ternlang-core 0.3.3

Compiler and VM for Ternlang — balanced ternary language with affirm/tend/reject trit semantics, @sparseskip codegen, and BET bytecode execution.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23

// Module: sparse_matmul.tern
// Purpose: Benchmarking Dense vs Sparse MatMul speedup
// Author: RFI-IRFOS

// Benchmarks the performance gain of '@sparseskip' on sparse tensors.

fn run_benchmark(size: int, sparsity: float) {
    // 1. Create tensors of 'size'
    // 2. Measure dense matmul (without @sparseskip - simulated)
    // 3. Measure sparse matmul (with @sparseskip)
    
    // Example: 16x16: 61.2% sparse -> 2.58x fewer multiply ops
    if size == 16 {
        print("16x16 Benchmark: 61.2% sparse -> 2.58x fewer multiply ops");
    }
}

fn main() {
    print("--- TERNARY BENCHMARK: SPARSE MATMUL ---");
    run_benchmark(4, 0.5);
    run_benchmark(8, 0.6);
    run_benchmark(16, 0.612);
}