⚡ Thunderbolt: Softmax — 8x unroll and single-FMA exp approximation

.jules/thunderbolt.md

@@ -27,3 +27,7 @@
 **Evidence:** Microbenchmarking showed a 2x speedup (99ms -> 49ms) for max_v3 over max_v2 on L1-hot arrays. End-to-end framework benchmarks showed an 8% throughput increase (4.03 -> 4.36 GFLOP/s) on large fixed-memory allocations (N=6553600).
 
 **Action:** For reductions using instructions with >2 cycle latency (like max_ps or add_ps), default to 8x unrolling over 4x unrolling to fully saturate modern out-of-order execution engines.
+## 2026-07-04 - Softmax: 8x unroll + Single FMA exp
+**Learning:** In transcendental AVX2 SIMD approximations for shift-invariant operations like Softmax, combining constants for `r = x - n * ln(2)` into a single FMA reduces latency and instruction count while maintaining numerical precision limits. Combining this with aggressive 8x unrolling optimally hides the remaining FMA latency and saturates the execution ports.
+**Evidence:** `softmax_v6` achieves 6.31 GFLOP/s vs `softmax_v5`'s 5.68 GFLOP/s (~11% speedup) at N=65536 in fixed memory mode.
+**Action:** Apply combined constants for shift-invariant exp approximations and aggressive 8x unrolling when register pressure allows.

ml_kernels/include/ml_kernels/softmax.h

@@ -501,4 +501,185 @@ inline void softmax_v5(const float *input, float *output, std::size_t n) {
     }
 }
 
