From 471cfe5ff7577b51adc37de61ca39ad63a2d0b40 Mon Sep 17 00:00:00 2001
From: Deven Desai <deven.desai.amd@gmail.com>
Date: Wed, 11 Jul 2018 09:22:04 -0400
Subject: renaming CUDA* to GPU* for some header files

---
 Eigen/src/Core/arch/CUDA/MathFunctions.h | 103 ----------
 Eigen/src/Core/arch/CUDA/PacketMath.h    | 341 -------------------------------
 2 files changed, 444 deletions(-)
 delete mode 100644 Eigen/src/Core/arch/CUDA/MathFunctions.h
 delete mode 100644 Eigen/src/Core/arch/CUDA/PacketMath.h

(limited to 'Eigen/src/Core/arch/CUDA')

diff --git a/Eigen/src/Core/arch/CUDA/MathFunctions.h b/Eigen/src/Core/arch/CUDA/MathFunctions.h
deleted file mode 100644
index ff6256ce0..000000000
--- a/Eigen/src/Core/arch/CUDA/MathFunctions.h
+++ /dev/null
@@ -1,103 +0,0 @@
-// This file is part of Eigen, a lightweight C++ template library
-// for linear algebra.
-//
-// Copyright (C) 2014 Benoit Steiner <benoit.steiner.goog@gmail.com>
-//
-// This Source Code Form is subject to the terms of the Mozilla
-// Public License v. 2.0. If a copy of the MPL was not distributed
-// with this file, You can obtain one at http://mozilla.org/MPL/2.0/.
-
-#ifndef EIGEN_MATH_FUNCTIONS_CUDA_H
-#define EIGEN_MATH_FUNCTIONS_CUDA_H
-
-namespace Eigen {
-
-namespace internal {
-
-// Make sure this is only available when targeting a GPU: we don't want to
-// introduce conflicts between these packet_traits definitions and the ones
-// we'll use on the host side (SSE, AVX, ...)
-#if defined(EIGEN_CUDACC) && defined(EIGEN_USE_GPU)
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE
-float4 plog<float4>(const float4& a)
-{
-  return make_float4(logf(a.x), logf(a.y), logf(a.z), logf(a.w));
-}
-
-template<>  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE
-double2 plog<double2>(const double2& a)
-{
-  using ::log;
-  return make_double2(log(a.x), log(a.y));
-}
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE
-float4 plog1p<float4>(const float4& a)
-{
-  return make_float4(log1pf(a.x), log1pf(a.y), log1pf(a.z), log1pf(a.w));
-}
-
-template<>  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE
-double2 plog1p<double2>(const double2& a)
-{
-  return make_double2(log1p(a.x), log1p(a.y));
-}
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE
-float4 pexp<float4>(const float4& a)
-{
-  return make_float4(expf(a.x), expf(a.y), expf(a.z), expf(a.w));
-}
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE
-double2 pexp<double2>(const double2& a)
-{
-  using ::exp;
-  return make_double2(exp(a.x), exp(a.y));
-}
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE
-float4 pexpm1<float4>(const float4& a)
-{
-  return make_float4(expm1f(a.x), expm1f(a.y), expm1f(a.z), expm1f(a.w));
-}
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE
-double2 pexpm1<double2>(const double2& a)
-{
-  return make_double2(expm1(a.x), expm1(a.y));
-}
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE
-float4 psqrt<float4>(const float4& a)
-{
-  return make_float4(sqrtf(a.x), sqrtf(a.y), sqrtf(a.z), sqrtf(a.w));
-}
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE
-double2 psqrt<double2>(const double2& a)
-{
-  using ::sqrt;
-  return make_double2(sqrt(a.x), sqrt(a.y));
-}
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE
-float4 prsqrt<float4>(const float4& a)
-{
-  return make_float4(rsqrtf(a.x), rsqrtf(a.y), rsqrtf(a.z), rsqrtf(a.w));
-}
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE
-double2 prsqrt<double2>(const double2& a)
-{
-  return make_double2(rsqrt(a.x), rsqrt(a.y));
-}
-
-
-#endif
-
-} // end namespace internal
-
-} // end namespace Eigen
-
-#endif // EIGEN_MATH_FUNCTIONS_CUDA_H
diff --git a/Eigen/src/Core/arch/CUDA/PacketMath.h b/Eigen/src/Core/arch/CUDA/PacketMath.h
deleted file mode 100644
index ab8e477f4..000000000
--- a/Eigen/src/Core/arch/CUDA/PacketMath.h
+++ /dev/null
@@ -1,341 +0,0 @@
-// This file is part of Eigen, a lightweight C++ template library
-// for linear algebra.
