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

Use VectMergeLow and VecMergeHigh

pull/795/head
Jeffrey Walton 2019-01-21 03:04:07 -05:00
parent 767d7e2a13
commit bbad4397c4
No known key found for this signature in database
GPG Key ID: B36AB348921B1838
2 changed files with 21 additions and 21 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

34
gf2n_simd.cpp
View File

@ -318,8 +318,8 @@ using CryptoPP::VecXor;
using CryptoPP::VecAnd; using CryptoPP::VecAnd;
using CryptoPP::VecPermute; using CryptoPP::VecPermute;
using CryptoPP::VecMergeHi; using CryptoPP::VecMergeHigh;
using CryptoPP::VecMergeLo; using CryptoPP::VecMergeLow;
using CryptoPP::VecShiftLeft; using CryptoPP::VecShiftLeft;
using CryptoPP::VecShiftRight; using CryptoPP::VecShiftRight;
using CryptoPP::VecRotateLeftOctet; using CryptoPP::VecRotateLeftOctet;
@ -336,16 +336,16 @@ F2N_Multiply_128x128_POWER8(uint64x2_p& c1, uint64x2_p& c0, const uint64x2_p& a,
c0 = VecPolyMultiply00LE(a, b); c0 = VecPolyMultiply00LE(a, b);
c1 = VecPolyMultiply11LE(a, b); c1 = VecPolyMultiply11LE(a, b);
t1 = VecMergeLo(a, a); t1 = VecMergeLow(a, a);
t1 = VecXor(a, t1); t1 = VecXor(a, t1);
t2 = VecMergeLo(b, b); t2 = VecMergeLow(b, b);
t2 = VecXor(b, t2); t2 = VecXor(b, t2);
t1 = VecPolyMultiply00LE(t1, t2); t1 = VecPolyMultiply00LE(t1, t2);
t1 = VecXor(c0, t1); t1 = VecXor(c0, t1);
t1 = VecXor(c1, t1); t1 = VecXor(c1, t1);
t2 = t1; t2 = t1;
t1 = VecMergeHi(z0, t1); t1 = VecMergeHigh(z0, t1);
t2 = VecMergeLo(t2, z0); t2 = VecMergeLow(t2, z0);
c0 = VecXor(c0, t1); c0 = VecXor(c0, t1);
c1 = VecXor(c1, t2); c1 = VecXor(c1, t2);
} }
@ -394,7 +394,7 @@ inline uint64x2_p ShiftLeft128_POWER8(uint64x2_p x)
x = VecShiftLeft<N>(x); x = VecShiftLeft<N>(x);
u = VecShiftRight<64-N>(u); u = VecShiftRight<64-N>(u);
v = VecMergeHi(z, u); v = VecMergeHigh(z, u);
x = VecOr(x, v); x = VecOr(x, v);
return x; return x;
} }
@ -411,41 +411,41 @@ GF2NT_233_Reduce_POWER8(uint64x2_p& c3, uint64x2_p& c2, uint64x2_p& c1, uint64x2
const uint64x2_p z0={0}; const uint64x2_p z0={0};
b1 = c1; a1 = c1; b1 = c1; a1 = c1;
a0 = VecMergeHi(c1, z0); a0 = VecMergeHigh(c1, z0);
a1 = VecShiftLeft<23>(a1); a1 = VecShiftLeft<23>(a1);
a1 = VecShiftRight<23>(a1); a1 = VecShiftRight<23>(a1);
c1 = VecOr(a1, a0); c1 = VecOr(a1, a0);
b2 = VecShiftRight<64-23>(c2); b2 = VecShiftRight<64-23>(c2);
c3 = ShiftLeft128_POWER8<23>(c3); c3 = ShiftLeft128_POWER8<23>(c3);
a0 = VecMergeLo(b2, z0); a0 = VecMergeLow(b2, z0);
c3 = VecOr(c3, a0); c3 = VecOr(c3, a0);
b1 = VecShiftRight<64-23>(b1); b1 = VecShiftRight<64-23>(b1);
c2 = ShiftLeft128_POWER8<23>(c2); c2 = ShiftLeft128_POWER8<23>(c2);
a0 = VecMergeLo(b1, z0); a0 = VecMergeLow(b1, z0);
c2 = VecOr(c2, a0); c2 = VecOr(c2, a0);
b3 = c3; b3 = c3;
b2 = VecShiftRight<64-10>(c2); b2 = VecShiftRight<64-10>(c2);
b3 = ShiftLeft128_POWER8<10>(b3); b3 = ShiftLeft128_POWER8<10>(b3);
a0 = VecMergeLo(b2, z0); a0 = VecMergeLow(b2, z0);
b3 = VecOr(b3, a0); b3 = VecOr(b3, a0);
a0 = VecMergeLo(c3, z0); a0 = VecMergeLow(c3, z0);
b3 = VecXor(b3, a0); b3 = VecXor(b3, a0);
b1 = VecShiftRight<64-23>(b3); b1 = VecShiftRight<64-23>(b3);
b3 = ShiftLeft128_POWER8<23>(b3); b3 = ShiftLeft128_POWER8<23>(b3);
b3 = VecMergeLo(b3, z0); b3 = VecMergeLow(b3, z0);
b3 = VecOr(b3, b1); b3 = VecOr(b3, b1);
c2 = VecXor(c2, b3); c2 = VecXor(c2, b3);
b3 = c3; b3 = c3;
b2 = VecShiftRight<64-10>(c2); b2 = VecShiftRight<64-10>(c2);
b3 = ShiftLeft128_POWER8<10>(b3); b3 = ShiftLeft128_POWER8<10>(b3);
b2 = VecMergeLo(b2, z0); b2 = VecMergeLow(b2, z0);
b3 = VecOr(b3, b2); b3 = VecOr(b3, b2);
b2 = c2; b2 = c2;
b2 = ShiftLeft128_POWER8<10>(b2); b2 = ShiftLeft128_POWER8<10>(b2);
a0 = VecMergeHi(z0, b2); a0 = VecMergeHigh(z0, b2);
c2 = VecXor(c2, a0); c2 = VecXor(c2, a0);
a0 = VecMergeHi(z0, b3); a0 = VecMergeHigh(z0, b3);
a1 = VecMergeLo(b2, z0); a1 = VecMergeLow(b2, z0);
a0 = VecOr(a0, a1); a0 = VecOr(a0, a1);
c3 = VecXor(c3, a0); c3 = VecXor(c3, a0);
c0 = VecXor(c0, c2); c0 = VecXor(c0, c2);

