benchFloat4()

template<typename Op>

int64_t autoresearch::simd_check::benchFloat4 ( int iters, Op op )

inline

Definition at line 1134 of file AutoResearchSimd.h.

                                             {
    float a0 = 1.5f, a1 = 2.3f, a2 = 0.7f, a3 = 3.1f;
    float b0 = 0.5f, b1 = 1.2f, b2 = 2.0f, b3 = 0.9f;
    uint32_t t0 = micros();
    for (int i = 0; i < iters; i++) {
        a0 = op(a0, b0); a1 = op(a1, b1);
        a2 = op(a2, b2); a3 = op(a3, b3);
        b0 = a0 + 0.001f; b1 = a1 + 0.001f;
        b2 = a2 + 0.001f; b3 = a3 + 0.001f;
    }
    uint32_t t1 = micros();
    uint32_t tmp; fl::memcpy(&tmp, &a0, sizeof(tmp));
    simdBenchmarkState().sink = tmp;
    return static_cast<int64_t>(t1 - t0);
}

References fl::memcpy(), fl::micros(), simdBenchmarkState(), and autoresearch::simd_check::SimdBenchmarkState::sink.

Referenced by runMultiplyBenchmark().

Here is the call graph for this function:

Here is the caller graph for this function: