avcodec/ac3: Implement sum_square_butterfly_float for aarch64 NEON

Signed-off-by: Geoff Hill <geoff@geoffhill.org>
Signed-off-by: Martin Storsjö <martin@martin.st>

libavcodec/aarch64/ac3dsp_init_aarch64.c

@@ -32,6 +32,10 @@ void ff_ac3_sum_square_butterfly_int32_neon(int64_t sum[4],
                                             const int32_t *coef0,
                                             const int32_t *coef1,
                                             int len);
+void ff_ac3_sum_square_butterfly_float_neon(float sum[4],
+                                            const float *coef0,
+                                            const float *coef1,
+                                            int len);
 
 av_cold void ff_ac3dsp_init_aarch64(AC3DSPContext *c)
 {
@@ -42,4 +46,5 @@ av_cold void ff_ac3dsp_init_aarch64(AC3DSPContext *c)
     c->extract_exponents = ff_ac3_extract_exponents_neon;
     c->float_to_fixed24 = ff_float_to_fixed24_neon;
     c->sum_square_butterfly_int32 = ff_ac3_sum_square_butterfly_int32_neon;
+    c->sum_square_butterfly_float = ff_ac3_sum_square_butterfly_float_neon;
 }

libavcodec/aarch64/ac3dsp_neon.S

@@ -87,3 +87,33 @@ function ff_ac3_sum_square_butterfly_int32_neon, export=1
         st1             {v0.1d-v3.1d}, [x0]
         ret
 endfunc
+
+function ff_ac3_sum_square_butterfly_float_neon, export=1
+        movi            v0.4s, #0
+        movi            v1.4s, #0
+        movi            v2.4s, #0
+        movi            v3.4s, #0
+1:      ld1             {v30.4s}, [x1], #16
+        ld1             {v31.4s}, [x2], #16
+        fadd            v16.4s, v30.4s, v31.4s
+        fsub            v17.4s, v30.4s, v31.4s
+        fmla            v0.4s, v30.4s, v30.4s
+        fmla            v1.4s, v31.4s, v31.4s
+        fmla            v2.4s, v16.4s, v16.4s
+        fmla            v3.4s, v17.4s, v17.4s
+        subs            w3, w3, #4
+        b.gt            1b
+        faddp           v0.4s, v0.4s, v0.4s
+        faddp           v0.2s, v0.2s, v0.2s
+        st1             {v0.s}[0], [x0], #4
+        faddp           v1.4s, v1.4s, v1.4s
+        faddp           v1.2s, v1.2s, v1.2s
+        st1             {v1.s}[0], [x0], #4
+        faddp           v2.4s, v2.4s, v2.4s
+        faddp           v2.2s, v2.2s, v2.2s
+        st1             {v2.s}[0], [x0], #4
+        faddp           v3.4s, v3.4s, v3.4s
+        faddp           v3.2s, v3.2s, v3.2s
+        st1             {v3.s}[0], [x0]
+        ret
+endfunc

tests/checkasm/ac3dsp.c

@@ -165,6 +165,31 @@ static void check_ac3_sum_square_butterfly_int32(AC3DSPContext *c) {
     report("ac3_sum_square_butterfly_int32");
 }
 
+static void check_ac3_sum_square_butterfly_float(AC3DSPContext *c) {
+    LOCAL_ALIGNED_32(float, lt, [ELEMS]);
+    LOCAL_ALIGNED_32(float, rt, [ELEMS]);
+    LOCAL_ALIGNED_16(float, v1, [4]);
+    LOCAL_ALIGNED_16(float, v2, [4]);
+
+    declare_func(void, float[4], const float *, const float *, int);
+
+    randomize_float(lt, ELEMS);
+    randomize_float(rt, ELEMS);
+
+    if (check_func(c->sum_square_butterfly_float,
+                   "ac3_sum_square_bufferfly_float")) {
+        call_ref(v1, lt, rt, ELEMS);
+        call_new(v2, lt, rt, ELEMS);
+
+        if (!float_near_ulp_array(v1, v2, 10, 4))
+            fail();
+
+        bench_new(v2, lt, rt, ELEMS);
+    }
+
+    report("ac3_sum_square_butterfly_float");
+}
+
 void checkasm_check_ac3dsp(void)
 {
     AC3DSPContext c;
@@ -174,4 +199,5 @@ void checkasm_check_ac3dsp(void)
     check_ac3_extract_exponents(&c);
     check_float_to_fixed24(&c);
     check_ac3_sum_square_butterfly_int32(&c);
+    check_ac3_sum_square_butterfly_float(&c);
 }