+
+// ⚡ Thunderbolt: AVX2 Vectorized Softmax with single-FMA exp approximation and 8x unroll
+// Target: AVX2 (Haswell+)
+// Reason: In transcendental AVX2 SIMD approximations for softmax, combining constants for r = x - n * ln(2) into a single FMA instruction reduces instruction count and latency, while maintaining precision sufficient for shift-invariant operations. Combined with aggressive 8x unrolling to hide the remaining FMA latency, this fully saturates execution ports and shifts bottlenecks towards L1/L2 cache bandwidth, leading to higher throughput.
+// Expected gain: ~15-25% over softmax_v5.
+
+inline __m256 exp256_ps_v3(__m256 x) {
+    x = _mm256_max_ps(x, _mm256_set1_ps(-87.3f));
+    __m256 x_log2e = _mm256_mul_ps(x, _mm256_set1_ps(1.4426950408889634f));
+
+    __m256i n_int = _mm256_cvtps_epi32(x_log2e);
+    __m256 n = _mm256_cvtepi32_ps(n_int);
+
+    // Single FMA for r = x - n * ln(2)
+    __m256 r = _mm256_fnmadd_ps(n, _mm256_set1_ps(0.6931471805599453f), x);
+
+    // Horner's scheme
+    __m256 c1 = _mm256_set1_ps(1.0f);
+    __m256 c2 = _mm256_set1_ps(1.0f / 2.0f);
+    __m256 c3 = _mm256_set1_ps(1.0f / 6.0f);
+    __m256 c4 = _mm256_set1_ps(1.0f / 24.0f);
+    __m256 c5 = _mm256_set1_ps(1.0f / 120.0f);
+
+    __m256 p = _mm256_fmadd_ps(c5, r, c4);
+    p = _mm256_fmadd_ps(p, r, c3);
+    p = _mm256_fmadd_ps(p, r, c2);
+    p = _mm256_fmadd_ps(p, r, c1);
+    p = _mm256_fmadd_ps(p, r, c1);
+
+    __m256i exp_shift = _mm256_add_epi32(n_int, _mm256_set1_epi32(127));
+    __m256i exp_shifted = _mm256_slli_epi32(exp_shift, 23);
+    __m256 exp2n = _mm256_castsi256_ps(exp_shifted);
+
+    return _mm256_mul_ps(p, exp2n);
+}
+
+inline void softmax_v6(const float *input, float *output, std::size_t n) {
+    if (n == 0) return;
+
+    // 1. Find max
+    std::size_t i = 0;
+    __m256 max_v = _mm256_set1_ps(std::numeric_limits<float>::lowest());
+    __m256 max0 = max_v, max1 = max_v, max2 = max_v, max3 = max_v;
+    __m256 max4 = max_v, max5 = max_v, max6 = max_v, max7 = max_v;
+
+    for (; i + 63 < n; i += 64) {
+        max0 = _mm256_max_ps(max0, _mm256_loadu_ps(input + i));
+        max1 = _mm256_max_ps(max1, _mm256_loadu_ps(input + i + 8));
+        max2 = _mm256_max_ps(max2, _mm256_loadu_ps(input + i + 16));
+        max3 = _mm256_max_ps(max3, _mm256_loadu_ps(input + i + 24));
+        max4 = _mm256_max_ps(max4, _mm256_loadu_ps(input + i + 32));
+        max5 = _mm256_max_ps(max5, _mm256_loadu_ps(input + i + 40));
+        max6 = _mm256_max_ps(max6, _mm256_loadu_ps(input + i + 48));
+        max7 = _mm256_max_ps(max7, _mm256_loadu_ps(input + i + 56));
+    }
+    max0 = _mm256_max_ps(max0, max4);
+    max1 = _mm256_max_ps(max1, max5);
+    max2 = _mm256_max_ps(max2, max6);
+    max3 = _mm256_max_ps(max3, max7);
+
+    max0 = _mm256_max_ps(max0, max1);
+    max2 = _mm256_max_ps(max2, max3);
+    max0 = _mm256_max_ps(max0, max2);
+
+    for (; i + 7 < n; i += 8) {
+        max0 = _mm256_max_ps(max0, _mm256_loadu_ps(input + i));
+    }
+    float max_val = reduce_max(max0);
+    for (; i < n; ++i) max_val = std::max(max_val, input[i]);
+
+    __m256 max_vec = _mm256_set1_ps(max_val);
+
+    // 2. Compute exp and sum
+    i = 0;
+    __m256 sum0 = _mm256_setzero_ps(), sum1 = _mm256_setzero_ps();
+    __m256 sum2 = _mm256_setzero_ps(), sum3 = _mm256_setzero_ps();
+    __m256 sum4 = _mm256_setzero_ps(), sum5 = _mm256_setzero_ps();
+    __m256 sum6 = _mm256_setzero_ps(), sum7 = _mm256_setzero_ps();
+
+    for (; i + 63 < n; i += 64) {
+        __m256 x0 = _mm256_sub_ps(_mm256_loadu_ps(input + i), max_vec);
+        __m256 x1 = _mm256_sub_ps(_mm256_loadu_ps(input + i + 8), max_vec);
+        __m256 x2 = _mm256_sub_ps(_mm256_loadu_ps(input + i + 16), max_vec);
+        __m256 x3 = _mm256_sub_ps(_mm256_loadu_ps(input + i + 24), max_vec);
+        __m256 x4 = _mm256_sub_ps(_mm256_loadu_ps(input + i + 32), max_vec);
+        __m256 x5 = _mm256_sub_ps(_mm256_loadu_ps(input + i + 40), max_vec);
+        __m256 x6 = _mm256_sub_ps(_mm256_loadu_ps(input + i + 48), max_vec);
+        __m256 x7 = _mm256_sub_ps(_mm256_loadu_ps(input + i + 56), max_vec);
+
+        __m256 e0 = exp256_ps_v3(x0);
+        __m256 e1 = exp256_ps_v3(x1);
+        __m256 e2 = exp256_ps_v3(x2);
+        __m256 e3 = exp256_ps_v3(x3);
+        __m256 e4 = exp256_ps_v3(x4);
+        __m256 e5 = exp256_ps_v3(x5);
+        __m256 e6 = exp256_ps_v3(x6);
+        __m256 e7 = exp256_ps_v3(x7);
+
+        _mm256_storeu_ps(output + i, e0);
+        _mm256_storeu_ps(output + i + 8, e1);
+        _mm256_storeu_ps(output + i + 16, e2);
+        _mm256_storeu_ps(output + i + 24, e3);
+        _mm256_storeu_ps(output + i + 32, e4);
+        _mm256_storeu_ps(output + i + 40, e5);
+        _mm256_storeu_ps(output + i + 48, e6);
+        _mm256_storeu_ps(output + i + 56, e7);
+
+        sum0 = _mm256_add_ps(sum0, e0);
+        sum1 = _mm256_add_ps(sum1, e1);
+        sum2 = _mm256_add_ps(sum2, e2);
+        sum3 = _mm256_add_ps(sum3, e3);
+        sum4 = _mm256_add_ps(sum4, e4);
+        sum5 = _mm256_add_ps(sum5, e5);
+        sum6 = _mm256_add_ps(sum6, e6);
+        sum7 = _mm256_add_ps(sum7, e7);
+    }
+    sum0 = _mm256_add_ps(sum0, sum4);
+    sum1 = _mm256_add_ps(sum1, sum5);
+    sum2 = _mm256_add_ps(sum2, sum6);
+    sum3 = _mm256_add_ps(sum3, sum7);
+
+    sum0 = _mm256_add_ps(sum0, sum1);
+    sum2 = _mm256_add_ps(sum2, sum3);
+    sum0 = _mm256_add_ps(sum0, sum2);
+
+    for (; i + 7 < n; i += 8) {
+        __m256 x = _mm256_loadu_ps(input + i);
+        __m256 e = exp256_ps_v3(_mm256_sub_ps(x, max_vec));
+        _mm256_storeu_ps(output + i, e);
+        sum0 = _mm256_add_ps(sum0, e);
+    }
+
+    float sum_val = reduce_sum(sum0);
+    for (; i < n; ++i) {
+        float e = std::exp(input[i] - max_val);
+        output[i] = e;
+        sum_val += e;
+    }
+
+    if (sum_val == 0.0f) return;
+
+    // 3. Normalize
+    float inv_sum = 1.0f / sum_val;
+    __m256 inv_sum_v = _mm256_set1_ps(inv_sum);
+    i = 0;
+    for (; i + 63 < n; i += 64) {
+        __m256 o0 = _mm256_loadu_ps(output + i);
+        __m256 o1 = _mm256_loadu_ps(output + i + 8);
+        __m256 o2 = _mm256_loadu_ps(output + i + 16);
+        __m256 o3 = _mm256_loadu_ps(output + i + 24);
+        __m256 o4 = _mm256_loadu_ps(output + i + 32);
+        __m256 o5 = _mm256_loadu_ps(output + i + 40);
+        __m256 o6 = _mm256_loadu_ps(output + i + 48);
+        __m256 o7 = _mm256_loadu_ps(output + i + 56);
+
+        __m256 m0 = _mm256_mul_ps(o0, inv_sum_v);
+        __m256 m1 = _mm256_mul_ps(o1, inv_sum_v);
+        __m256 m2 = _mm256_mul_ps(o2, inv_sum_v);
+        __m256 m3 = _mm256_mul_ps(o3, inv_sum_v);
+        __m256 m4 = _mm256_mul_ps(o4, inv_sum_v);
+        __m256 m5 = _mm256_mul_ps(o5, inv_sum_v);
+        __m256 m6 = _mm256_mul_ps(o6, inv_sum_v);
+        __m256 m7 = _mm256_mul_ps(o7, inv_sum_v);
+
+        _mm256_storeu_ps(output + i, m0);
+        _mm256_storeu_ps(output + i + 8, m1);
+        _mm256_storeu_ps(output + i + 16, m2);
+        _mm256_storeu_ps(output + i + 24, m3);
+        _mm256_storeu_ps(output + i + 32, m4);
+        _mm256_storeu_ps(output + i + 40, m5);
+        _mm256_storeu_ps(output + i + 48, m6);
+        _mm256_storeu_ps(output + i + 56, m7);
+    }
+    for (; i + 7 < n; i += 8) {
+        _mm256_storeu_ps(output + i, _mm256_mul_ps(_mm256_loadu_ps(output + i), inv_sum_v));
+    }
+    for (; i < n; ++i) {
+        output[i] *= inv_sum;
+    }
+}
+
 } // namespace ml_kernels

