From 07a46c66535ce0f50c2d4706335d2093868a1ee9 Mon Sep 17 00:00:00 2001
From: Anthony Blake <anthonix@me.com>
Date: Tue, 6 Nov 2012 10:42:47 +1300
Subject: 2D has SIMD transpose (still needs work though)

---
 src/ffts_nd.c | 36 ++++++++++++++++++++++++++++++++----
 src/ffts_nd.h |  6 +++++-
 2 files changed, 37 insertions(+), 5 deletions(-)

(limited to 'src')

diff --git a/src/ffts_nd.c b/src/ffts_nd.c
index 4e72e81..086b98c 100644
--- a/src/ffts_nd.c
+++ b/src/ffts_nd.c
@@ -48,13 +48,41 @@ void ffts_free_nd(ffts_plan_t *p) {
 	free(p);
 }
 
-void ffts_transpose(uint64_t *in, uint64_t *out, int w, int h) {
+inline void ffts_transpose(uint64_t *in, uint64_t *out, int w, int h) {
 
 	size_t i,j;
 
-	for(i=0;i<w;i++) {
-		for(j=0;j<h;j++) {
-			out[i*h + j] = in[j*w + i];
+	for(i=0;i<w;i+=2) {
+		for(j=0;j<h;j+=2) {
+#ifdef __ARM_NEON__
+//			out[i*h + j] = in[j*w + i];
+			float32x4_t Q0 = vld1q_f32(in + j*w + i);
+			float32x4_t Q1 = vld1q_f32(in + j*w + i + w);
+
+			float32x2x2_t t0;
+			float32x2x2_t t1;
+			t0.val[0] = vget_low_f32(Q0);
+			t0.val[1] = vget_high_f32(Q0);
+			t1.val[0] = vget_low_f32(Q1);
+			t1.val[1] = vget_high_f32(Q1);
+
+			__asm__ ("vswp %0,%1\n\t"
+					: "+w" (t0.val[1]), "+w" (t1.val[0])
+					:
+					);
+
+			Q0 = vcombine_f32(t0.val[0], t0.val[1]);
+			Q1 = vcombine_f32(t1.val[0], t1.val[1]);
+			vst1q_f32(out + i*h + j, Q0);
+			vst1q_f32(out + i*h + j + h, Q1);
+#else
+			__m128d q0 = _mm_load_pd(in + j*w + i);
+			__m128d q1 = _mm_load_pd(in + j*w + i + w);
+			__m128d t0 = _mm_shuffle_pd(q0, q1, _MM_SHUFFLE2(0, 0));
+			__m128d t1 = _mm_shuffle_pd(q0, q1, _MM_SHUFFLE2(1, 1));
+			_mm_store_pd(out + i*h + j, t0);
+			_mm_store_pd(out + i*h + j + h, t1);
+#endif
 		}
 	}
 }
diff --git a/src/ffts_nd.h b/src/ffts_nd.h
index 2eef287..4d2474d 100644
--- a/src/ffts_nd.h
+++ b/src/ffts_nd.h
@@ -40,7 +40,11 @@
 
 #include "ffts.h"
 
-
+#ifdef __ARM_NEON__
+	#include <arm_neon.h>
+#else
+	#include <xmmintrin.h>
+#endif
 
 
 #endif
-- 
cgit v1.1