Merge pull request #1203 from stanmoore1/kk_update

# Change Log

## [2.7.24](https://github.com/kokkos/kokkos/tree/2.7.24) (2018-11-04)

[Full Changelog](https://github.com/kokkos/kokkos/compare/2.7.00...2.7.24)

**Implemented enhancements:**

- DualView: Add non-templated functions for sync, need\_sync, view, modify [\#1858](https://github.com/kokkos/kokkos/issues/1858)

- DualView: Avoid needlessly allocates and initializes modify\_host and modify\_device flag views [\#1831](https://github.com/kokkos/kokkos/issues/1831)

- DualView: Incorrect deduction of "not device type" [\#1659](https://github.com/kokkos/kokkos/issues/1659)

- BuildSystem: Add KOKKOS\_ENABLE\_CXX14 and KOKKOS\_ENABLE\_CXX17 [\#1602](https://github.com/kokkos/kokkos/issues/1602)

- BuildSystem: Installed kokkos\_generated\_settings.cmake contains build directories instead of install directories [\#1838](https://github.com/kokkos/kokkos/issues/1838)

- BuildSystem: KOKKOS\_ARCH: add ticks to printout of improper arch setting [\#1649](https://github.com/kokkos/kokkos/issues/1649)

- BuildSystem: Make core/src/Makefile for Cuda use needed nvcc\_wrapper [\#1296](https://github.com/kokkos/kokkos/issues/1296)

- Build: Support PGI as host compiler for NVCC [\#1828](https://github.com/kokkos/kokkos/issues/1828)

- Build: Many Warnings Fixed e.g.[\#1786](https://github.com/kokkos/kokkos/issues/1786)

- Capability: OffsetView with non-zero begin index [\#567](https://github.com/kokkos/kokkos/issues/567)

- Capability: Reductions into device side view [\#1788](https://github.com/kokkos/kokkos/issues/1788)

- Capability: Add max\_size to Kokkos::Array [\#1760](https://github.com/kokkos/kokkos/issues/1760)

- Capability: View Assignment: LayoutStride -\> LayoutLeft and LayoutStride -\> LayoutRight [\#1594](https://github.com/kokkos/kokkos/issues/1594)

- Capability: Atomic function allow implicit conversion of update argument [\#1571](https://github.com/kokkos/kokkos/issues/1571)

- Capability: Add team\_size\_max with tagged functors [\#663](https://github.com/kokkos/kokkos/issues/663)

- Capability: Fix allignment of views from Kokkos\_ScratchSpace should use different alignment [\#1700](https://github.com/kokkos/kokkos/issues/1700)

- Capabilitiy: create\_mirror\_view\_and\_copy for DynRankView [\#1651](https://github.com/kokkos/kokkos/issues/1651)

- Capability: DeepCopy HBWSpace / HostSpace [\#548](https://github.com/kokkos/kokkos/issues/548)

- ROCm: support team vector scan  [\#1645](https://github.com/kokkos/kokkos/issues/1645)

- ROCm:  Merge from rocm-hackathon2 [\#1636](https://github.com/kokkos/kokkos/issues/1636)

- ROCm:  Add ParallelScanWithTotal [\#1611](https://github.com/kokkos/kokkos/issues/1611)

- ROCm: Implement MDRange in ROCm [\#1314](https://github.com/kokkos/kokkos/issues/1314)

- ROCm: Implement Reducers for Nested Parallelism Levels [\#963](https://github.com/kokkos/kokkos/issues/963)

- ROCm: Add asynchronous deep copy [\#959](https://github.com/kokkos/kokkos/issues/959)

- Tests: Memory pool test seems to allocate 8GB [\#1830](https://github.com/kokkos/kokkos/issues/1830)

- Tests: Add unit\_test for team\_broadcast [\#734](https://github.com/kokkos/kokkos/issues/734)

**Fixed bugs:**

- BuildSystem: Makefile.kokkos gets gcc-toolchain wrong if gcc is cached [\#1841](https://github.com/kokkos/kokkos/issues/1841)

- BuildSystem: kokkos\_generated\_settings.cmake placement is inconsistent [\#1771](https://github.com/kokkos/kokkos/issues/1771)

- BuildSystem: Invalid escape sequence \. in kokkos\_functions.cmake [\#1661](https://github.com/kokkos/kokkos/issues/1661)

- BuildSystem: Problem in Kokkos generated cmake file [\#1770](https://github.com/kokkos/kokkos/issues/1770)

