Best Use & Engineering Application Guide

Engineering Category: Balanced Ternary & b158 Quantization

Best Use in Production: Packing and unpacking balanced ternary (-1, 0, +1) activation lanes into dense 64-bit registers for C64-constrained inference pipelines.

When to Select This Capsule: Sole canonical implementation for its 64-vector functional equivalence class. Low-latency L1-resident kernel (3767491/200000 cyc/op); well-suited for high-frequency pipeline stages.

Snap-Circuit Wiring Role: Specialized Multi-Pin Stage `([i8;5])->u8` — wire matching input pins via `snap_dag` or `snap_pipeline`.

Hardware _rdtsc: 3767491/200000 cycles/op
Latency: 2044161/400000 ns/op
Throughput: 400000000/2044161 ops/us
Compiled ELF64 .text: 156 B (50 x86_64 insns)
C64 64KB RAM Budget: 379/65536 (379 B source)
Ternary Vote: + (score 1) | Salience: 36045/416
Pure-Rust Capsule Source Code (379 B)
pub const fn pack_5_trits_base243(trits: [i8; 5]) -> u8 {
    let mut acc: u16 = 0;
    let mut pow3: u16 = 1;
    let mut i = 0;
    while i < 5 {
        let digit: u16 = if trits[i] < 0 {
            0
        } else if trits[i] > 0 {
            2
        } else {
            1
        };
        acc += digit * pow3;
        pow3 *= 3;
        i += 1;
    }
    acc as u8
}
rustc -C opt-level=3 x86_64 Disassembly (156 B .text / 50 insns)
pack_5_trits_base243:
    mov    eax, edi
    shr    eax, 16
    test    al, al
    setne    cl
    mov    sil, 9
    shl    sil, cl
    xor    edx, edx
    test    al, al
    movzx    eax, sil
    cmovs    eax, edx
    mov    esi, edi
    shr    esi, 8
    test    sil, sil
    setne    cl
    mov    r8b, 3
    shl    r8b, cl
    test    sil, sil
    movzx    esi, r8b
    cmovs    esi, edx
    test    dil, dil
    setne    cl
    mov    r8b, 1
    shl    r8b, cl
    test    dil, dil
    movzx    ecx, r8b
    cmovs    ecx, edx
    mov    r8d, edi
    shr    r8d, 24
    add    sil, cl
    test    r8d, r8d
    setne    cl
    mov    r9b, 27
    shl    r9b, cl
    test    r8b, r8b
    movzx    r8d, r9b
    cmovs    r8d, edx
    add    r8b, al
    add    r8b, sil
    mov    rax, rdi
    shr    rax, 32
    shl    rax, 56
    setne    cl
    mov    al, 81
    shl    al, cl
    movabs    rcx, 549755813888
    test    rdi, rcx
    movzx    eax, al
    cmovne    eax, edx
    add    al, r8b
    ret

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

Global RankCapsule & DomainHardware _rdtscLatencyThroughputELF .textTernary VoteAuthor Aye
#948pack_5_trits_base243 (THIS CAPSULE)
domain: ayecast-x86_64 | id: a32f837266521437c9a5ea6b9e1a0890
3767491/200000 cyc/op2044161/400000 ns/op400000000/2044161 ops/us156 B .text (50 insns)+ (1)aye::Hue+Gemini(0+1)::x86_64 @ 432Hz