diff --git a/Eigen/src/Core/GenericPacketMath.h b/Eigen/src/Core/GenericPacketMath.h index 914f16e73..8ace18174 100644 --- a/Eigen/src/Core/GenericPacketMath.h +++ b/Eigen/src/Core/GenericPacketMath.h @@ -167,6 +167,10 @@ ei_pload(const typename ei_unpacket_traits::type* from) { return *from; template inline Packet ei_ploadu(const typename ei_unpacket_traits::type* from) { return *from; } +/** \internal \returns a packet with elements of \a *from duplicated, e.g.: (from[0],from[0],from[1],from[1]) */ +template inline Packet +ei_ploaddup(const typename ei_unpacket_traits::type* from) { return *from; } + /** \internal \returns a packet with constant coefficients \a a, e.g.: (a,a,a,a) */ template inline Packet ei_pset1(const typename ei_unpacket_traits::type& a) { return a; } diff --git a/Eigen/src/Core/arch/SSE/Complex.h b/Eigen/src/Core/arch/SSE/Complex.h index 8184159c7..baa25907c 100644 --- a/Eigen/src/Core/arch/SSE/Complex.h +++ b/Eigen/src/Core/arch/SSE/Complex.h @@ -73,9 +73,12 @@ template<> EIGEN_STRONG_INLINE Packet2cf ei_pmul(const Packet2cf& a, { // TODO optimize it for SSE3 and 4 #ifdef EIGEN_VECTORIZE_SSE3 - return Packet2cf(_mm_addsub_ps(_mm_mul_ps(ei_vec4f_swizzle1(a.v, 0, 0, 2, 2), b.v), - _mm_mul_ps(ei_vec4f_swizzle1(a.v, 1, 1, 3, 3), + return Packet2cf(_mm_addsub_ps(_mm_mul_ps(_mm_moveldup_ps(a.v), b.v), + _mm_mul_ps(_mm_movehdup_ps(a.v), ei_vec4f_swizzle1(b.v, 1, 0, 3, 2)))); +// return Packet2cf(_mm_addsub_ps(_mm_mul_ps(ei_vec4f_swizzle1(a.v, 0, 0, 2, 2), b.v), +// _mm_mul_ps(ei_vec4f_swizzle1(a.v, 1, 1, 3, 3), +// ei_vec4f_swizzle1(b.v, 1, 0, 3, 2)))); #else const __m128 mask = _mm_castsi128_ps(_mm_setr_epi32(0x80000000,0x00000000,0x80000000,0x00000000)); return Packet2cf(_mm_add_ps(_mm_mul_ps(ei_vec4f_swizzle1(a.v, 0, 0, 2, 2), b.v), @@ -224,6 +227,12 @@ template<> EIGEN_STRONG_INLINE Packet2cf ei_pdiv(const Packet2cf& a, return Packet2cf(_mm_div_ps(res.v,_mm_add_ps(s,_mm_castsi128_ps(_mm_shuffle_epi32(_mm_castps_si128(s), 0xb1))))); } +EIGEN_STRONG_INLINE Packet2cf ei_pcplxflip/**/(const Packet2cf& x) +{ + return Packet2cf(ei_vec4f_swizzle1(x.v, 1, 0, 3, 2)); +} + + //---------- double ---------- struct Packet1cd { @@ -268,9 +277,13 @@ template<> EIGEN_STRONG_INLINE Packet1cd ei_pmul(const Packet1cd& a, { // TODO optimize it for SSE3 and 4 #ifdef EIGEN_VECTORIZE_SSE3 - return Packet1cd(_mm_addsub_pd(_mm_mul_pd(ei_vec2d_swizzle1(a.v, 0, 0), b.v), - _mm_mul_pd(ei_vec2d_swizzle1(a.v, 1, 1), - ei_vec2d_swizzle1(b.v, 1, 0)))); +// return Packet1cd(_mm_addsub_pd(_mm_mul_pd(a.v, b.v), +// _mm_mul_pd(a.v, b.v/*ei_vec2d_swizzle1(b.v, 1, 0)*/))); + return Packet1cd(_mm_add_pd(_mm_mul_pd(a.v, b.v), + _mm_mul_pd(a.v, ei_vec2d_swizzle1(b.v, 1, 0)))); +// return Packet1cd(_mm_addsub_pd(_mm_mul_pd(ei_vec2d_swizzle1(a.v, 0, 0), b.v), +// _mm_mul_pd(ei_vec2d_swizzle1(a.v, 1, 1), +// ei_vec2d_swizzle1(b.v, 1, 0)))); #else const __m128d mask = _mm_castsi128_pd(_mm_set_epi32(0x0,0x0,0x80000000,0x0)); return Packet1cd(_mm_add_pd(_mm_mul_pd(ei_vec2d_swizzle1(a.v, 0, 0), b.v), @@ -286,14 +299,14 @@ template<> EIGEN_STRONG_INLINE Packet1cd ei_pandnot(const Packet1cd& // FIXME force unaligned load, this is a temporary fix template<> EIGEN_STRONG_INLINE Packet1cd ei_pload (const std::complex* from) -{ EIGEN_DEBUG_ALIGNED_LOAD return Packet1cd(ei_ploadu((const double*)from)); } +{ EIGEN_DEBUG_ALIGNED_LOAD return Packet1cd(ei_pload((const double*)from)); } template<> EIGEN_STRONG_INLINE Packet1cd ei_ploadu(const std::complex* from) { EIGEN_DEBUG_UNALIGNED_LOAD return Packet1cd(ei_ploadu((const double*)from)); } template<> EIGEN_STRONG_INLINE Packet1cd ei_pset1(const std::complex& from) { /* here we really have to use unaligned loads :( */ return ei_ploadu(&from); } // FIXME force unaligned store, this is a temporary fix -template<> EIGEN_STRONG_INLINE void ei_pstore >(std::complex * to, const Packet1cd& from) { EIGEN_DEBUG_ALIGNED_STORE ei_pstoreu((double*)to, from.v); } +template<> EIGEN_STRONG_INLINE void ei_pstore >(std::complex * to, const Packet1cd& from) { EIGEN_DEBUG_ALIGNED_STORE ei_pstore((double*)to, from.v); } template<> EIGEN_STRONG_INLINE void ei_pstoreu >(std::complex * to, const Packet1cd& from) { EIGEN_DEBUG_UNALIGNED_STORE ei_pstoreu((double*)to, from.v); } template<> EIGEN_STRONG_INLINE void ei_prefetch >(const std::complex * addr) { _mm_prefetch((const char*)(addr), _MM_HINT_T0); } @@ -415,4 +428,9 @@ template<> EIGEN_STRONG_INLINE Packet1cd ei_pdiv(const Packet1cd& a, return Packet1cd(_mm_div_pd(res.v, _mm_add_pd(s,_mm_shuffle_pd(s, s, 0x1)))); } +EIGEN_STRONG_INLINE Packet1cd ei_pcplxflip/**/(const Packet1cd& x) +{ + return Packet1cd(ei_preverse(x.v)); +} + #endif // EIGEN_COMPLEX_SSE_H diff --git a/Eigen/src/Core/arch/SSE/PacketMath.h b/Eigen/src/Core/arch/SSE/PacketMath.h index 53a9bcf56..87ebb783a 100644 --- a/Eigen/src/Core/arch/SSE/PacketMath.h +++ b/Eigen/src/Core/arch/SSE/PacketMath.h @@ -45,7 +45,9 @@ template<> struct ei_is_arithmetic<__m128d> { enum { ret = true }; }; #define ei_vec2d_swizzle1(v,p,q) \ (_mm_castsi128_pd(_mm_shuffle_epi32( _mm_castpd_si128(v), ((q*2+1)<<6|(q*2)<<4|(p*2+1)<<2|(p*2))))) - +// #define ei_vec2d_swizzle1(v,p,q) \ + (_mm_shuffle_pd(v,v, (q)<<1|(p) )) + #define ei_vec4f_swizzle2(a,b,p,q,r,s) \ (_mm_shuffle_ps( (a), (b), ((s)<<6|(r)<<4|(q)<<2|(p)))) @@ -255,6 +257,21 @@ template<> EIGEN_STRONG_INLINE Packet4i ei_ploadu(const int* from) } #endif +template<> EIGEN_STRONG_INLINE Packet4f ei_ploaddup(const float* from) +{ + Packet4f tmp; + tmp = _mm_loadl_pi(tmp,(__m64*)from); + return ei_vec4f_swizzle1(tmp, 0, 0, 1, 1); +} +template<> EIGEN_STRONG_INLINE Packet2d ei_ploaddup(const double* from) +{ return ei_pset1(from[0]); } +template<> EIGEN_STRONG_INLINE Packet4i ei_ploaddup(const int* from) +{ + Packet4i tmp; + tmp = _mm_loadl_epi64(reinterpret_cast(from)); + return ei_vec4i_swizzle1(tmp, 0, 0, 1, 1); +} + template<> EIGEN_STRONG_INLINE void ei_pstore(float* to, const Packet4f& from) { EIGEN_DEBUG_ALIGNED_STORE _mm_store_ps(to, from); } template<> EIGEN_STRONG_INLINE void ei_pstore(double* to, const Packet2d& from) { EIGEN_DEBUG_ALIGNED_STORE _mm_store_pd(to, from); } template<> EIGEN_STRONG_INLINE void ei_pstore(int* to, const Packet4i& from) { EIGEN_DEBUG_ALIGNED_STORE _mm_store_si128(reinterpret_cast(to), from); } diff --git a/Eigen/src/Core/products/GeneralBlockPanelKernel.h b/Eigen/src/Core/products/GeneralBlockPanelKernel.h index ee3e135ef..a5297a31f 100644 --- a/Eigen/src/Core/products/GeneralBlockPanelKernel.h +++ b/Eigen/src/Core/products/GeneralBlockPanelKernel.h @@ -163,23 +163,34 @@ inline void computeProductBlockingSizes(std::ptrdiff_t& k, std::ptrdiff_t& m, st // #define MADD(CJ,A,B,C,T) T = B; T = CJ.pmul(A,T); C = ei_padd(C,T); #endif -/* optimized GEneral packed Block * packed Panel product kernel - * - * Mixing type logic: C += A * B - * | A | B | comments - * |real |cplx | no vectorization yet, would require to pack A with duplication - * |cplx |real | easy vectorization +/* Vectorization logic + * real*real: unpack rhs to constant packets, ... + * + * cd*cd : unpack rhs to (b_r,b_r), (b_i,b_i), mul to get (a_r b_r,a_i b_r) (a_r b_i,a_i b_i), + * storing each res packet into two packets (2x2), + * at the end combine them: swap the second and addsub them + * cf*cf : same but with 2x4 blocks + * cplx*real : unpack rhs to constant packets, ... + * real*cplx : load lhs as (a0,a0,a1,a1), and mul as usual */ -template -struct ei_gebp_kernel +template +class ei_gebp_traits { +public: + typedef _LhsScalar LhsScalar; + typedef _RhsScalar RhsScalar; typedef typename ei_scalar_product_traits::ReturnType ResScalar; enum { + ConjLhs = _ConjLhs, + ConjRhs = _ConjRhs, Vectorizable = ei_product_blocking_traits::Vectorizable, LhsPacketSize = Vectorizable ? ei_packet_traits::size : 1, RhsPacketSize = Vectorizable ? ei_packet_traits::size : 1, - ResPacketSize = Vectorizable ? ei_packet_traits::size : 1 + ResPacketSize = Vectorizable ? ei_packet_traits::size : 1, + + LhsProgress = LhsPacketSize, + RhsProgress = RhsPacketSize }; typedef typename ei_packet_traits::type _LhsPacket; @@ -190,80 +201,381 @@ struct ei_gebp_kernel typedef typename ei_meta_if::ret RhsPacket; typedef typename ei_meta_if::ret ResPacket; + typedef ResPacket AccPacket; + + EIGEN_STRONG_INLINE void initAcc(AccPacket& p) + { + p = ei_pset1(ResScalar(0)); + } + + EIGEN_STRONG_INLINE void unpackRhs(DenseIndex n, const RhsScalar* rhs, RhsScalar* b) + { + for(DenseIndex k=0; k(rhs[k])); + } + + EIGEN_STRONG_INLINE void loadRhs(const RhsScalar* b, RhsPacket& dest) const + { + dest = ei_pload(b); + } + + EIGEN_STRONG_INLINE void loadLhs(const LhsScalar* a, LhsPacket& dest) const + { + dest = ei_pload(a); + } + + EIGEN_STRONG_INLINE void madd(const LhsPacket& a, const RhsPacket& b, AccPacket& c, AccPacket& tmp) const + { + tmp = b; tmp = ei_pmul(a,tmp); c = ei_padd(c,tmp); + } + + EIGEN_STRONG_INLINE void acc(const AccPacket& c, const ResPacket& alpha, ResPacket& r) const + { + r = ei_pmadd(c,alpha,r); + } + +protected: +// ei_conj_helper cj; +// ei_conj_helper pcj; +}; + +template +class ei_gebp_traits, RealScalar, _ConjLhs, false> +{ +public: + typedef std::complex LhsScalar; + typedef RealScalar RhsScalar; + typedef typename ei_scalar_product_traits::ReturnType ResScalar; + + enum { + ConjLhs = _ConjLhs, + ConjRhs = false, + Vectorizable = ei_product_blocking_traits::Vectorizable, + LhsPacketSize = Vectorizable ? ei_packet_traits::size : 1, + RhsPacketSize = Vectorizable ? ei_packet_traits::size : 1, + ResPacketSize = Vectorizable ? ei_packet_traits::size : 1, + + LhsProgress = LhsPacketSize, + RhsProgress = RhsPacketSize + }; + + typedef typename ei_packet_traits::type _LhsPacket; + typedef typename ei_packet_traits::type _RhsPacket; + typedef typename ei_packet_traits::type _ResPacket; + + typedef typename ei_meta_if::ret LhsPacket; + typedef typename ei_meta_if::ret RhsPacket; + typedef typename ei_meta_if::ret