- BuildSystem: invalid file names on windows [\#1671](https://github.com/kokkos/kokkos/issues/1671)

- Tests: reducers min/max\_loc test fails randomly due to multiple min values and thus multiple valid locations [\#1681](https://github.com/kokkos/kokkos/issues/1681)

- Tests: cuda.scatterview unit test causes "Bus error" when force\_uvm and enable\_lambda are enabled [\#1852](https://github.com/kokkos/kokkos/issues/1852)

- Tests: cuda.cxx11 unit test fails when force\_uvm and enable\_lambda are enabled [\#1850](https://github.com/kokkos/kokkos/issues/1850)

- Tests: threads.reduce\_device\_view\_range\_policy failing with Cuda/8.0.44 and RDC [\#1836](https://github.com/kokkos/kokkos/issues/1836)

- Build: compile error when compiling Kokkos with hwloc 2.0.1 \(on OSX 10.12.6, with g++ 7.2.0\) [\#1506](https://github.com/kokkos/kokkos/issues/1506)

- Build: dual\_view.view broken with UVM [\#1834](https://github.com/kokkos/kokkos/issues/1834)

- Build: White cuda/9.2 + gcc/7.2 warnings triggering errors  [\#1833](https://github.com/kokkos/kokkos/issues/1833)

- Build: warning: enum constant in boolean context [\#1813](https://github.com/kokkos/kokkos/issues/1813)

- Capability: Fix overly conservative max\_team\_size thingy [\#1808](https://github.com/kokkos/kokkos/issues/1808)

- DynRankView: Ctors taking ViewAllocateWithoutInitializing broken [\#1783](https://github.com/kokkos/kokkos/issues/1783)

- Cuda: Apollo cuda.team\_broadcast test fail with clang-6.0 [\#1762](https://github.com/kokkos/kokkos/issues/1762)

- Cuda: Clang spurious test failure in impl\_view\_accessible [\#1753](https://github.com/kokkos/kokkos/issues/1753)

- Cuda: Kokkos::complex\<double\> atomic deadlocks with Clang 6 Cuda build with -O0 [\#1752](https://github.com/kokkos/kokkos/issues/1752)

- Cuda: LayoutStride Test fails for UVM as default memory space [\#1688](https://github.com/kokkos/kokkos/issues/1688)

- Cuda: Scan wrong values on Volta [\#1676](https://github.com/kokkos/kokkos/issues/1676)

- Cuda: Kokkos::deep\_copy error with CudaUVM and Kokkos::Serial spaces [\#1652](https://github.com/kokkos/kokkos/issues/1652)

- Cuda: cudaErrorInvalidConfiguration with debug build [\#1647](https://github.com/kokkos/kokkos/issues/1647)

- Cuda: parallel\_for with TeamPolicy::team\_size\_recommended with launch bounds not working -- reported by Daniel Holladay [\#1283](https://github.com/kokkos/kokkos/issues/1283)

- Cuda: Using KOKKOS\_CLASS\_LAMBDA in a class with Kokkos::Random\_XorShift64\_Pool member data [\#1696](https://github.com/kokkos/kokkos/issues/1696)

- Long Build Times on Darwin [\#1721](https://github.com/kokkos/kokkos/issues/1721)

- Capability: Typo in Kokkos\_Sort.hpp - BinOp3D - wrong comparison [\#1720](https://github.com/kokkos/kokkos/issues/1720)

- Buffer overflow in SharedAllocationRecord in Kokkos\_HostSpace.cpp [\#1673](https://github.com/kokkos/kokkos/issues/1673)

- Serial unit test failure [\#1632](https://github.com/kokkos/kokkos/issues/1632)

## [2.7.00](https://github.com/kokkos/kokkos/tree/2.7.00) (2018-05-24)

[Full Changelog](https://github.com/kokkos/kokkos/compare/2.6.00...2.7.00)

  #------------ GENERATE HEADER AND SOURCE FILES -------------------------------

  execute_process(

    COMMAND ${KOKKOS_SETTINGS} make -f ${KOKKOS_SRC_PATH}/cmake/Makefile.generate_cmake_settings CXX=${CMAKE_CXX_COMPILER} generate_build_settings

    COMMAND ${KOKKOS_SETTINGS} make -f ${KOKKOS_SRC_PATH}/cmake/Makefile.generate_cmake_settings CXX=${CMAKE_CXX_COMPILER} PREFIX=${CMAKE_INSTALL_PREFIX} generate_build_settings

    WORKING_DIRECTORY "${Kokkos_BINARY_DIR}"

    OUTPUT_FILE ${Kokkos_BINARY_DIR}/core_src_make.out

    RESULT_VARIABLE GEN_SETTINGS_RESULT

  endif()

  include(${Kokkos_BINARY_DIR}/kokkos_generated_settings.cmake)

  install(FILES ${Kokkos_BINARY_DIR}/kokkos_generated_settings.cmake DESTINATION lib/cmake/Kokkos)

  install(FILES ${Kokkos_BINARY_DIR}/kokkos_generated_settings.cmake DESTINATION ${CMAKE_INSTALL_PREFIX})

  string(REPLACE " " ";" KOKKOS_TPL_INCLUDE_DIRS "${KOKKOS_GMAKE_TPL_INCLUDE_DIRS}")

  string(REPLACE " " ";" KOKKOS_TPL_LIBRARY_DIRS "${KOKKOS_GMAKE_TPL_LIBRARY_DIRS}")

  string(REPLACE " " ";" KOKKOS_TPL_LIBRARY_NAMES "${KOKKOS_GMAKE_TPL_LIBRARY_NAMES}")

endif

CXXFLAGS=$(CCFLAGS)

# Options: Cuda,ROCm,OpenMP,Pthreads,Qthreads,Serial

# Options: Cuda,ROCm,OpenMP,Pthread,Qthreads,Serial

KOKKOS_DEVICES ?= "OpenMP"

#KOKKOS_DEVICES ?= "Pthreads"

#KOKKOS_DEVICES ?= "Pthread"

# Options: 

# Intel:    KNC,KNL,SNB,HSW,BDW,SKX

# NVIDIA:   Kepler,Kepler30,Kepler32,Kepler35,Kepler37,Maxwell,Maxwell50,Maxwell52,Maxwell53,Pascal60,Pascal61,Volta70,Volta72

KOKKOS_DEBUG ?= "no"

# Options: hwloc,librt,experimental_memkind

KOKKOS_USE_TPLS ?= ""

# Options: c++11,c++1z

# Options: c++11,c++14,c++1y,c++17,c++1z,c++2a

KOKKOS_CXX_STANDARD ?= "c++11"

# Options: aggressive_vectorization,disable_profiling,disable_deprecated_code,enable_large_mem_tests

KOKKOS_OPTIONS ?= ""

# Option for setting ETI path

KOKKOS_ETI_PATH ?= ${KOKKOS_PATH}/core/src/eti

KOKKOS_CMAKE ?= "no"

# Default settings specific options.

# Options: force_uvm,use_ldg,rdc,enable_lambda

# Check for general settings.

KOKKOS_INTERNAL_ENABLE_DEBUG := $(call kokkos_has_string,$(KOKKOS_DEBUG),yes)

KOKKOS_INTERNAL_ENABLE_CXX11 := $(call kokkos_has_string,$(KOKKOS_CXX_STANDARD),c++11)

KOKKOS_INTERNAL_ENABLE_CXX14 := $(call kokkos_has_string,$(KOKKOS_CXX_STANDARD),c++14)

KOKKOS_INTERNAL_ENABLE_CXX1Y := $(call kokkos_has_string,$(KOKKOS_CXX_STANDARD),c++1y)

KOKKOS_INTERNAL_ENABLE_CXX17 := $(call kokkos_has_string,$(KOKKOS_CXX_STANDARD),c++17)

KOKKOS_INTERNAL_ENABLE_CXX1Z := $(call kokkos_has_string,$(KOKKOS_CXX_STANDARD),c++1z)

KOKKOS_INTERNAL_ENABLE_CXX2A := $(call kokkos_has_string,$(KOKKOS_CXX_STANDARD),c++2a)

# Check for external libraries.

KOKKOS_INTERNAL_USE_HWLOC := $(call kokkos_has_string,$(KOKKOS_USE_TPLS),hwloc)

KOKKOS_INTERNAL_COMPILER_APPLE_CLANG := $(call kokkos_has_string,$(KOKKOS_CXX_VERSION),apple-darwin)

KOKKOS_INTERNAL_COMPILER_HCC         := $(call kokkos_has_string,$(KOKKOS_CXX_VERSION),HCC)

# Check Host Compiler if using NVCC through nvcc_wrapper

ifeq ($(KOKKOS_INTERNAL_COMPILER_NVCC), 1)

