C2 GEMM/IGEMM QS8/QC8 NEON microkernels with FP32 requantization
PiperOrigin-RevId: 382429758
diff --git a/src/qc8-igemm/gen/1x8c2-minmax-fp32-neon-mlal-padal-dup.c b/src/qc8-igemm/gen/1x8c2-minmax-fp32-neon-mlal-padal-dup.c
new file mode 100644
index 0000000..9d7e04f
--- /dev/null
+++ b/src/qc8-igemm/gen/1x8c2-minmax-fp32-neon-mlal-padal-dup.c
@@ -0,0 +1,233 @@
+// Auto-generated file. Do not edit!
+// Template: src/qs8-igemm/c2-neon-mull-padal-dup.c.in
+// Generator: tools/xngen
+//
+// Copyright 2021 Google LLC
+//
+// This source code is licensed under the BSD-style license found in the
+// LICENSE file in the root directory of this source tree.
+
+#include <assert.h>
+
+#include <arm_neon.h>
+
+#include <xnnpack/gemm.h>
+#include <xnnpack/math.h>
+
+
+void xnn_qc8_igemm_minmax_fp32_ukernel_1x8c2__neon_mlal_padal_dup(
+ size_t mr,
+ size_t nc,
+ size_t kc,
+ size_t ks,
+ const int8_t** restrict a,
+ const void* restrict w,
+ int8_t* restrict c,
+ size_t cm_stride,
+ size_t cn_stride,
+ size_t a_offset,
+ const int8_t* zero,
+ const union xnn_qs8_minmax_params params[restrict XNN_MIN_ELEMENTS(1)]) XNN_DISABLE_TSAN XNN_DISABLE_MSAN
+{
+ assert(mr != 0);
+ assert(mr <= 1);
+ assert(nc != 0);
+ assert(kc != 0);
+ assert(ks != 0);
+ assert(ks % (1 * sizeof(void*)) == 0);
+ assert(a_offset % sizeof(int8_t) == 0);
+ assert(a != NULL);
+ assert(w != NULL);
+ assert(c != NULL);
+
+ kc = round_up_po2(kc, 2 * sizeof(int8_t));
+ int8_t* c0 = c;
+
+ do {
+ int32x4_t vacc0x0123 = vld1q_s32(w); w = (const void*) ((uintptr_t) w + 4 * sizeof(int32_t));
+ int32x4_t vacc0x4567 = vld1q_s32(w); w = (const void*) ((uintptr_t) w + 4 * sizeof(int32_t));
+
+ size_t p = ks;
+ do {
+ const int8_t* restrict a0 = a[0];
+ if XNN_UNPREDICTABLE(a0 != zero) {
+ a0 = (const int8_t*) ((uintptr_t) a0 + a_offset);
+ }
+ a += 1;
+
+ size_t k = kc;
+
+ while (k >= 16 * sizeof(int8_t)) {
+ const int8x8_t va0x0 = vld1_s8(a0); a0 += 8;
+ const int8x8_t va0x1 = vld1_s8(a0); a0 += 8;
+
+ const int8x8_t vb0123c0x0 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+ const int8x8_t vb4567c0x0 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+ const int8x8_t vb0123c1x0 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+ const int8x8_t vb4567c1x0 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+ const int8x8_t vb0123c2x0 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+ const int8x8_t vb4567c2x0 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+ const int8x8_t vb0123c3x0 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+ const int8x8_t vb4567c3x0 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+
+ int16x8_t vprod0x0123c0 = vmull_s8(vb0123c0x0, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0x0), 0)));
+ const int8x8_t vb0123c0x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+ vprod0x0123c0 = vmlal_s8(vprod0x0123c0, vb0123c0x1, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0x1), 0)));
+ vacc0x0123 = vpadalq_s16(vacc0x0123, vprod0x0123c0);
+ int16x8_t vprod0x4567c0 = vmull_s8(vb4567c0x0, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0x0), 0)));
+ const int8x8_t vb4567c0x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+ vprod0x4567c0 = vmlal_s8(vprod0x4567c0, vb4567c0x1, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0x1), 0)));
+ vacc0x4567 = vpadalq_s16(vacc0x4567, vprod0x4567c0);
+ int16x8_t vprod0x0123c1 = vmull_s8(vb0123c1x0, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0x0), 1)));
+ const int8x8_t vb0123c1x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+ vprod0x0123c1 = vmlal_s8(vprod0x0123c1, vb0123c1x1, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0x1), 1)));
+ vacc0x0123 = vpadalq_s16(vacc0x0123, vprod0x0123c1);
+ int16x8_t vprod0x4567c1 = vmull_s8(vb4567c1x0, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0x0), 1)));
+ const int8x8_t vb4567c1x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+ vprod0x4567c1 = vmlal_s8(vprod0x4567c1, vb4567c1x1, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0x1), 1)));