ResPacket; + + typedef ResPacket AccPacket; + + EIGEN_STRONG_INLINE void initAcc(AccPacket& p) + { + p = ei_pset1(ResScalar(0)); + } + + EIGEN_STRONG_INLINE void unpackRhs(DenseIndex n, const RhsScalar* rhs, RhsScalar* b) + { + for(DenseIndex k=0; k(rhs[k])); + } + + EIGEN_STRONG_INLINE void loadRhs(const RhsScalar* b, RhsPacket& dest) const + { + dest = ei_pload(b); + } + + EIGEN_STRONG_INLINE void loadLhs(const LhsScalar* a, LhsPacket& dest) const + { + dest = ei_pload(a); + } + + EIGEN_STRONG_INLINE void madd(const LhsPacket& a, const RhsPacket& b, AccPacket& c, RhsPacket& tmp) const + { + madd_impl(a, b, c, tmp, typename ei_meta_if::ret()); + } + + EIGEN_STRONG_INLINE void madd_impl(const LhsPacket& a, const RhsPacket& b, AccPacket& c, RhsPacket& tmp, const ei_meta_true&) const + { + tmp = b; tmp = ei_pmul(a.v,tmp); c.v = ei_padd(c.v,tmp); + } + + EIGEN_STRONG_INLINE void madd_impl(const LhsScalar& a, const RhsScalar& b, ResScalar& c, RhsScalar& /*tmp*/, const ei_meta_false&) const + { + c += a * b; + } + + EIGEN_STRONG_INLINE void acc(const AccPacket& c, const ResPacket& alpha, ResPacket& r) const + { + r = ei_pmadd(c,alpha,r); + } + +protected: +// ei_conj_helper cj; +// ei_conj_helper pcj; +}; + +template +class ei_gebp_traits, std::complex, _ConjLhs, _ConjRhs > +{ +public: + typedef std::complex Scalar; + typedef std::complex LhsScalar; + typedef std::complex RhsScalar; + typedef std::complex ResScalar; + + enum { + ConjLhs = _ConjLhs, + ConjRhs = _ConjRhs, + Vectorizable = ei_packet_traits::Vectorizable + && ei_packet_traits::Vectorizable, + RealPacketSize = Vectorizable ? ei_packet_traits::size : 1, + ResPacketSize = Vectorizable ? ei_packet_traits::size : 1, + + LhsProgress = ResPacketSize, + RhsProgress = Vectorizable ? 2*ResPacketSize : 1 + }; + + typedef typename ei_packet_traits::type RealPacket; + typedef typename ei_packet_traits::type ScalarPacket; + struct DoublePacket + { + RealPacket first; + RealPacket second; + }; + + typedef typename ei_meta_if::ret LhsPacket; + typedef typename ei_meta_if::ret RhsPacket; + typedef typename ei_meta_if::ret ResPacket; + typedef typename ei_meta_if::ret AccPacket; + + EIGEN_STRONG_INLINE void initAcc(Scalar& p) { p = Scalar(0); } + + EIGEN_STRONG_INLINE void initAcc(DoublePacket& p) + { + p.first = ei_pset1(RealScalar(0)); + p.second = ei_pset1(RealScalar(0)); + } + + /* Unpack the rhs coeff such that each complex coefficient is spread into + * two packects containing respectively the real and imaginary coefficient + * duplicated as many time as needed: (x+iy) => [x, ..., x] [y, ..., y] + */ + EIGEN_STRONG_INLINE void unpackRhs(DenseIndex n, const Scalar* rhs, Scalar* b) + { + for(DenseIndex k=0; k(ei_real(rhs[k]))); + ei_pstore((RealScalar*)&b[k*ResPacketSize*2+ResPacketSize], ei_pset1(ei_imag(rhs[k]))); + } + else + b[k] = rhs[k]; + } + } + + EIGEN_STRONG_INLINE void loadRhs(const RhsScalar* b, ResPacket& dest) const { dest = *b; } + + EIGEN_STRONG_INLINE void loadRhs(const RhsScalar* b, DoublePacket& dest) const + { + dest.first = ei_pload((const RealScalar*)b); + dest.second = ei_pload((const RealScalar*)(b+ResPacketSize)); + } + + // nothing special here + EIGEN_STRONG_INLINE void loadLhs(const LhsScalar* a, LhsPacket& dest) const + { + dest = ei_pload((const typename ei_unpacket_traits::type*)(a)); + } + + EIGEN_STRONG_INLINE void madd(const LhsPacket& a, const RhsPacket& b, DoublePacket& c, RhsPacket& /*tmp*/) const + { + c.first = ei_padd(ei_pmul(a,b.first), c.first); + c.second = ei_padd(ei_pmul(a,b.second),c.second); + } + + EIGEN_STRONG_INLINE void madd(const LhsPacket& a, const RhsPacket& b, ResPacket& c, RhsPacket& /*tmp*/) const + { + c = cj.pmadd(a,b,c); + } + + EIGEN_STRONG_INLINE void acc(const Scalar& c, const Scalar& alpha, Scalar& r) const { r += alpha * c; } + + EIGEN_STRONG_INLINE void acc(const DoublePacket& c, const ResPacket& alpha, ResPacket& r) const + { + // assemble c + ResPacket tmp; + if((!ConjLhs)&&(!ConjRhs)) + { + tmp = ei_pcplxflip(ei_pconj(ResPacket(c.second))); + tmp = ei_padd(ResPacket(c.first),tmp); + } + else if((!ConjLhs)&&(ConjRhs)) + { + tmp = ei_pconj(ei_pcplxflip(ResPacket(c.second))); + tmp = ei_padd(ResPacket(c.first),tmp); + } + else if((ConjLhs)&&(!ConjRhs)) + { + tmp = ei_pcplxflip(ResPacket(c.second)); + tmp = ei_padd(ei_pconj(ResPacket(c.first)),tmp); + } + else if((ConjLhs)&&(ConjRhs)) + { + tmp = ei_pcplxflip(ResPacket(c.second)); + tmp = ei_padd(ei_pconj(ResPacket(c.first)),tmp); + } + + r = ei_pmadd(tmp,alpha,r); + } + +protected: + ei_conj_helper cj; +}; + +/* Specialization for real * complex. + * The only subtility is how the lhs coefficients are loaded. + */ +template +struct ei_product_blocking_traits > +{ + typedef std::complex Scalar; + enum { + Vectorizable = ei_packet_traits::Vectorizable, + PacketSize = ei_packet_traits::size, + nr = 4, + mr = 2*PacketSize + }; +}; + +template +class ei_gebp_traits, false, _ConjRhs > +{ +public: + typedef