  KOKKOS_INTERNAL_COMPILER_NVCC_WRAPPER := $(strip $(shell echo $(CXX) | grep nvcc_wrapper | wc -l))

  ifeq ($(KOKKOS_INTERNAL_COMPILER_NVCC_WRAPPER), 1)

    KOKKOS_CXX_HOST_VERSION             := $(strip $(shell $(CXX) $(CXXFLAGS) --host-version       2>&1))

    KOKKOS_INTERNAL_COMPILER_PGI    := $(call kokkos_has_string,$(KOKKOS_CXX_HOST_VERSION),PGI)

    KOKKOS_INTERNAL_COMPILER_INTEL  := $(call kokkos_has_string,$(KOKKOS_CXX_HOST_VERSION),Intel Corporation)

    KOKKOS_INTERNAL_COMPILER_CLANG  := $(call kokkos_has_string,$(KOKKOS_CXX_HOST_VERSION),clang)

  endif

endif

ifeq ($(KOKKOS_INTERNAL_COMPILER_CLANG), 2)

  KOKKOS_INTERNAL_COMPILER_CLANG = 1

endif

# Set C++11 flags.

ifeq ($(KOKKOS_INTERNAL_COMPILER_PGI), 1)

  KOKKOS_INTERNAL_CXX11_FLAG := --c++11

  KOKKOS_INTERNAL_CXX14_FLAG := --c++14

  #KOKKOS_INTERNAL_CXX17_FLAG := --c++17

else

  ifeq ($(KOKKOS_INTERNAL_COMPILER_XL), 1)

     KOKKOS_INTERNAL_CXX11_FLAG := -std=c++11

     #KOKKOS_INTERNAL_CXX14_FLAG := -std=c++14

     KOKKOS_INTERNAL_CXX1Y_FLAG := -std=c++1y

     #KOKKOS_INTERNAL_CXX17_FLAG := -std=c++17

     #KOKKOS_INTERNAL_CXX1Z_FLAG := -std=c++1Z

     #KOKKOS_INTERNAL_CXX2A_FLAG := -std=c++2a

  else

    ifeq ($(KOKKOS_INTERNAL_COMPILER_CRAY), 1)

      KOKKOS_INTERNAL_CXX11_FLAG := -hstd=c++11

      KOKKOS_INTERNAL_CXX14_FLAG := -hstd=c++14

      #KOKKOS_INTERNAL_CXX1Y_FLAG := -hstd=c++1y

      #KOKKOS_INTERNAL_CXX17_FLAG := -hstd=c++17

      #KOKKOS_INTERNAL_CXX1Z_FLAG := -hstd=c++1z

      #KOKKOS_INTERNAL_CXX2A_FLAG := -hstd=c++2a

    else

      ifeq ($(KOKKOS_INTERNAL_COMPILER_HCC), 1)

        KOKKOS_INTERNAL_CXX11_FLAG := 

      else

        KOKKOS_INTERNAL_CXX11_FLAG := --std=c++11

        KOKKOS_INTERNAL_CXX14_FLAG := --std=c++14

        KOKKOS_INTERNAL_CXX1Y_FLAG := --std=c++1y

        KOKKOS_INTERNAL_CXX17_FLAG := --std=c++17

        KOKKOS_INTERNAL_CXX1Z_FLAG := --std=c++1z

        KOKKOS_INTERNAL_CXX2A_FLAG := --std=c++2a

      endif

    endif

  endif

#CPPFLAGS is now unused

KOKKOS_CPPFLAGS =

ifneq ($(KOKKOS_CMAKE), yes)

  KOKKOS_CXXFLAGS = -I./ -I$(KOKKOS_PATH)/core/src -I$(KOKKOS_PATH)/containers/src -I$(KOKKOS_PATH)/algorithms/src -I$(KOKKOS_ETI_PATH)

endif

KOKKOS_TPL_INCLUDE_DIRS =

KOKKOS_TPL_LIBRARY_DIRS =

KOKKOS_TPL_LIBRARY_NAMES =

KOKKOS_LIBS = -ldl

KOKKOS_TPL_LIBRARY_NAMES += dl

ifneq ($(KOKKOS_CMAKE), yes)

  KOKKOS_LDFLAGS = -L$(shell pwd)