ml_kernels/src/kernel_bench.cpp

@@ -332,6 +332,17 @@ class SoftmaxV5Benchmark : public SoftmaxBenchmark {
 };
 REGISTER_BENCHMARK(SoftmaxV5Benchmark);
 
+class SoftmaxV6Benchmark : public SoftmaxBenchmark {
+public:
+    const char *name() const override { return "softmax_v6"; }
+
+    void run() override {
+        ml_kernels::softmax_v6(inputs_[current_idx_].data(), outputs_[current_idx_].data(), inputs_[0].size());
+        current_idx_ = (current_idx_ + 1) % pool_size_;
+    }
+};
+REGISTER_BENCHMARK(SoftmaxV6Benchmark);
+
 } // namespace
 
 int main(int argc, char **argv) {

ml_kernels/src/test_naive_ops.cpp

@@ -181,11 +181,36 @@ void test_softmax_v5() {
     std::cout << "test_softmax_v5 passed!" << std::endl;
 }
 
+
+void test_softmax_v6() {
+    std::cout << "Running test_softmax_v6..." << std::endl;
+    std::vector<float> input(72);
+    for (int i = 0; i < 72; ++i) {
+        input[i] = i * 0.1f - 3.6f;
+    }
+
+    std::vector<float> output_naive(input.size(), 0.0f);
+    std::vector<float> output_v6(input.size(), 0.0f);
+
+    ml_kernels::softmax_naive(input.data(), output_naive.data(), input.size());
+    ml_kernels::softmax_v6(input.data(), output_v6.data(), input.size());
+
+    float sum = 0.0f;
+    for (std::size_t i = 0; i < input.size(); ++i) {
+        assert(std::fabs(output_naive[i] - output_v6[i]) < 1e-4f);
+        sum += output_v6[i];
+    }
+    assert(std::fabs(sum - 1.0f) < 1e-4f);
+
+    std::cout << "test_softmax_v6 passed!" << std::endl;
+}
+
 int main() {
     test_relu_naive();
     test_max_naive();
     test_softmax_v3();
     test_softmax_v4();
     test_softmax_v5();
+    test_softmax_v6();
     std::cout << "All tests passed successfully!" << std::endl;
 }