resampler: add floating point neon optimization

Change-Id: I0684ea002b2b02be5dae01a6c00ffbdf28b0e265

1 files changed, 89 insertions, 1 deletions

diff --git a/libspeex/resample_neon.h b/libspeex/resample_neon.h
index 784ee2f..8a97429 100644
--- a/libspeex/resample_neon.h
+++ b/libspeex/resample_neon.h
@@ -38,7 +38,18 @@
 #include <arm_neon.h>
 
 #ifdef FIXED_POINT
-
+static inline int32_t saturate_32bit_to_16bit(int32_t a) {
+    int32_t ret;
+    asm volatile ("vmov.s32 d24[0], %[a]\n"
+                  "vqmovn.s32 d24, q12\n"
+                  "vmov.s16 %[ret], d24[0]\n"
+                  : [ret] "=&r" (ret)
+                  : [a] "r" (a)
+                  : "q12", "d24", "d25" );
+    return ret;
+}
+#undef WORD2INT
+#define WORD2INT(x) (saturate_32bit_to_16bit(x))
 
 #define OVERRIDE_INNER_PRODUCT_SINGLE
 static inline int32_t inner_product_single(const int16_t *a, const int16_t *b, unsigned int len)
@@ -97,4 +108,81 @@ static inline int32_t inner_product_single(const int16_t *a, const int16_t *b, u
     return ret;
 }
 
+#elif defined(FLOATING_POINT)
+
+static inline int32_t saturate_float_to_16bit(float a) {
+    int32_t ret;
+    asm ("vmov.f32 d24[0], %[a]\n"
+         "vcvt.s32.f32 d24, d24, #15\n"
+         "vqrshrn.s32 d24, q12, #15\n"
+         "vmov.s16 %[ret], d24[0]\n"
+         : [ret] "=&r" (ret)
+         : [a] "r" (a)
+         : "q12", "d24", "d25" );
+    return ret;
+}
+#undef WORD2INT
+#define WORD2INT(x) (saturate_float_to_16bit(x))
+
+#define OVERRIDE_INNER_PRODUCT_SINGLE
+static inline float inner_product_single(const float *a, const float *b, unsigned int len)
+{
+    float ret;
+    uint32_t remainder = len % 16;
+    len = len - remainder;
+
+    asm volatile ("	 cmp %[len], #0\n"
+		  "	 bne 1f\n"
+		  "	 vld1.32 {q4}, [%[a]]!\n"
+		  "	 vld1.32 {q8}, [%[b]]!\n"
+		  "	 subs %[remainder], %[remainder], #4\n"
+		  "	 vmul.f32 q0, q4, q8\n"
+		  "      beq 5f\n"
+		  "	 b 4f\n"
+		  "1:"
+		  "	 vld1.32 {q4, q5}, [%[a]]!\n"
+		  "	 vld1.32 {q8, q9}, [%[b]]!\n"
+		  "	 vld1.32 {q6, q7}, [%[a]]!\n"
+		  "	 vld1.32 {q10, q11}, [%[b]]!\n"
+		  "	 subs %[len], %[len], #16\n"
+		  "	 vmul.f32 q0, q4, q8\n"
+		  "	 vmul.f32 q1, q5, q9\n"
+		  "	 vmul.f32 q2, q6, q10\n"
+		  "	 vmul.f32 q3, q7, q11\n"
+		  "	 beq 3f\n"
+		  "2:"
+		  "	 vld1.32 {q4, q5}, [%[a]]!\n"
+		  "	 vld1.32 {q8, q9}, [%[b]]!\n"
+		  "	 vld1.32 {q6, q7}, [%[a]]!\n"
+		  "	 vld1.32 {q10, q11}, [%[b]]!\n"
+		  "	 subs %[len], %[len], #16\n"
+		  "	 vmla.f32 q0, q4, q8\n"
+		  "	 vmla.f32 q1, q5, q9\n"
+		  "	 vmla.f32 q2, q6, q10\n"
+		  "	 vmla.f32 q3, q7, q11\n"
+		  "	 bne 2b\n"
+		  "3:"
+		  "	 vadd.f32 q4, q0, q1\n"
+		  "	 vadd.f32 q5, q2, q3\n"
+		  "	 vadd.f32 q0, q4, q5\n"
+		  "	 cmp %[remainder], #0\n"
+		  "	 beq 5f\n"
+		  "4:"
+		  "	 vld1.32 {q6}, [%[a]]!\n"
+		  "	 vld1.32 {q10}, [%[b]]!\n"
+		  "	 subs %[remainder], %[remainder], #4\n"
+		  "	 vmla.f32 q0, q6, q10\n"
+		  "	 bne 4b\n"
+		  "5:"
+		  "	 vadd.f32 d0, d0, d1\n"
+		  "	 vpadd.f32 d0, d0, d0\n"
+		  "	 vmov.f32 %[ret], d0[0]\n"
+		  : [ret] "=&r" (ret), [a] "+r" (a), [b] "+r" (b),
+		    [len] "+l" (len), [remainder] "+l" (remainder)
+		  :
+		  : "cc", "q0", "q1", "q2", "q3", "q4", "q5", "q6", "q7", "q8",
+		    "q10", "q11");
+    return ret;
+}
+
 #endif