  # CXXLDFLAGS is used together with CXXFLAGS in a combined compile/link command

  KOKKOS_CXXLDFLAGS = -L$(shell pwd)

endif

KOKKOS_LINK_FLAGS = 

KOKKOS_SRC =

KOKKOS_HEADERS =

ifeq ($(KOKKOS_INTERNAL_USE_CUDA), 1)

  tmp := $(call kokkos_append_header,"\#define KOKKOS_ENABLE_CUDA")

  tmp := $(call kokkos_append_header,"\#define KOKKOS_COMPILER_CUDA_VERSION $(KOKKOS_INTERNAL_COMPILER_NVCC_VERSION)")

endif

ifeq ($(KOKKOS_INTERNAL_USE_ROCM), 1)

  tmp := $(call kokkos_append_header,'\#define KOKKOS_ENABLE_ROCM')

  tmp := $(call kokkos_append_header,'\#define KOKKOS_IMPL_ROCM_CLANG_WORKAROUND 1')

endif

ifeq ($(KOKKOS_INTERNAL_USE_OPENMPTARGET), 1)

  KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_CXX11_FLAG)

  tmp := $(call kokkos_append_header,"\#define KOKKOS_ENABLE_CXX11")

endif

ifeq ($(KOKKOS_INTERNAL_ENABLE_CXX14), 1)

  KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_CXX14_FLAG)

  tmp := $(call kokkos_append_header,"\#define KOKKOS_ENABLE_CXX14")

endif

ifeq ($(KOKKOS_INTERNAL_ENABLE_CXX1Y), 1)

  KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_CXX1Y_FLAG)

  tmp := $(call kokkos_append_header,"\#define KOKKOS_ENABLE_CXX14")

endif

ifeq ($(KOKKOS_INTERNAL_ENABLE_CXX17), 1)

  KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_CXX17_FLAG)

  tmp := $(call kokkos_append_header,"\#define KOKKOS_ENABLE_CXX17")

endif

ifeq ($(KOKKOS_INTERNAL_ENABLE_CXX1Z), 1)

  KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_CXX1Z_FLAG)

  tmp := $(call kokkos_append_header,"\#define KOKKOS_ENABLE_CXX11")

  tmp := $(call kokkos_append_header,"\#define KOKKOS_ENABLE_CXX1Z")

  tmp := $(call kokkos_append_header,"\#define KOKKOS_ENABLE_CXX17")

endif

ifeq ($(KOKKOS_INTERNAL_ENABLE_CXX2A), 1)

  KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_CXX2A_FLAG)

  tmp := $(call kokkos_append_header,"\#define KOKKOS_ENABLE_CXX20")

endif

ifeq ($(KOKKOS_INTERNAL_ENABLE_DEBUG), 1)

ifeq ($(KOKKOS_INTERNAL_USE_HWLOC), 1)

  ifneq ($(HWLOC_PATH),)

    ifneq ($(KOKKOS_CMAKE), yes)

      KOKKOS_CXXFLAGS += -I$(HWLOC_PATH)/include

    endif

    KOKKOS_LDFLAGS += -L$(HWLOC_PATH)/lib

    KOKKOS_CXXLDFLAGS += -L$(HWLOC_PATH)/lib

    KOKKOS_TPL_INCLUDE_DIRS += $(HWLOC_PATH)/include

ifeq ($(KOKKOS_INTERNAL_USE_MEMKIND), 1)

  ifneq ($(MEMKIND_PATH),)

    ifneq ($(KOKKOS_CMAKE), yes)

      KOKKOS_CXXFLAGS += -I$(MEMKIND_PATH)/include

    endif

    KOKKOS_LDFLAGS += -L$(MEMKIND_PATH)/lib

    KOKKOS_CXXLDFLAGS += -L$(MEMKIND_PATH)/lib

    KOKKOS_TPL_INCLUDE_DIRS += $(MEMKIND_PATH)/include

endif

  KOKKOS_HEADERS += $(wildcard $(KOKKOS_PATH)/core/src/Cuda/*.hpp)

  ifneq ($(CUDA_PATH),)

    ifneq ($(KOKKOS_CMAKE), yes)

      KOKKOS_CXXFLAGS += -I$(CUDA_PATH)/include

    endif

    KOKKOS_LDFLAGS += -L$(CUDA_PATH)/lib64

    KOKKOS_CXXLDFLAGS += -L$(CUDA_PATH)/lib64

    KOKKOS_TPL_INCLUDE_DIRS += $(CUDA_PATH)/include

  KOKKOS_SRC += $(wildcard $(KOKKOS_PATH)/core/src/Qthreads/*.cpp)

  KOKKOS_HEADERS += $(wildcard $(KOKKOS_PATH)/core/src/Qthreads/*.hpp)

  ifneq ($(QTHREADS_PATH),)

    ifneq ($(KOKKOS_CMAKE), yes)