+ vacc0x4567 = vpadalq_s16(vacc0x4567, vprod0x4567c1);
+ int16x8_t vprod0x0123c2 = vmull_s8(vb0123c2x0, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0x0), 2)));
+ const int8x8_t vb0123c2x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+ vprod0x0123c2 = vmlal_s8(vprod0x0123c2, vb0123c2x1, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0x1), 2)));
+ vacc0x0123 = vpadalq_s16(vacc0x0123, vprod0x0123c2);
+ int16x8_t vprod0x4567c2 = vmull_s8(vb4567c2x0, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0x0), 2)));
+ const int8x8_t vb4567c2x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+ vprod0x4567c2 = vmlal_s8(vprod0x4567c2, vb4567c2x1, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0x1), 2)));
+ vacc0x4567 = vpadalq_s16(vacc0x4567, vprod0x4567c2);
+ int16x8_t vprod0x0123c3 = vmull_s8(vb0123c3x0, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0x0), 3)));
+ const int8x8_t vb0123c3x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+ vprod0x0123c3 = vmlal_s8(vprod0x0123c3, vb0123c3x1, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0x1), 3)));
+ vacc0x0123 = vpadalq_s16(vacc0x0123, vprod0x0123c3);
+ int16x8_t vprod0x4567c3 = vmull_s8(vb4567c3x0, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0x0), 3)));
+ const int8x8_t vb4567c3x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+ vprod0x4567c3 = vmlal_s8(vprod0x4567c3, vb4567c3x1, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0x1), 3)));
+ vacc0x4567 = vpadalq_s16(vacc0x4567, vprod0x4567c3);
+
+ k -= 16 * sizeof(int8_t);
+ }
+
+ if (k >= 8 * sizeof(int8_t)) {
+ const int8x8_t va0 = vld1_s8(a0); a0 += 8;
+
+ const int8x8_t vb0123c0 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+ const int8x8_t vb4567c0 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+ const int8x8_t vb0123c1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+ const int8x8_t vb4567c1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+ const int8x8_t vb0123c2 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+ const int8x8_t vb4567c2 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+ const int8x8_t vb0123c3 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+ const int8x8_t vb4567c3 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+
+ const int16x8_t vprod0x0123c0 = vmull_s8(vb0123c0, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0), 0)));
+ const int16x8_t vprod0x0123c1 = vmull_s8(vb0123c1, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0), 1)));
+ const int16x8_t vprod0x0123c2 = vmull_s8(vb0123c2, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0), 2)));
+ const int16x8_t vprod0x0123c3 = vmull_s8(vb0123c3, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0), 3)));
+ vacc0x0123 = vpadalq_s16(vacc0x0123, vprod0x0123c0);
+ vacc0x0123 = vpadalq_s16(vacc0x0123, vprod0x0123c1);
+ vacc0x0123 = vpadalq_s16(vacc0x0123, vprod0x0123c2);
+ vacc0x0123 = vpadalq_s16(vacc0x0123, vprod0x0123c3);
+ const int16x8_t vprod0x4567c0 = vmull_s8(vb4567c0, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0), 0)));
+ const int16x8_t vprod0x4567c1 = vmull_s8(vb4567c1, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0), 1)));
+ const int16x8_t vprod0x4567c2 = vmull_s8(vb4567c2, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0), 2)));
+ const int16x8_t vprod0x4567c3 = vmull_s8(vb4567c3, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0), 3)));
+ vacc0x4567 = vpadalq_s16(vacc0x4567, vprod0x4567c0);
+ vacc0x4567 = vpadalq_s16(vacc0x4567, vprod0x4567c1);
+ vacc0x4567 = vpadalq_s16(vacc0x4567, vprod0x4567c2);
+ vacc0x4567 = vpadalq_s16(vacc0x4567, vprod0x4567c3);
+
+ k -= 8 * sizeof(int8_t);
+ }
+
+ if XNN_UNLIKELY(k != 0) {
+ const int8x8_t va0 = vld1_s8(a0); a0 = (const int8_t*) ((uintptr_t) a0 + k);
+
+ const int8x8_t vb0123c0 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+ const int8x8_t vb4567c0 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+