-//
-// Copyright (C) 2014 Benoit Steiner <benoit.steiner.goog@gmail.com>
-//
-// This Source Code Form is subject to the terms of the Mozilla
-// Public License v. 2.0. If a copy of the MPL was not distributed
-// with this file, You can obtain one at http://mozilla.org/MPL/2.0/.
-
-#ifndef EIGEN_PACKET_MATH_CUDA_H
-#define EIGEN_PACKET_MATH_CUDA_H
-
-namespace Eigen {
-
-namespace internal {
-
-// Make sure this is only available when targeting a GPU: we don't want to
-// introduce conflicts between these packet_traits definitions and the ones
-// we'll use on the host side (SSE, AVX, ...)
-#if defined(EIGEN_CUDACC) && defined(EIGEN_USE_GPU)
-template<> struct is_arithmetic<float4>  { enum { value = true }; };
-template<> struct is_arithmetic<double2> { enum { value = true }; };
-
-template<> struct packet_traits<float> : default_packet_traits
-{
-  typedef float4 type;
-  typedef float4 half;
-  enum {
-    Vectorizable = 1,
-    AlignedOnScalar = 1,
-    size=4,
-    HasHalfPacket = 0,
-
-    HasDiv  = 1,
-    HasSin  = 0,
-    HasCos  = 0,
-    HasLog  = 1,
-    HasExp  = 1,
-    HasSqrt = 1,
-    HasRsqrt = 1,
-    HasLGamma = 1,
-    HasDiGamma = 1,
-    HasZeta = 1,
-    HasPolygamma = 1,
-    HasErf = 1,
-    HasErfc = 1,
-    HasI0e = 1,
-    HasI1e = 1,
-    HasIGamma = 1,
-    HasIGammaDerA = 1,
-    HasGammaSampleDerAlpha = 1,
-    HasIGammac = 1,
-    HasBetaInc = 1,
-
-    HasBlend = 0,
-  };
-};
-
-template<> struct packet_traits<double> : default_packet_traits
-{
-  typedef double2 type;
-  typedef double2 half;
-  enum {
-    Vectorizable = 1,
-    AlignedOnScalar = 1,
-    size=2,
-    HasHalfPacket = 0,
-
-    HasDiv  = 1,
-    HasLog  = 1,
-    HasExp  = 1,
-    HasSqrt = 1,
-    HasRsqrt = 1,
-    HasLGamma = 1,
-    HasDiGamma = 1,
-    HasZeta = 1,
-    HasPolygamma = 1,
-    HasErf = 1,
-    HasErfc = 1,
-    HasI0e = 1,
-    HasI1e = 1,
-    HasIGamma = 1,
-    HasIGammaDerA = 1,
-    HasGammaSampleDerAlpha = 1,
-    HasIGammac = 1,
-    HasBetaInc = 1,
-
-    HasBlend = 0,
-  };
-};
-
-
-template<> struct unpacket_traits<float4>  { typedef float  type; enum {size=4, alignment=Aligned16}; typedef float4 half; };
-template<> struct unpacket_traits<double2> { typedef double type; enum {size=2, alignment=Aligned16}; typedef double2 half; };
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE float4 pset1<float4>(const float&  from) {
-  return make_float4(from, from, from, from);
-}
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE double2 pset1<double2>(const double& from) {
-  return make_double2(from, from);
-}
-
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE float4 plset<float4>(const float& a) {
-  return make_float4(a, a+1, a+2, a+3);
-}
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE double2 plset<double2>(const double& a) {
-  return make_double2(a, a+1);
-}
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE float4 padd<float4>(const float4& a, const float4& b) {
-  return make_float4(a.x+b.x, a.y+b.y, a.z+b.z, a.w+b.w);
-}
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE double2 padd<double2>(const double2& a, const double2& b) {
-  return make_double2(a.x+b.x, a.y+b.y);
-}
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE float4 psub<float4>(const float4& a, const float4& b) {
-  return make_float4(a.x-b.x, a.y-b.y, a.z-b.z, a.w-b.w);
-}
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE double2 psub<double2>(const double2& a, const double2& b) {
-  return make_double2(a.x-b.x, a.y-b.y);
-}
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE float4 pnegate(const float4& a) {
-  return make_float4(-a.x, -a.y, -a.z, -a.w);
-}
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE double2 pnegate(const double2& a) {
-  return make_double2(-a.x, -a.y);
