arne
/
cryptopp
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Update comments

pull/676/head
Jeffrey Walton 2018-06-23 04:49:10 -04:00
parent 9980d30734
commit d0b5dac162
No known key found for this signature in database
GPG Key ID: B36AB348921B1838
1 changed files with 24 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

28
lea-simd.cpp
View File

@ -154,7 +154,12 @@ inline uint32x4_t UnpackNEON(const uint32x4_t& a, const uint32x4_t& b, const uin
template <> template <>
inline uint32x4_t UnpackNEON<0>(const uint32x4_t& a, const uint32x4_t& b, const uint32x4_t& c, const uint32x4_t& d) inline uint32x4_t UnpackNEON<0>(const uint32x4_t& a, const uint32x4_t& b, const uint32x4_t& c, const uint32x4_t& d)
{ {
// LEA is little-endian oriented, so there is no need for a separate shuffle. //uint32x4_t r(a);
//r = vsetq_lane_u32(vgetq_lane_u32(b, 0), r, 1);
//r = vsetq_lane_u32(vgetq_lane_u32(c, 0), r, 2);
//r = vsetq_lane_u32(vgetq_lane_u32(d, 0), r, 3);
//return r;
const uint32x4_t r1 = UnpackLow32(a, b); const uint32x4_t r1 = UnpackLow32(a, b);
const uint32x4_t r2 = UnpackLow32(c, d); const uint32x4_t r2 = UnpackLow32(c, d);
return UnpackLow64(r1, r2); return UnpackLow64(r1, r2);
@ -163,7 +168,12 @@ inline uint32x4_t UnpackNEON<0>(const uint32x4_t& a, const uint32x4_t& b, const
template <> template <>
inline uint32x4_t UnpackNEON<1>(const uint32x4_t& a, const uint32x4_t& b, const uint32x4_t& c, const uint32x4_t& d) inline uint32x4_t UnpackNEON<1>(const uint32x4_t& a, const uint32x4_t& b, const uint32x4_t& c, const uint32x4_t& d)
{ {
// LEA is little-endian oriented, so there is no need for a separate shuffle. //uint32x4_t r(a);
//r = vsetq_lane_u32(vgetq_lane_u32(b, 1), r, 1);
//r = vsetq_lane_u32(vgetq_lane_u32(c, 1), r, 2);
//r = vsetq_lane_u32(vgetq_lane_u32(d, 1), r, 3);
//return r;
const uint32x4_t r1 = UnpackLow32(a, b); const uint32x4_t r1 = UnpackLow32(a, b);
const uint32x4_t r2 = UnpackLow32(c, d); const uint32x4_t r2 = UnpackLow32(c, d);
return UnpackHigh64(r1, r2); return UnpackHigh64(r1, r2);
@ -172,7 +182,12 @@ inline uint32x4_t UnpackNEON<1>(const uint32x4_t& a, const uint32x4_t& b, const
template <> template <>
inline uint32x4_t UnpackNEON<2>(const uint32x4_t& a, const uint32x4_t& b, const uint32x4_t& c, const uint32x4_t& d) inline uint32x4_t UnpackNEON<2>(const uint32x4_t& a, const uint32x4_t& b, const uint32x4_t& c, const uint32x4_t& d)
{ {
// LEA is little-endian oriented, so there is no need for a separate shuffle. //uint32x4_t r(a);
//r = vsetq_lane_u32(vgetq_lane_u32(b, 2), r, 1);
//r = vsetq_lane_u32(vgetq_lane_u32(c, 2), r, 2);
//r = vsetq_lane_u32(vgetq_lane_u32(d, 2), r, 3);
//return r;
const uint32x4_t r1 = UnpackHigh32(a, b); const uint32x4_t r1 = UnpackHigh32(a, b);
const uint32x4_t r2 = UnpackHigh32(c, d); const uint32x4_t r2 = UnpackHigh32(c, d);
return UnpackLow64(r1, r2); return UnpackLow64(r1, r2);
@ -181,7 +196,12 @@ inline uint32x4_t UnpackNEON<2>(const uint32x4_t& a, const uint32x4_t& b, const
template <> template <>
inline uint32x4_t UnpackNEON<3>(const uint32x4_t& a, const uint32x4_t& b, const uint32x4_t& c, const uint32x4_t& d) inline uint32x4_t UnpackNEON<3>(const uint32x4_t& a, const uint32x4_t& b, const uint32x4_t& c, const uint32x4_t& d)
{ {
// LEA is little-endian oriented, so there is no need for a separate shuffle. //uint32x4_t r(a);
//r = vsetq_lane_u32(vgetq_lane_u32(b, 3), r, 1);
//r = vsetq_lane_u32(vgetq_lane_u32(c, 3), r, 2);
//r = vsetq_lane_u32(vgetq_lane_u32(d, 3), r, 3);
//return r;
const uint32x4_t r1 = UnpackHigh32(a, b); const uint32x4_t r1 = UnpackHigh32(a, b);
const uint32x4_t r2 = UnpackHigh32(c, d); const uint32x4_t r2 = UnpackHigh32(c, d);
return UnpackHigh64(r1, r2); return UnpackHigh64(r1, r2);