      KOKKOS_CXXFLAGS += -I$(QTHREADS_PATH)/include

    endif

    KOKKOS_LDFLAGS += -L$(QTHREADS_PATH)/lib

    KOKKOS_CXXLDFLAGS += -L$(QTHREADS_PATH)/lib

    KOKKOS_TPL_INCLUDE_DIRS += $(QTHREADS_PATH)/include

  * GCC 4.8.4

  * GCC 4.9.3

  * GCC 5.1.0

  * GCC 5.3.0

  * GCC 5.5.0

  * GCC 6.1.0

  * GCC 7.2.0

  * GCC 7.3.0

  * GCC 8.1.0

  * Intel 15.0.2

  * Intel 16.0.1

  * Intel 17.1.043

  * Intel 17.0.1

  * Intel 17.4.196

  * Intel 18.0.128

  * Intel 18.2.128

  * Clang 3.6.1

  * Clang 3.7.1

  * Clang 3.8.1

  * Clang 3.9.0

  * Clang 4.0.0

  * Clang 4.0.0 for CUDA (CUDA Toolkit 8.0.44)

  * Clang 6.0.0 for CUDA (CUDA Toolkit 9.1)

  * PGI 17.10

  * NVCC 7.0 for CUDA (with gcc 4.8.4)

  * Clang 6.0.0 for CUDA (CUDA Toolkit 9.0)

  * Clang 7.0.0 for CUDA (CUDA Toolkit 9.1)

  * PGI 18.7

  * NVCC 7.5 for CUDA (with gcc 4.8.4)

  * NVCC 8.0.44 for CUDA (with gcc 5.3.0)

  * NVCC 9.1 for CUDA (with gcc 6.1.0)

### Primary tested compilers on Power 8 are:

  * GCC 5.4.0 (OpenMP,Serial)

  * IBM XL 13.1.6 (OpenMP, Serial)

  * NVCC 8.0.44 for CUDA (with gcc 5.4.0)

  * NVCC 9.0.103 for CUDA (with gcc 6.3.0 and XL 13.1.6)

  * GCC 6.4.0 (OpenMP,Serial)

  * GCC 7.2.0 (OpenMP,Serial)

  * IBM XL 16.1.0 (OpenMP, Serial)

  * NVCC 9.2.88 for CUDA (with gcc 7.2.0 and XL 16.1.0)

### Primary tested compilers on Intel KNL are:

  * GCC 6.2.0

  * Intel 16.4.258 (with gcc 4.7.2)

  * Intel 17.2.174 (with gcc 4.9.3)

  * Intel 18.0.128 (with gcc 4.9.3)

  * Intel 18.2.199 (with gcc 4.9.3)

### Primary tested compilers on ARM

  * GCC 6.1.0 

### Primary tested compilers on ARM (Cavium ThunderX2)

  * GCC 7.2.0 

  * ARM/Clang 18.4.0

### Other compilers working:

  * X86:

   - Cygwin 2.1.0 64bit with gcc 4.9.3

   - GCC 8.1.0 (not warning free)

### Known non-working combinations:

  * Power8:

    typedef Random_XorShift64<DeviceType> generator_type;

    typedef DeviceType device_type;

    KOKKOS_INLINE_FUNCTION

    Random_XorShift64_Pool() {

      num_states_ = 0;

    }

#endif

    }

    KOKKOS_INLINE_FUNCTION

    Random_XorShift64_Pool(const Random_XorShift64_Pool& src):

      locks_(src.locks_),

      state_(src.state_),

      num_states_(src.num_states_)

    {}

    KOKKOS_INLINE_FUNCTION

    Random_XorShift64_Pool operator = (const Random_XorShift64_Pool& src) {

      locks_ = src.locks_;

      state_ = src.state_;

    typedef DeviceType device_type;

    KOKKOS_INLINE_FUNCTION

    Random_XorShift1024_Pool() {

      num_states_ = 0;

    }

#endif

    }

    KOKKOS_INLINE_FUNCTION

    Random_XorShift1024_Pool(const Random_XorShift1024_Pool& src):

      locks_(src.locks_),

      state_(src.state_),

      num_states_(src.num_states_)

    {}

    KOKKOS_INLINE_FUNCTION

    Random_XorShift1024_Pool operator = (const Random_XorShift1024_Pool& src) {

      locks_ = src.locks_;

      state_ = src.state_;