add    eax, edi
    movzx    edi, cx
    add    edi, edi
    movzx    eax, ax
    shl    eax, 8
    xor    edx, edx
    div    edi
    shl    ecx, 8
    movzx    edi, cx
    add    edi, eax
    mov    eax, edi
    mul    edi
    seto    cl
    cmp    eax, r8d
    seta    al
    or    al, cl
    movzx    eax, al
    sub    edi, eax
    shr    esi
    mov    ecx, esi
    shr    edi, cl
    mov    rax, rdi
    ret
.LBB7_1:
    xor    edi, edi
    mov    rax, rdi
    ret
.LBB7_3:
    bsr    rsi, r8
    xor    rsi, 63
    mov    ecx, esi
    and    cl, 62
    shl    r8, cl
    mov    rax, r8
    shr    rax, 56
    lea    rcx, [rip + .Lanon.a270e05f789020653a98a620bea8a67e.3]
    movzx    edi, byte ptr [rcx + 2*rax]
    movzx    ecx, byte ptr [rcx + 2*rax + 1]
    mov    r9d, edi
    add    r9d, edi
    shl    ecx, 4
    mov    rax, r8
    shr    rax, 52
    and    eax, 15
    or    eax, ecx
    xor    edx, edx
    div    r9w
    mov    r9, r8
    shr    r9, 48
    xor    ecx, ecx
    shl    dil, 4
    movzx    edi, dil
    add    edi, eax
    shl    edx, 4
    and    r9d, 15
    or    r9d, edx
    imul    eax, eax
    sub    r9w, ax
    lea    eax, [rdi + rdi - 1]
    cmovae    eax, ecx
    sbb    di, 0
    add    eax, r9d
    movzx    r9d, di
    add    r9d, r9d
    movzx    edx, ax
    shl    edx, 8
    mov    rax, r8
    shr    rax, 40
    movzx    eax, al
    or    eax, edx
    xor    edx, edx
    div    r9d
    mov    r9, r8
    shr    r9, 32
    shl    edi, 8
    movzx    edi, di
    add    edi, eax
    shl    edx, 8
    movzx    r9d, r9b
    or    r9d, edx
    imul    eax, eax
    sub    r9d, eax
    lea    eax, [rdi + rdi - 1]
    cmovae    eax, ecx
    lea    edx, [r9 + rax]
    sbb    edi, 0
    lea    rcx, [rdi + rdi]
    shl    rdx, 16
    mov    eax, r8d
    shr    eax, 16
    or    rax, rdx
    or    rdx, rcx
    shr    rdx, 32
    test    dx, dx
    je    .LBB7_4
    xor    edx, edx
    div    rcx
    jmp    .LBB7_6
.LBB7_4:
    xor    edx, edx
    div    ecx
.LBB7_6:
    shl    edi, 16
    add    rdi, rax
    mov    rax, rdi
    mul    rdi
    seto    cl
    cmp    rax, r8
    seta    al
    or    al, cl
    movzx    eax, al
    sub    rdi, rax
    shr    esi
    mov    ecx, esi
    shr    rdi, cl
    mov    rax, rdi
    ret

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

Global RankCapsule & DomainHardware _rdtscLatencyThroughputELF .textTernary VoteAuthor Aye
#630fx_sqrt_q16 (THIS CAPSULE)
domain: fixed-point | id: 46fe68ba0b3218784ddec6a6647bf96d
6471/500 cyc/op404479/100000 ns/op100000000/404479 ops/us478 B .text (154 insns)+ (1)aye::Peter+Omni(0+1)::aarch64 @ 432Hz