8
ppc_simd.h
View File

@ -1214,7 +1214,7 @@ inline uint32x4_p VecShiftLeft(const uint32x4_p vec)
/// vec_mergeh /// vec_mergeh
/// \since Crypto++ 8.1 /// \since Crypto++ 8.1
template <class T> template <class T>
inline T VecMergeHi(const T vec1, const T vec2) inline T VecMergeHigh(const T vec1, const T vec2)
{ {
return vec_mergeh(vec1, vec2); return vec_mergeh(vec1, vec2);
} }
@ -1228,7 +1228,7 @@ inline T VecMergeHi(const T vec1, const T vec2)
/// vec_mergel /// vec_mergel
/// \since Crypto++ 8.1 /// \since Crypto++ 8.1
template <class T> template <class T>
inline T VecMergeLo(const T vec1, const T vec2) inline T VecMergeLow(const T vec1, const T vec2)
{ {
return vec_mergel(vec1, vec2); return vec_mergel(vec1, vec2);
} }
@ -1364,7 +1364,7 @@ inline T VecGetLow(const T val)
{ {
#if (CRYPTOPP_BIG_ENDIAN) && (_ARCH_PWR8) #if (CRYPTOPP_BIG_ENDIAN) && (_ARCH_PWR8)
const T zero = {0}; const T zero = {0};
return (T)VecMergeLo((uint64x2_p)zero, (uint64x2_p)val); return (T)VecMergeLow((uint64x2_p)zero, (uint64x2_p)val);
#else #else
return VecShiftRightOctet<8>(VecShiftLeftOctet<8>(val)); return VecShiftRightOctet<8>(VecShiftLeftOctet<8>(val));
#endif #endif
@ -1386,7 +1386,7 @@ inline T VecGetHigh(const T val)
{ {
#if (CRYPTOPP_BIG_ENDIAN) && (_ARCH_PWR8) #if (CRYPTOPP_BIG_ENDIAN) && (_ARCH_PWR8)
const T zero = {0}; const T zero = {0};
return (T)VecMergeHi((uint64x2_p)zero, (uint64x2_p)val); return (T)VecMergeHigh((uint64x2_p)zero, (uint64x2_p)val);
#else #else
return VecShiftRightOctet<8>(val); return VecShiftRightOctet<8>(val);
#endif #endif