Merge branch 'next'

Conflicts:
	host/lib/usrp/usrp2/codec_impl.cpp

1 files changed, 61 insertions, 0 deletions

diff --git a/host/lib/convert/convert_with_neon.cpp b/host/lib/convert/convert_with_neon.cpp
new file mode 100644
index 000000000..3d677db5b
--- /dev/null
+++ b/host/lib/convert/convert_with_neon.cpp
@@ -0,0 +1,61 @@
+//
+// Copyright 2011-2011 Ettus Research LLC
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU General Public License for more details.
+//
+// You should have received a copy of the GNU General Public License
+// along with this program.  If not, see <http://www.gnu.org/licenses/>.
+//
+
+#include "convert_common.hpp"
+#include <arm_neon.h>
+
+using namespace uhd::convert;
+
+DECLARE_CONVERTER(convert_fc32_1_to_item32_1_nswap, PRIORITY_CUSTOM){
+    const fc32_t *input = reinterpret_cast<const fc32_t *>(inputs[0]);
+    item32_t *output = reinterpret_cast<item32_t *>(outputs[0]);
+
+    size_t i;
+
+    float32x4_t Q0 = vdupq_n_f32(shorts_per_float);
+    for (i=0; i < (nsamps & ~0x03); i+=2) {
+        float32x4_t Q1 = vld1q_f32(reinterpret_cast<const float *>(&input[i]));
+        float32x4_t Q2 = vmulq_f32(Q1, Q0);
+        int32x4_t Q3 = vcvtq_s32_f32(Q2);
+        int16x4_t D8 = vmovn_s32(Q3);
+        int16x4_t D9 = vrev32_s16(D8);
+        vst1_s16((reinterpret_cast<int16_t *>(&output[i])), D9);
+    }
+
+    for (; i < nsamps; i++)
+        output[i] = fc32_to_item32(input[i]);
+}
+
+DECLARE_CONVERTER(convert_item32_1_to_fc32_1_nswap, PRIORITY_CUSTOM){
+    const item32_t *input = reinterpret_cast<const item32_t *>(inputs[0]);
+    fc32_t *output = reinterpret_cast<fc32_t *>(outputs[0]);
+
+    size_t i;
+
+    float32x4_t Q1 = vdupq_n_f32(floats_per_short);
+    for (i=0; i < (nsamps & ~0x03); i+=2) {
+        int16x4_t D0 = vld1_s16(reinterpret_cast<const int16_t *>(&input[i]));
+        int16x4_t D1 = vrev32_s16(D0);
+        int32x4_t Q2 = vmovl_s16(D1);
+        float32x4_t Q3 = vcvtq_f32_s32(Q2);
+        float32x4_t Q4 = vmulq_f32(Q3, Q1);
+        vst1q_f32((reinterpret_cast<float *>(&output[i])), Q4);
+    }
+
+    for (; i < nsamps; i++)
+        output[i] = item32_to_fc32(input[i]);
+}