std::complex Scalar; + typedef RealScalar LhsScalar; + typedef Scalar RhsScalar; + typedef Scalar ResScalar; + + enum { + ConjLhs = false, + ConjRhs = _ConjRhs, + Vectorizable = ei_packet_traits::Vectorizable + && ei_packet_traits::Vectorizable, + LhsPacketSize = Vectorizable ? ei_packet_traits::size : 1, + RhsPacketSize = Vectorizable ? ei_packet_traits::size : 1, + ResPacketSize = Vectorizable ? ei_packet_traits::size : 1, + + LhsProgress = ResPacketSize, + RhsProgress = ResPacketSize + }; + + typedef typename ei_packet_traits::type _LhsPacket; + typedef typename ei_packet_traits::type _RhsPacket; + typedef typename ei_packet_traits::type _ResPacket; + + typedef typename ei_meta_if::ret LhsPacket; + typedef typename ei_meta_if::ret RhsPacket; + typedef typename ei_meta_if::ret ResPacket; + + typedef ResPacket AccPacket; + + EIGEN_STRONG_INLINE void initAcc(AccPacket& p) + { + p = ei_pset1(ResScalar(0)); + } + + EIGEN_STRONG_INLINE void unpackRhs(DenseIndex n, const RhsScalar* rhs, RhsScalar* b) + { + for(DenseIndex k=0; k(rhs[k])); + } + + EIGEN_STRONG_INLINE void loadRhs(const RhsScalar* b, RhsPacket& dest) const + { + dest = ei_pload(b); + } + + EIGEN_STRONG_INLINE void loadLhs(const LhsScalar* a, LhsPacket& dest) const + { + dest = ei_ploaddup(a); +// dest = ei_pload(a); + } + + EIGEN_STRONG_INLINE void madd(const LhsPacket& a, const RhsPacket& b, AccPacket& c, RhsPacket& tmp) const + { + madd_impl(a, b, c, tmp, typename ei_meta_if::ret()); + } + + EIGEN_STRONG_INLINE void madd_impl(const LhsPacket& a, const RhsPacket& b, AccPacket& c, RhsPacket& tmp, const ei_meta_true&) const + { + tmp = b; tmp.v = ei_pmul(a,tmp.v); c = ei_padd(c,tmp); + } + + EIGEN_STRONG_INLINE void madd_impl(const LhsScalar& a, const RhsScalar& b, ResScalar& c, RhsScalar& /*tmp*/, const ei_meta_false&) const + { + c += a * b; + } + + EIGEN_STRONG_INLINE void acc(const AccPacket& c, const ResPacket& alpha, ResPacket& r) const + { + r = cj.pmadd(alpha,c,r); + } + +protected: + ei_conj_helper cj; +}; + +/* optimized GEneral packed Block * packed Panel product kernel + * + * Mixing type logic: C += A * B + * | A | B | comments + * |real |cplx | no vectorization yet, would require to pack A with duplication + * |cplx |real | easy vectorization + */ +template +struct ei_gebp_kernel +{ + typedef ei_gebp_traits Traits; + typedef typename Traits::ResScalar ResScalar; + typedef typename Traits::LhsPacket LhsPacket; + typedef typename Traits::RhsPacket RhsPacket; + typedef typename Traits::ResPacket ResPacket; + typedef typename Traits::AccPacket AccPacket; + + enum { + Vectorizable = Traits::Vectorizable, + LhsProgress = Traits::LhsProgress, + RhsProgress = Traits::RhsProgress, + ResPacketSize = Traits::ResPacketSize + }; + EIGEN_FLATTEN_ATTRIB void operator()(ResScalar* res, Index resStride, const LhsScalar* blockA, const RhsScalar* blockB, Index rows, Index depth, Index cols, ResScalar alpha, Index strideA=-1, Index strideB=-1, Index offsetA=0, Index offsetB=0, RhsScalar* unpackedB = 0) { + Traits traits; + if(strideA==-1) strideA = depth; if(strideB==-1) strideB = depth; ei_conj_helper cj; - ei_conj_helper pcj; +// ei_conj_helper pcj; Index packet_cols = (cols/nr) * nr; const Index peeled_mc = (rows/mr)*mr; - const Index peeled_mc2 = peeled_mc + (rows-peeled_mc >= LhsPacketSize ? LhsPacketSize : 0); + // FIXME: + const Index peeled_mc2 = peeled_mc + (rows-peeled_mc >= LhsProgress ? LhsProgress : 0); const Index peeled_kc = (depth/4)*4; if(unpackedB==0) - unpackedB = const_cast(blockB - strideB * nr * RhsPacketSize); + unpackedB = const_cast(blockB - strideB * nr * RhsProgress); // loops on each micro vertical panel of rhs (depth x nr) for(Index j2=0; j2(blB[k])); - /*Scalar* dest = unpackedB; - for(Index k=0; k(blB+0*PacketSize); - C1[0] = ei_pload(blB+1*PacketSize); - C2[0] = ei_pload(blB+2*PacketSize); - C3[0] = ei_pload(blB+3*PacketSize); - - ei_punpackp(C0); - ei_punpackp(C1); - ei_punpackp(C2); - ei_punpackp(C3); - - ei_pstore(dest+ 0*PacketSize, C0[0]); - ei_pstore(dest+ 1*PacketSize, C0[1]); - ei_pstore(dest+ 2*PacketSize, C0[2]); - ei_pstore(dest+ 3*PacketSize, C0[3]); - - ei_pstore(dest+ 4*PacketSize, C1[0]); - ei_pstore(dest+ 5*PacketSize, C1[1]); - ei_pstore(dest+ 6*PacketSize, C1[2]); - ei_pstore(dest+ 7*PacketSize, C1[3]); - - ei_pstore(dest+ 8*PacketSize, C2[0]); - ei_pstore(dest+ 9*PacketSize, C2[1]); - ei_pstore(dest+10*PacketSize, C2[2]); - ei_pstore(dest+11*PacketSize, C2[3]); - - ei_pstore(dest+12*PacketSize, C3[0]); - ei_pstore(dest+13*PacketSize, C3[1]); - ei_pstore(dest+14*PacketSize, C3[2]); - ei_pstore(dest+15*PacketSize, C3[3]); - - blB += 4*PacketSize; - dest += 16*PacketSize; - }*/ - } - // loops on each micro horizontal panel of lhs (mr x depth) + // loops on each largest micro horizontal panel of lhs (mr x depth) // => we select a mr x nr micro block of res which is entirely // stored into mr/packet_size x nr registers. for(Index i=0; i(ResScalar(0)); - C1 = ei_pset1(ResScalar(0)); - if(nr==4) C2 = ei_pset1(ResScalar(0)); - if(nr==4) C3 = ei_pset1(ResScalar(0)); - C4 = ei_pset1(ResScalar(0)); - C5 = ei_pset1(ResScalar(0)); - if(nr==4) C6 = ei_pset1(ResScalar(0)); - if(nr==4) C7 = ei_pset1(ResScalar(0)); + AccPacket C0, C1, C2, C3, C4, C5, C6, C7; + traits.initAcc(C0); + traits.initAcc(C1); + if(nr==4) traits.initAcc(C2); + if(nr==4) traits.initAcc(C3); + traits.initAcc(C4); + traits.initAcc(C5); + if(nr==4) traits.initAcc(C6); + if(nr==4) traits.initAcc(C7); ResScalar* r0 = &res[(j2+0)*resStride + i]; ResScalar* r1 = r0 + resStride; @@ -304,115 +614,114 @@ struct ei_gebp_kernel { LhsPacket A0, A1; RhsPacket B0; - #ifndef EIGEN_HAS_FUSE_CJMADD RhsPacket T0; - #endif -EIGEN_ASM_COMMENT("mybegin"); - A0 = ei_pload(&blA[0*LhsPacketSize]); - A1 = ei_pload(&blA[1*LhsPacketSize]); - B0 = ei_pload(&blB[0*RhsPacketSize]); - MADD(pcj,A0,B0,C0,T0); - MADD(pcj,A1,B0,C4,B0); - B0 = ei_pload(&blB[1*RhsPacketSize]); - MADD(pcj,A0,B0,C1,T0); - MADD(pcj,A1,B0,C5,B0); + +EIGEN_ASM_COMMENT("mybegin2"); + traits.loadLhs(&blA[0*LhsProgress], A0); + traits.loadLhs(&blA[1*LhsProgress], A1); + traits.loadRhs(&blB[0*RhsProgress], B0); + traits.madd(A0,B0,C0,T0); + traits.madd(A1,B0,C4,B0); + traits.loadRhs(&blB[1*RhsProgress], B0); + traits.madd(A0,B0,C1,T0); + traits.madd(A1,B0,C5,B0); - A0 = ei_pload(&blA[2*LhsPacketSize]); - A1 = ei_pload(&blA[3*LhsPacketSize]); - B0 = ei_pload(&blB[2*RhsPacketSize]); - MADD(pcj,A0,B0,C0,T0); - MADD(pcj,A1,B0,C4,B0); - B0 = ei_pload(&blB[3*RhsPacketSize]); - MADD(pcj,A0,B0,C1,T0); - MADD(pcj,A1,B0,C5,B0); + traits.loadLhs(&blA[2*LhsProgress], A0); + traits.loadLhs(&blA[3*LhsProgress], A1); + traits.loadRhs(&blB[2*RhsProgress], B0); + traits.madd(A0,B0,C0,T0); + traits.madd(A1,B0,C4,B0); + traits.loadRhs(&blB[3*RhsProgress], B0); + traits.madd(A0,B0,C1,T0); + traits.madd(A1,B0,C5,B0); - A0 = ei_pload(&blA[4*LhsPacketSize]); - A1 = ei_pload(&blA[5*LhsPacketSize]); - B0 = ei_pload(&blB[4*RhsPacketSize]); - MADD(pcj,A0,B0,C0,T0); - MADD(pcj,A1,B0,C4,B0); - B0 = ei_pload(&blB[5*RhsPacketSize]); - MADD(pcj,A0,B0,C1,T0); - MADD(pcj,A1,B0,C5,B0); + traits.loadLhs(&blA[4*LhsProgress], A0); + traits.loadLhs(&blA[5*LhsProgress], A1); + traits.loadRhs(&blB[4*RhsProgress], B0); + traits.madd(A0,B0,C0,T0); + traits.madd(A1,B0,C4,B0); + traits.loadRhs(&blB[5*RhsProgress], B0); + traits.madd(A0,B0,C1,T0); + traits.madd(A1,B0,C5,B0); - A0 = ei_pload(&blA[6*LhsPacketSize]); - A1 = ei_pload(&blA[7*LhsPacketSize]); - B0 = ei_pload(&blB[6*RhsPacketSize]); - MADD(pcj,A0,B0,C0,T0); - MADD(pcj,A1,B0,C4,B0); - B0 = ei_pload(&blB[7*RhsPacketSize]); - MADD(pcj,A0,B0,C1,T0); - MADD(pcj,A1,B0,C5,B0); + traits.loadLhs(&blA[6*LhsProgress], A0); + traits.loadLhs(&blA[7*LhsProgress], A1); + traits.loadRhs(&blB[6*RhsProgress], B0); + traits.madd(A0,B0,C0,T0); + traits.madd(A1,B0,C4,B0); + traits.loadRhs(&blB[7*RhsProgress], B0); + traits.madd(A0,B0,C1,T0); + traits.madd(A1,B0,C5,B0); EIGEN_ASM_COMMENT("myend"); } else { +EIGEN_ASM_COMMENT("mybegin4"); LhsPacket A0, A1; RhsPacket B0, B1, B2, B3; - #ifndef EIGEN_HAS_FUSE_CJMADD RhsPacket T0; - #endif - A0 = ei_pload(&blA[0*LhsPacketSize]); - A1 = ei_pload(&blA[1*LhsPacketSize]); - B0 = ei_pload(&blB[0*RhsPacketSize]); - B1 = ei_pload(&blB[1*RhsPacketSize]); + + traits.loadLhs(&blA[0*LhsProgress], A0); + traits.loadLhs(&blA[1*LhsProgress], A1); + traits.loadRhs(&blB[0*RhsProgress], B0); + traits.loadRhs(&blB[1*RhsProgress], B1); - MADD(pcj,A0,B0,C0,T0); - B2 = ei_pload(&blB[2*RhsPacketSize]); - MADD(pcj,A1,B0,C4,B0); - B3 = ei_pload(&blB[3*RhsPacketSize]); - B0 = ei_pload(&blB[4*RhsPacketSize]); - MADD(pcj,A0,B1,C1,T0); - MADD(pcj,A1,B1,C5,B1); - B1 = ei_pload(&blB[5*RhsPacketSize]); - MADD(pcj,A0,B2,C2,T0); - MADD(pcj,A1,B2,C6,B2); - B2 = ei_pload(&blB[6*RhsPacketSize]); - MADD(pcj,A0,B3,C3,T0); - A0 = ei_pload(&blA[2*LhsPacketSize]); - MADD(pcj,A1,B3,C7,B3); - A1 = ei_pload(&blA[3*LhsPacketSize]); - B3 = ei_pload(&blB[7*RhsPacketSize]); - MADD(pcj,A0,B0,C0,T0); - MADD(pcj,A1,B0,C4,B0); - B0 = ei_pload(&blB[8*RhsPacketSize]); - MADD(pcj,A0,B1,C1,T0); - MADD(pcj,A1,B1,C5,B1); - B1 = ei_pload(&blB[9*RhsPacketSize]); - MADD(pcj,A0,B2,C2,T0); - MADD(pcj,A1,B2,C6,B2); - B2 = ei_pload(&blB[10*RhsPacketSize]); - MADD(pcj,A0,B3,C3,T0); - A0 = ei_pload(&blA[4*LhsPacketSize]); - MADD(pcj,A1,B3,C7,B3); - A1 = ei_pload(&blA[5*LhsPacketSize]); - B3 = ei_pload(&blB[11*RhsPacketSize]); + traits.madd(A0,B0,C0,T0); + traits.loadRhs(&blB[2*RhsProgress], B2); + traits.madd(A1,B0,C4,B0); + traits.loadRhs(&blB[3*RhsProgress], B3); + traits.loadRhs(&blB[4*RhsProgress], B0); + traits.madd(A0,B1,C1,T0); + traits.madd(A1,B1,C5,B1); + traits.loadRhs(&blB[5*RhsProgress], B1); + traits.madd(A0,B2,C2,T0); + traits.madd(A1,B2,C6,B2); + traits.loadRhs(&blB[6*RhsProgress], B2); + traits.madd(A0,B3,C3,T0); + traits.loadLhs(&blA[2*LhsProgress], A0); + traits.madd(A1,B3,C7,B3); + traits.loadLhs(&blA[3*LhsProgress], A1); + traits.loadRhs(&blB[7*RhsProgress], B3); + traits.madd(A0,B0,C0,T0); + traits.madd(A1,B0,C4,B0); + traits.loadRhs(&blB[8*RhsProgress], B0); + traits.madd(A0,B1,C1,T0); + traits.madd(A1,B1,C5,B1); + traits.loadRhs(&blB[9*RhsProgress], B1); + traits.madd(A0,B2,C2,T0); + traits.madd(A1,B2,C6,B2); + traits.loadRhs(&blB[10*RhsProgress], B2); + traits.madd(A0,B3,C3,T0); + traits.loadLhs(&blA[4*LhsProgress], A0); + traits.madd(A1,B3,C7,B3); + traits.loadLhs(&blA[5*LhsProgress], A1); + traits.loadRhs(&blB[11*RhsProgress], B3); - MADD(pcj,A0,B0,C0,T0); - MADD(pcj,A1,B0,C4,B0); - B0 = ei_pload(&blB[12*RhsPacketSize]); - MADD(pcj,A0,B1,C1,T0); - MADD(pcj,A1,B1,C5,B1); - B1 = ei_pload(&blB[13*RhsPacketSize]); - MADD(pcj,A0,B2,C2,T0); - MADD(pcj,A1,B2,C6,B2); - B2 = ei_pload(&blB[14*RhsPacketSize]); - MADD(pcj,A0,B3,C3,T0); - A0 = ei_pload(&blA[6*LhsPacketSize]); - MADD(pcj,A1,B3,C7,B3); - A1 = ei_pload(&blA[7*LhsPacketSize]); - B3 = ei_pload(&blB[15*RhsPacketSize]); - MADD(pcj,A0,B0,C0,T0); - MADD(pcj,A1,B0,C4,B0); - MADD(pcj,A0,B1,C1,T0); - MADD(pcj,A1,B1,C5,B1); - MADD(pcj,A0,B2,C2,T0); - MADD(pcj,A1,B2,C6,B2); - MADD(pcj,A0,B3,C3,T0); - MADD(pcj,A1,B3,C7,B3); + traits.madd(A0,B0,C0,T0); + traits.madd(A1,B0,C4,B0); + traits.loadRhs(&blB[12*RhsProgress], B0); + traits.madd(A0,B1,C1,T0); + traits.madd(A1,B1,C5,B1); + traits.loadRhs(&blB[13*RhsProgress], B1); + traits.madd(A0,B2,C2,T0); + traits.madd(A1,B2,C6,B2); + traits.loadRhs(&blB[14*RhsProgress], B2); + traits.madd(A0,B3,C3,T0); + traits.loadLhs(&blA[6*LhsProgress], A0); + traits.madd(A1,B3,C7,B3); + traits.loadLhs(&blA[7*LhsProgress], A1); + traits.loadRhs(&blB[15*RhsProgress], B3); + traits.madd(A0,B0,C0,T0); + traits.madd(A1,B0,C4,B0); + traits.madd(A0,B1,C1,T0); + traits.madd(A1,B1,C5,B1); + traits.madd(A0,B2,C2,T0); + traits.madd(A1,B2,C6,B2); + traits.madd(A0,B3,C3,T0); + traits.madd(A1,B3,C7,B3); } - blB += 4*nr*RhsPacketSize; + blB += 4*nr*RhsProgress; blA += 4*mr; } // process remaining peeled loop @@ -422,45 +731,41 @@ EIGEN_ASM_COMMENT("myend"); { LhsPacket A0, A1; RhsPacket B0; - #ifndef EIGEN_HAS_FUSE_CJMADD RhsPacket T0; - #endif - A0 = ei_pload(&blA[0*LhsPacketSize]); - A1 = ei_pload(&blA[1*LhsPacketSize]); - B0 = ei_pload(&blB[0*RhsPacketSize]); - MADD(pcj,A0,B0,C0,T0); - MADD(pcj,A1,B0,C4,B0); - B0 = ei_pload(&blB[1*RhsPacketSize]); - MADD(pcj,A0,B0,C1,T0); - MADD(pcj,A1,B0,C5,B0); + traits.loadLhs(&blA[0*LhsProgress], A0); + traits.loadLhs(&blA[1*LhsProgress], A1); + traits.loadRhs(&blB[0*RhsProgress], B0); + traits.madd(A0,B0,C0,T0); + traits.madd(A1,B0,C4,B0); + traits.loadRhs(&blB[1*RhsProgress], B0); + traits.madd(A0,B0,C1,T0); + traits.madd(A1,B0,C5,B0); } else { LhsPacket A0, A1; RhsPacket B0, B1, B2, B3; - #ifndef EIGEN_HAS_FUSE_CJMADD RhsPacket T0; - #endif - A0 = ei_pload(&blA[0*LhsPacketSize]); - A1 = ei_pload(&blA[1*LhsPacketSize]); - B0 = ei_pload(&blB[0*RhsPacketSize]); - B1 = ei_pload(&blB[1*RhsPacketSize]); + traits.loadLhs(&blA[0*LhsProgress], A0); + traits.loadLhs(&blA[1*LhsProgress], A1); + traits.loadRhs(&blB[0*RhsProgress], B0); + traits.loadRhs(&blB[1*RhsProgress], B1); - MADD(pcj,A0,B0,C0,T0); - B2 = ei_pload(&blB[2*RhsPacketSize]); - MADD(pcj,A1,B0,C4,B0); - B3 = ei_pload(&blB[3*RhsPacketSize]); - MADD(pcj,A0,B1,C1,T0); - MADD(pcj,A1,B1,C5,B1); - MADD(pcj,A0,B2,C2,T0); - MADD(pcj,A1,B2,C6,B2); - MADD(pcj,A0,B3,C3,T0); - MADD(pcj,A1,B3,C7,B3); + traits.madd(A0,B0,C0,T0); + traits.loadRhs(&blB[2*RhsProgress], B2); + traits.madd(A1,B0,C4,B0); + traits.loadRhs(&blB[3*RhsProgress], B3); + traits.madd(A0,B1,C1,T0); + traits.madd(A1,B1,C5,B1); + traits.madd(A0,B2,C2,T0); + traits.madd(A1,B2,C6,B2); + traits.madd(A0,B3,C3,T0); + traits.madd(A1,B3,C7,B3); } - blB += nr*RhsPacketSize; + blB += nr*RhsProgress; blA += mr; } @@ -476,36 +781,37 @@ EIGEN_ASM_COMMENT("myend"); if(nr==4) R6 = ei_ploadu(r2 + ResPacketSize); if(nr==4) R7 = ei_ploadu(r3 + ResPacketSize); - C0 = ei_pmadd(C0, alphav, R0); - C1 = ei_pmadd(C1, alphav, R1); - if(nr==4) C2 = ei_pmadd(C2, alphav, R2); - if(nr==4) C3 = ei_pmadd(C3, alphav, R3); - C4 = ei_pmadd(C4, alphav, R4); - C5 = ei_pmadd(C5, alphav, R5); - if(nr==4) C6 = ei_pmadd(C6, alphav, R6); - if(nr==4) C7 = ei_pmadd(C7, alphav, R7); + traits.acc(C0, alphav, R0); + traits.acc(C1, alphav, R1); + if(nr==4) traits.acc(C2, alphav, R2); + if(nr==4) traits.acc(C3, alphav, R3); + traits.acc(C4, alphav, R4); + traits.acc(C5, alphav, R5); + if(nr==4) traits.acc(C6, alphav, R6); + if(nr==4) traits.acc(C7, alphav, R7); - ei_pstoreu(r0, C0); - ei_pstoreu(r1, C1); - if(nr==4) ei_pstoreu(r2, C2); - if(nr==4) ei_pstoreu(r3, C3); - ei_pstoreu(r0 + ResPacketSize, C4); - ei_pstoreu(r1 + ResPacketSize, C5); - if(nr==4) ei_pstoreu(r2 + ResPacketSize, C6); - if(nr==4) ei_pstoreu(r3 + ResPacketSize, C7); + ei_pstoreu(r0, R0); + ei_pstoreu(r1, R1); + if(nr==4) ei_pstoreu(r2, R2); + if(nr==4) ei_pstoreu(r3, R3); + ei_pstoreu(r0 + ResPacketSize, R4); + ei_pstoreu(r1 + ResPacketSize, R5); + if(nr==4) ei_pstoreu(r2 + ResPacketSize, R6); + if(nr==4) ei_pstoreu(r3 + ResPacketSize, R7); } - if(rows-peeled_mc>=LhsPacketSize) + + if(rows-peeled_mc>=LhsProgress) { Index i = peeled_mc; - const LhsScalar* blA = &blockA[i*strideA+offsetA*LhsPacketSize]; + const LhsScalar* blA = &blockA[i*strideA+offsetA*LhsProgress]; ei_prefetch(&blA[0]); // gets res block as register - ResPacket C0, C1, C2, C3; - C0 = ei_pset1(ResScalar(0)); - C1 = ei_pset1(ResScalar(0)); - if(nr==4) C2 = ei_pset1(ResScalar(0)); - if(nr==4) C3 = ei_pset1(ResScalar(0)); + AccPacket C0, C1, C2, C3; + traits.initAcc(C0); + traits.initAcc(C1); + if(nr==4) traits.initAcc(C2); + if(nr==4) traits.initAcc(C3); // performs "inner" product const RhsScalar* blB = unpackedB; @@ -516,75 +822,75 @@ EIGEN_ASM_COMMENT("myend"); LhsPacket A0; RhsPacket B0, B1; - A0 = ei_pload(&blA[0*LhsPacketSize]); - B0 = ei_pload(&blB[0*RhsPacketSize]); - B1 = ei_pload(&blB[1*RhsPacketSize]); - MADD(pcj,A0,B0,C0,B0); - B0 = ei_pload(&blB[2*RhsPacketSize]); - MADD(pcj,A0,B1,C1,B1); - A0 = ei_pload(&blA[1*LhsPacketSize]); - B1 = ei_pload(&blB[3*RhsPacketSize]); - MADD(pcj,A0,B0,C0,B0); - B0 = ei_pload(&blB[4*RhsPacketSize]); - MADD(pcj,A0,B1,C1,B1); - A0 = ei_pload(&blA[2*LhsPacketSize]); - B1 = ei_pload(&blB[5*RhsPacketSize]); - MADD(pcj,A0,B0,C0,B0); - B0 = ei_pload(&blB[6*RhsPacketSize]); - MADD(pcj,A0,B1,C1,B1); - A0 = ei_pload(&blA[3*LhsPacketSize]); - B1 = ei_pload(&blB[7*RhsPacketSize]); - MADD(pcj,A0,B0,C0,B0); - MADD(pcj,A0,B1,C1,B1); + traits.loadLhs(&blA[0*LhsProgress], A0); + traits.loadRhs(&blB[0*RhsProgress], B0); + traits.loadRhs(&blB[1*RhsProgress], B1); + traits.madd(A0,B0,C0,B0); + traits.loadRhs(&blB[2*RhsProgress], B0); + traits.madd(A0,B1,C1,B1); + traits.loadLhs(&blA[1*LhsProgress], A0); + traits.loadRhs(&blB[3*RhsProgress], B1); + traits.madd(A0,B0,C0,B0); + traits.loadRhs(&blB[4*RhsProgress], B0); + traits.madd(A0,B1,C1,B1); + traits.loadLhs(&blA[2*LhsProgress], A0); + traits.loadRhs(&blB[5*RhsProgress], B1); + traits.madd(A0,B0,C0,B0); + traits.loadRhs(&blB[6*RhsProgress], B0); + traits.madd(A0,B1,C1,B1); + traits.loadLhs(&blA[3*LhsProgress], A0); + traits.loadRhs(&blB[7*RhsProgress], B1); + traits.madd(A0,B0,C0,B0); + traits.madd(A0,B1,C1,B1); } else { LhsPacket A0; RhsPacket B0, B1, B2, B3; - A0 = ei_pload(&blA[0*LhsPacketSize]); - B0 = ei_pload(&blB[0*RhsPacketSize]); - B1 = ei_pload(&blB[1*RhsPacketSize]); + traits.loadLhs(&blA[0*LhsProgress], A0); + traits.loadRhs(&blB[0*RhsProgress], B0); + traits.loadRhs(&blB[1*RhsProgress], B1); - MADD(pcj,A0,B0,C0,B0); - B2 = ei_pload(&blB[2*RhsPacketSize]); - B3 = ei_pload(&blB[3*RhsPacketSize]); - B0 = ei_pload(&blB[4*RhsPacketSize]); - MADD(pcj,A0,B1,C1,B1); - B1 = ei_pload(&blB[5*RhsPacketSize]); - MADD(pcj,A0,B2,C2,B2); - B2 = ei_pload(&blB[6*RhsPacketSize]); - MADD(pcj,A0,B3,C3,B3); - A0 = ei_pload(&blA[1*LhsPacketSize]); - B3 = ei_pload(&blB[7*RhsPacketSize]); - MADD(pcj,A0,B0,C0,B0); - B0 = ei_pload(&blB[8*RhsPacketSize]); - MADD(pcj,A0,B1,C1,B1); - B1 = ei_pload(&blB[9*RhsPacketSize]); - MADD(pcj,A0,B2,C2,B2); - B2 = ei_pload(&blB[10*RhsPacketSize]); - MADD(pcj,A0,B3,C3,B3); - A0 = ei_pload(&blA[2*LhsPacketSize]); - B3 = ei_pload(&blB[11*RhsPacketSize]); + traits.madd(A0,B0,C0,B0); + traits.loadRhs(&blB[2*RhsProgress], B2); + traits.loadRhs(&blB[3*RhsProgress], B3); + traits.loadRhs(&blB[4*RhsProgress], B0); + traits.madd(A0,B1,C1,B1); + traits.loadRhs(&blB[5*RhsProgress], B1); + traits.madd(A0,B2,C2,B2); + traits.loadRhs(&blB[6*RhsProgress], B2); + traits.madd(A0,B3,C3,B3); + traits.loadLhs(&blA[1*LhsProgress], A0); + traits.loadRhs(&blB[7*RhsProgress], B3); + traits.madd(A0,B0,C0,B0); + traits.loadRhs(&blB[8*RhsProgress], B0); + traits.madd(A0,B1,C1,B1); + traits.loadRhs(&blB[9*RhsProgress], B1); + traits.madd(A0,B2,C2,B2); + traits.loadRhs(&blB[10*RhsProgress], B2); + traits.madd(A0,B3,C3,B3); + traits.loadLhs(&blA[2*LhsProgress], A0); + traits.loadRhs(&blB[11*RhsProgress], B3); - MADD(pcj,A0,B0,C0,B0); - B0 = ei_pload(&blB[12*RhsPacketSize]); - MADD(pcj,A0,B1,C1,B1); - B1 = ei_pload(&blB[13*RhsPacketSize]); - MADD(pcj,A0,B2,C2,B2); - B2 = ei_pload(&blB[14*RhsPacketSize]); - MADD(pcj,A0,B3,C3,B3); + traits.madd(A0,B0,C0,B0); + traits.loadRhs(&blB[12*RhsProgress], B0); + traits.madd(A0,B1,C1,B1); + traits.loadRhs(&blB[13*RhsProgress], B1); + traits.madd(A0,B2,C2,B2); + traits.loadRhs(&blB[4*RhsProgress], B2); + traits.madd(A0,B3,C3,B3); - A0 = ei_pload(&blA[3*LhsPacketSize]); - B3 = ei_pload(&blB[15*RhsPacketSize]); - MADD(pcj,A0,B0,C0,B0); - MADD(pcj,A0,B1,C1,B1); - MADD(pcj,A0,B2,C2,B2); - MADD(pcj,A0,B3,C3,B3); + traits.loadLhs(&blA[3*LhsProgress], A0); + traits.loadRhs(&blB[15*RhsProgress], B3); + traits.madd(A0,B0,C0,B0); + traits.madd(A0,B1,C1,B1); + traits.madd(A0,B2,C2,B2); + traits.madd(A0,B3,C3,B3); } - blB += 4*nr*RhsPacketSize; - blA += 4*LhsPacketSize; + blB += nr*4*RhsProgress; + blA += 4*LhsProgress; } // process remaining peeled loop for(Index