+ const int16x8_t vprod0x0123c0 = vmull_s8(vb0123c0, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0), 0)));
+ vacc0x0123 = vpadalq_s16(vacc0x0123, vprod0x0123c0);
+ const int16x8_t vprod0x4567c0 = vmull_s8(vb4567c0, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0), 0)));
+ vacc0x4567 = vpadalq_s16(vacc0x4567, vprod0x4567c0);
+
+ if (k > 2 * sizeof(int8_t)) {
+ const int8x8_t vb0123c1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+ const int8x8_t vb4567c1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+
+ const int16x8_t vprod0x0123c1 = vmull_s8(vb0123c1, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0), 1)));
+ vacc0x0123 = vpadalq_s16(vacc0x0123, vprod0x0123c1);
+ const int16x8_t vprod0x4567c1 = vmull_s8(vb4567c1, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0), 1)));
+ vacc0x4567 = vpadalq_s16(vacc0x4567, vprod0x4567c1);
+
+ if (k > 4 * sizeof(int8_t)) {
+ const int8x8_t vb0123c2 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+ const int8x8_t vb4567c2 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof(int8_t));
+
+ const int16x8_t vprod0x0123c2 = vmull_s8(vb0123c2, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0), 2)));
+ vacc0x0123 = vpadalq_s16(vacc0x0123, vprod0x0123c2);
+ const int16x8_t vprod0x4567c2 = vmull_s8(vb4567c2, vreinterpret_s8_s16(vdup_lane_s16(vreinterpret_s16_s8(va0), 2)));
+ vacc0x4567 = vpadalq_s16(vacc0x4567, vprod0x4567c2);
+ }
+ }
+ }
+ p -= 1 * sizeof(void*);
+ } while (p != 0);
+
+ float32x4_t vfpacc0x0123 = vcvtq_f32_s32(vacc0x0123);
+ float32x4_t vfpacc0x4567 = vcvtq_f32_s32(vacc0x4567);
+
+ const float32x4_t vscale0123 = vld1q_f32((const float*) w); w = (const void*) ((const float*) w + 4);
+ vfpacc0x0123 = vmulq_f32(vfpacc0x0123, vscale0123);
+ const float32x4_t vscale4567 = vld1q_f32((const float*) w); w = (const void*) ((const float*) w + 4);
+ vfpacc0x4567 = vmulq_f32(vfpacc0x4567, vscale4567);
+
+ const float32x4_t voutput_min_less_zero_point = vld1q_dup_f32(¶ms->neon_fp32.output_min_less_zero_point);
+ vfpacc0x0123 = vmaxq_f32(vfpacc0x0123, voutput_min_less_zero_point);
+ vfpacc0x4567 = vmaxq_f32(vfpacc0x4567, voutput_min_less_zero_point);
+
+ const float32x4_t voutput_max_less_zero_point = vld1q_dup_f32(¶ms->neon_fp32.output_max_less_zero_point);
+ vfpacc0x0123 = vminq_f32(vfpacc0x0123, voutput_max_less_zero_point);
+ vfpacc0x4567 = vminq_f32(vfpacc0x4567, voutput_max_less_zero_point);
+
+ const float32x4_t vmagic_bias = vld1q_dup_f32(¶ms->neon_fp32.magic_bias);
+ vacc0x0123 = vreinterpretq_s32_f32(vaddq_f32(vfpacc0x0123, vmagic_bias));
+ vacc0x4567 = vreinterpretq_s32_f32(vaddq_f32(vfpacc0x4567, vmagic_bias));
+
+ const int32x4_t vmagic_bias_less_zero_point = vld1q_dup_s32(¶ms->neon_fp32.magic_bias_less_zero_point);
+ vacc0x0123 = vsubq_s32(vacc0x0123, vmagic_bias_less_zero_point);
+ vacc0x4567 = vsubq_s32(vacc0x4567, vmagic_bias_less_zero_point);
+
+#if XNN_ARCH_ARM64
+ const int16x8_t vacc0x01234567 = vuzp1q_s16(vreinterpretq_s16_s32(vacc0x0123), vreinterpretq_s16_s32(vacc0x4567));
+
+ int8x8_t vout0x01234567 = vmovn_s16(vacc0x01234567);
+#else
+ const int16x8_t vacc0x01234567 = vcombine_s16(vmovn_s32(vacc0x0123), vmovn_s32(vacc0x4567));
+
+ int8x8_t vout0x01234567 = vmovn_s16(vacc0x01234567);
+#endif
+
+ if (nc >= 8) {
+ vst1_s8(c0 + 0, vout0x01234567);
+
+ c0 = (int8_t*) ((uintptr_t) c0 + cn_stride);
+
+ a = (const int8_t**restrict) ((uintptr_t) a - ks);
+
+ nc -= 8;
+ } else {
+ if (nc & 4) {
+ vst1_lane_u32(__builtin_assume_aligned(c0, 1), vreinterpret_u32_s8(vout0x01234567), 0); c0 += 4;
+ vout0x01234567 = vext_s8(vout0x01234567, vout0x01234567, 4);
+ }
+ if (nc & 2) {
+ vst1_lane_u16(__builtin_assume_aligned(c0, 1), vreinterpret_u16_s8(vout0x01234567), 0); c0 += 2;
+ vout0x01234567 = vext_s8(vout0x01234567, vout0x01234567, 2);
+ }
+ if (nc & 1) {
+ vst1_lane_s8(c0, vout0x01234567, 0);
+ }
+
+ nc = 0;
+ }
+ } while (nc != 0);
+}