-}
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE float4 pconj(const float4& a) { return a; }
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE double2 pconj(const double2& a) { return a; }
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE float4 pmul<float4>(const float4& a, const float4& b) {
-  return make_float4(a.x*b.x, a.y*b.y, a.z*b.z, a.w*b.w);
-}
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE double2 pmul<double2>(const double2& a, const double2& b) {
-  return make_double2(a.x*b.x, a.y*b.y);
-}
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE float4 pdiv<float4>(const float4& a, const float4& b) {
-  return make_float4(a.x/b.x, a.y/b.y, a.z/b.z, a.w/b.w);
-}
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE double2 pdiv<double2>(const double2& a, const double2& b) {
-  return make_double2(a.x/b.x, a.y/b.y);
-}
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE float4 pmin<float4>(const float4& a, const float4& b) {
-  return make_float4(fminf(a.x, b.x), fminf(a.y, b.y), fminf(a.z, b.z), fminf(a.w, b.w));
-}
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE double2 pmin<double2>(const double2& a, const double2& b) {
-  return make_double2(fmin(a.x, b.x), fmin(a.y, b.y));
-}
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE float4 pmax<float4>(const float4& a, const float4& b) {
-  return make_float4(fmaxf(a.x, b.x), fmaxf(a.y, b.y), fmaxf(a.z, b.z), fmaxf(a.w, b.w));
-}
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE double2 pmax<double2>(const double2& a, const double2& b) {
-  return make_double2(fmax(a.x, b.x), fmax(a.y, b.y));
-}
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE float4 pload<float4>(const float* from) {
-  return *reinterpret_cast<const float4*>(from);
-}
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE double2 pload<double2>(const double* from) {
-  return *reinterpret_cast<const double2*>(from);
-}
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE float4 ploadu<float4>(const float* from) {
-  return make_float4(from[0], from[1], from[2], from[3]);
-}
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE double2 ploadu<double2>(const double* from) {
-  return make_double2(from[0], from[1]);
-}
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE float4 ploaddup<float4>(const float*   from) {
-  return make_float4(from[0], from[0], from[1], from[1]);
-}
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE double2 ploaddup<double2>(const double*  from) {
-  return make_double2(from[0], from[0]);
-}
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE void pstore<float>(float*   to, const float4& from) {
-  *reinterpret_cast<float4*>(to) = from;
-}
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE void pstore<double>(double* to, const double2& from) {
-  *reinterpret_cast<double2*>(to) = from;
-}
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE void pstoreu<float>(float*  to, const float4& from) {
-  to[0] = from.x;
-  to[1] = from.y;
-  to[2] = from.z;
-  to[3] = from.w;
-}
-
-template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE void pstoreu<double>(double* to, const double2& from) {
-  to[0] = from.x;
-  to[1] = from.y;
-}
-
-template<>
-EIGEN_DEVICE_FUNC EIGEN_ALWAYS_INLINE float4 ploadt_ro<float4, Aligned>(const float* from) {
-#if defined(EIGEN_CUDA_ARCH) && EIGEN_CUDA_ARCH >= 350
-  return __ldg((const float4*)from);
-#else
-  return make_float4(from[0], from[1], from[2], from[3]);
-#endif
-}
-template<>
-EIGEN_DEVICE_FUNC EIGEN_ALWAYS_INLINE double2 ploadt_ro<double2, Aligned>(const double* from) {
-#if defined(EIGEN_CUDA_ARCH) && EIGEN_CUDA_ARCH >= 350
-  return __ldg((const double2*)from);
-#else
-  return make_double2(from[0], from[1]);
-#endif
-}
-
-template<>