k=peeled_kc; k(&blA[0*LhsPacketSize]); - B0 = ei_pload(&blB[0*RhsPacketSize]); - B1 = ei_pload(&blB[1*RhsPacketSize]); - MADD(pcj,A0,B0,C0,B0); - MADD(pcj,A0,B1,C1,B1); + traits.loadLhs(&blA[0*LhsProgress], A0); + traits.loadRhs(&blB[0*RhsProgress], B0); + traits.loadRhs(&blB[1*RhsProgress], B1); + traits.madd(A0,B0,C0,B0); + traits.madd(A0,B1,C1,B1); } else { LhsPacket A0; RhsPacket B0, B1, B2, B3; - A0 = ei_pload(&blA[0*LhsPacketSize]); - B0 = ei_pload(&blB[0*RhsPacketSize]); - B1 = ei_pload(&blB[1*RhsPacketSize]); - B2 = ei_pload(&blB[2*RhsPacketSize]); - B3 = ei_pload(&blB[3*RhsPacketSize]); + traits.loadLhs(&blA[0*LhsProgress], A0); + traits.loadRhs(&blB[0*RhsProgress], B0); + traits.loadRhs(&blB[1*RhsProgress], B1); + traits.loadRhs(&blB[2*RhsProgress], B2); + traits.loadRhs(&blB[3*RhsProgress], B3); - MADD(pcj,A0,B0,C0,B0); - MADD(pcj,A0,B1,C1,B1); - MADD(pcj,A0,B2,C2,B2); - MADD(pcj,A0,B3,C3,B3); + traits.madd(A0,B0,C0,B0); + traits.madd(A0,B1,C1,B1); + traits.madd(A0,B2,C2,B2); + traits.madd(A0,B3,C3,B3); } - blB += nr*RhsPacketSize; - blA += LhsPacketSize; + blB += nr*RhsProgress; + blA += LhsProgress; } ResPacket R0, R1, R2, R3; @@ -634,15 +940,15 @@ EIGEN_ASM_COMMENT("myend"); if(nr==4) R2 = ei_ploadu(r2); if(nr==4) R3 = ei_ploadu(r3); - C0 = ei_pmadd(C0, alphav, R0); - C1 = ei_pmadd(C1, alphav, R1); - if(nr==4) C2 = ei_pmadd(C2, alphav, R2); - if(nr==4) C3 = ei_pmadd(C3, alphav, R3); + traits.acc(C0, alphav, R0); + traits.acc(C1, alphav, R1); + if(nr==4) traits.acc(C2, alphav, R2); + if(nr==4) traits.acc(C3, alphav, R3); - ei_pstoreu(r0, C0); - ei_pstoreu(r1, C1); - if(nr==4) ei_pstoreu(r2, C2); - if(nr==4) ei_pstoreu(r3, C3); + ei_pstoreu(r0, R0); + ei_pstoreu(r1, R1); + if(nr==4) ei_pstoreu(r2, R2); + if(nr==4) ei_pstoreu(r3, R3); } for(Index i=peeled_mc2; i do the same but with nr==1 for(Index j2=packet_cols; j2(blB[k])); + traits.unpackRhs(depth, &blockB[j2*strideB+offsetB], unpackedB); +// const RhsScalar* blB = &blockB[j2*strideB+offsetB]; +// for(Index k=0; k(blB[k])); } for(Index i=0; i(ResScalar(0)); - C4 = ei_pset1(ResScalar(0)); - + AccPacket C0, C4; + traits.initAcc(C0); + traits.initAcc(C4); const RhsScalar* blB = unpackedB; for(Index k=0; k(&blA[0*LhsPacketSize]); - A1 = ei_pload(&blA[1*LhsPacketSize]); - B0 = ei_pload(&blB[0*RhsPacketSize]); - MADD(pcj,A0,B0,C0,T0); - MADD(pcj,A1,B0,C4,B0); + traits.loadLhs(&blA[0*LhsProgress], A0); + traits.loadLhs(&blA[1*LhsProgress], A1); + traits.loadRhs(&blB[0*RhsProgress], B0); + traits.madd(A0,B0,C0,T0); + traits.madd(A1,B0,C4,B0); - blB += RhsPacketSize; - blA += mr; + blB += RhsProgress; + blA += 2*LhsProgress; } ResPacket R0, R4; ResPacket alphav = ei_pset1(alpha); @@ -742,33 +1045,37 @@ EIGEN_ASM_COMMENT("myend"); R0 = ei_ploadu(r0); R4 = ei_ploadu(r0+ResPacketSize); - C0 = ei_pmadd(C0, alphav, R0); - C4 = ei_pmadd(C4, alphav, R4); + traits.acc(C0, alphav, R0); + traits.acc(C4, alphav, R4); - ei_pstoreu(r0, C0); - ei_pstoreu(r0+ResPacketSize, C4); + ei_pstoreu(r0, R0); + ei_pstoreu(r0+ResPacketSize, R4); } - if(rows-peeled_mc>=LhsPacketSize) + if(rows-peeled_mc>=LhsProgress) { Index i = peeled_mc; - const LhsScalar* blA = &blockA[i*strideA+offsetA*LhsPacketSize]; + const LhsScalar* blA = &blockA[i*strideA+offsetA*LhsProgress]; ei_prefetch(&blA[0]); - ResPacket C0 = ei_pset1(ResScalar(0)); + AccPacket C0; + traits.initAcc(C0); const RhsScalar* blB = unpackedB; for(Index k=0; k(blA); - RhsPacket B0 = ei_pload(blB); - MADD(pcj, A0, B0, C0, B0); - blB += RhsPacketSize; - blA += LhsPacketSize; + LhsPacket A0; + RhsPacket B0; + traits.loadLhs(blA, A0); + traits.loadRhs(blB, B0); + traits.madd(A0, B0, C0, B0); + blB += RhsProgress; + blA += LhsProgress; } ResPacket alphav = ei_pset1(alpha); ResPacket R0 = ei_ploadu(&res[(j2+0)*resStride + i]); - ei_pstoreu(&res[(j2+0)*resStride + i], ei_pmadd(C0, alphav, R0)); + traits.acc(C0, alphav, R0); + ei_pstoreu(&res[(j2+0)*resStride + i], R0); } for(Index i=peeled_mc2; i::size*Blocking::nr; + std::size_t sizeW = kc*ei_packet_traits::size*Blocking::nr*2; LhsScalar *blockA = blocking.blockA()==0 ? ei_aligned_stack_new(LhsScalar, sizeA) : blocking.blockA(); RhsScalar *blockB = blocking.blockB()==0 ? ei_aligned_stack_new(RhsScalar, sizeB) : blocking.blockB(); RhsScalar *blockW = blocking.blockW()==0 ? ei_aligned_stack_new(RhsScalar, sizeW) : blocking.blockW(); diff --git a/Eigen/src/Core/util/BlasUtil.h b/Eigen/src/Core/util/BlasUtil.h index 797f6b0c4..a5e88c9d7 100644 --- a/Eigen/src/Core/util/BlasUtil.h +++ b/Eigen/src/Core/util/BlasUtil.h @@ -160,8 +160,8 @@ struct ei_product_blocking_traits enum { Vectorizable = ei_packet_traits::Vectorizable && ei_packet_traits::Vectorizable - && (ei_is_same_type::ret - || (NumTraits::IsComplex && !NumTraits::IsComplex)), + /*&& (ei_is_same_type::ret + || (NumTraits::IsComplex && !NumTraits::IsComplex))*/, LhsPacketSize = Vectorizable ? ei_packet_traits::size : 1, NumberOfRegisters = EIGEN_ARCH_DEFAULT_NUMBER_OF_REGISTERS,