-EIGEN_DEVICE_FUNC EIGEN_ALWAYS_INLINE float4 ploadt_ro<float4, Unaligned>(const float* from) {
-#if defined(EIGEN_CUDA_ARCH) && EIGEN_CUDA_ARCH >= 350
-  return make_float4(__ldg(from+0), __ldg(from+1), __ldg(from+2), __ldg(from+3));
-#else
-  return make_float4(from[0], from[1], from[2], from[3]);
-#endif
-}
-template<>
-EIGEN_DEVICE_FUNC EIGEN_ALWAYS_INLINE double2 ploadt_ro<double2, Unaligned>(const double* from) {
-#if defined(EIGEN_CUDA_ARCH) && EIGEN_CUDA_ARCH >= 350
-  return make_double2(__ldg(from+0), __ldg(from+1));
-#else
-  return make_double2(from[0], from[1]);
-#endif
-}
-
-template<> EIGEN_DEVICE_FUNC inline float4 pgather<float, float4>(const float* from, Index stride) {
-  return make_float4(from[0*stride], from[1*stride], from[2*stride], from[3*stride]);
-}
-
-template<> EIGEN_DEVICE_FUNC inline double2 pgather<double, double2>(const double* from, Index stride) {
-  return make_double2(from[0*stride], from[1*stride]);
-}
-
-template<> EIGEN_DEVICE_FUNC inline void pscatter<float, float4>(float* to, const float4& from, Index stride) {
-  to[stride*0] = from.x;
-  to[stride*1] = from.y;
-  to[stride*2] = from.z;
-  to[stride*3] = from.w;
-}
-template<> EIGEN_DEVICE_FUNC inline void pscatter<double, double2>(double* to, const double2& from, Index stride) {
-  to[stride*0] = from.x;
-  to[stride*1] = from.y;
-}
-
-template<> EIGEN_DEVICE_FUNC inline float  pfirst<float4>(const float4& a) {
-  return a.x;
-}
-template<> EIGEN_DEVICE_FUNC inline double pfirst<double2>(const double2& a) {
-  return a.x;
-}
-
-template<> EIGEN_DEVICE_FUNC inline float  predux<float4>(const float4& a) {
-  return a.x + a.y + a.z + a.w;
-}
-template<> EIGEN_DEVICE_FUNC inline double predux<double2>(const double2& a) {
-  return a.x + a.y;
-}
-
-template<> EIGEN_DEVICE_FUNC inline float  predux_max<float4>(const float4& a) {
-  return fmaxf(fmaxf(a.x, a.y), fmaxf(a.z, a.w));
-}
-template<> EIGEN_DEVICE_FUNC inline double predux_max<double2>(const double2& a) {
-  return fmax(a.x, a.y);
-}
-
-template<> EIGEN_DEVICE_FUNC inline float  predux_min<float4>(const float4& a) {
-  return fminf(fminf(a.x, a.y), fminf(a.z, a.w));
-}
-template<> EIGEN_DEVICE_FUNC inline double predux_min<double2>(const double2& a) {
-  return fmin(a.x, a.y);
-}
-
-template<> EIGEN_DEVICE_FUNC inline float  predux_mul<float4>(const float4& a) {
-  return a.x * a.y * a.z * a.w;
-}
-template<> EIGEN_DEVICE_FUNC inline double predux_mul<double2>(const double2& a) {
-  return a.x * a.y;
-}
-
-template<> EIGEN_DEVICE_FUNC inline float4  pabs<float4>(const float4& a) {
-  return make_float4(fabsf(a.x), fabsf(a.y), fabsf(a.z), fabsf(a.w));
-}
-template<> EIGEN_DEVICE_FUNC inline double2 pabs<double2>(const double2& a) {
-  return make_double2(fabs(a.x), fabs(a.y));
-}
-
-EIGEN_DEVICE_FUNC inline void
-ptranspose(PacketBlock<float4,4>& kernel) {
-  float tmp = kernel.packet[0].y;
-  kernel.packet[0].y = kernel.packet[1].x;
-  kernel.packet[1].x = tmp;
-
-  tmp = kernel.packet[0].z;
-  kernel.packet[0].z = kernel.packet[2].x;
-  kernel.packet[2].x = tmp;
-
-  tmp = kernel.packet[0].w;
-  kernel.packet[0].w = kernel.packet[3].x;
-  kernel.packet[3].x = tmp;
-
-  tmp = kernel.packet[1].z;
-  kernel.packet[1].z = kernel.packet[2].y;
-  kernel.packet[2].y = tmp;
-
-  tmp = kernel.packet[1].w;
-  kernel.packet[1].w = kernel.packet[3].y;
-  kernel.packet[3].y = tmp;
-
-  tmp = kernel.packet[2].w;
-  kernel.packet[2].w = kernel.packet[3].z;
-  kernel.packet[3].z = tmp;
-}
-
-EIGEN_DEVICE_FUNC inline void
-ptranspose(PacketBlock<double2,2>& kernel) {
-  double tmp = kernel.packet[0].y;
-  kernel.packet[0].y = kernel.packet[1].x;
-  kernel.packet[1].x = tmp;
-}
-
-#endif
-
-} // end namespace internal
-
-} // end namespace Eigen
-
-
-#endif // EIGEN_PACKET_MATH_CUDA_H
-- 
cgit v1.2.3