[Quad] Add fabsq (#375)

Co-authored-by: shibatch <[email protected]>

CMakeLists.txt

@@ -5,7 +5,7 @@ option(BUILD_SHARED_LIBS "Build shared libs" ON)
 option(BUILD_STATIC_TEST_BINS "Build statically linked test executables" OFF)
 option(ENABLE_LTO "Enable LTO on GCC or ThinLTO on clang" OFF)
 option(BUILD_LIBM "libsleef will be built." ON)
-option(BUILD_DFT "libsleefdft will be built." ON)
+option(BUILD_DFT "libsleefdft will be built." OFF)
 option(BUILD_QUAD "libsleefquad will be built." OFF)
 option(BUILD_GNUABI_LIBS "libsleefgnuabi will be built." ON)
 option(BUILD_TESTS "Tests will be built." ON)

Configure.cmake

@@ -85,6 +85,7 @@ if((CMAKE_SYSTEM_PROCESSOR MATCHES "x86") OR (CMAKE_SYSTEM_PROCESSOR MATCHES "AM
   set(SLEEF_ARCH_X86 ON CACHE INTERNAL "True for x86 architecture.")
 
   set(CLANG_FLAGS_ENABLE_PURECFMA_SCALAR "-mavx2;-mfma;-fno-strict-aliasing")
+  set(CLANG_FLAGS_ENABLE_PUREC_SCALAR "-fno-strict-aliasing")
 
 elseif(CMAKE_SYSTEM_PROCESSOR MATCHES "aarch64|arm64")
   set(SLEEF_ARCH_AARCH64 ON CACHE INTERNAL "True for Aarch64 architecture.")
@@ -98,16 +99,19 @@ elseif(CMAKE_SYSTEM_PROCESSOR MATCHES "arm")
   set(COMPILER_SUPPORTS_NEON32VFPV4 1)
 
   set(CLANG_FLAGS_ENABLE_PURECFMA_SCALAR "-mfpu=vfpv4;-fno-strict-aliasing")
+  set(CLANG_FLAGS_ENABLE_PUREC_SCALAR "-fno-strict-aliasing")
 
 elseif(CMAKE_SYSTEM_PROCESSOR MATCHES "^(powerpc|ppc)64")
   set(SLEEF_ARCH_PPC64 ON CACHE INTERNAL "True for PPC64 architecture.")
 
   set(CLANG_FLAGS_ENABLE_PURECFMA_SCALAR "-mvsx;-fno-strict-aliasing")
+  set(CLANG_FLAGS_ENABLE_PUREC_SCALAR "-fno-strict-aliasing")
 
 elseif(CMAKE_SYSTEM_PROCESSOR MATCHES "s390x")
   set(SLEEF_ARCH_S390X ON CACHE INTERNAL "True for IBM Z architecture.")
 
   set(CLANG_FLAGS_ENABLE_PURECFMA_SCALAR "-march=z14;-mzvector;-fno-strict-aliasing")
+  set(CLANG_FLAGS_ENABLE_PUREC_SCALAR "-fno-strict-aliasing")
 endif()
 
 set(COMPILER_SUPPORTS_PUREC_SCALAR 1)

Jenkinsfile

@@ -14,7 +14,7 @@ pipeline {
 			 rm -rf build
  			 mkdir build
 			 cd build
-			 cmake -GNinja -DCMAKE_INSTALL_PREFIX=../install -DSLEEF_SHOW_CONFIG=1 -DENFORCE_TESTER3=TRUE -DBUILD_INLINE_HEADERS=TRUE -DBUILD_QUAD=TRUE -DENFORCE_SVE=TRUE -DEMULATOR=qemu-aarch64 ..
+			 cmake -GNinja -DCMAKE_INSTALL_PREFIX=../install -DSLEEF_SHOW_CONFIG=1 -DENFORCE_TESTER3=TRUE -DBUILD_INLINE_HEADERS=TRUE -DBUILD_QUAD=TRUE -DBUILD_DFT=TRUE -DENFORCE_SVE=TRUE -DEMULATOR=qemu-aarch64 ..
 			 ninja
 			 export OMP_WAIT_POLICY=passive
 		         export CTEST_OUTPUT_ON_FAILURE=TRUE
@@ -34,7 +34,7 @@ pipeline {
 			 rm -rf build
  			 mkdir build
 			 cd build
-			 cmake -GNinja -DCMAKE_INSTALL_PREFIX=../install -DSLEEF_SHOW_CONFIG=1 -DENFORCE_TESTER3=TRUE -DBUILD_INLINE_HEADERS=TRUE -DFORCE_AAVPCS=On -DENABLE_GNUABI=On -DBUILD_QUAD=TRUE -DENFORCE_SVE=TRUE -DEMULATOR=qemu-aarch64 ..
+			 cmake -GNinja -DCMAKE_INSTALL_PREFIX=../install -DSLEEF_SHOW_CONFIG=1 -DENFORCE_TESTER3=TRUE -DBUILD_INLINE_HEADERS=TRUE -DFORCE_AAVPCS=On -DENABLE_GNUABI=On -DBUILD_QUAD=TRUE -DBUILD_DFT=TRUE -DENFORCE_SVE=TRUE -DEMULATOR=qemu-aarch64 ..
 			 ninja
 			 export OMP_WAIT_POLICY=passive
 		         export CTEST_OUTPUT_ON_FAILURE=TRUE
@@ -53,7 +53,7 @@ pipeline {
 			 rm -rf build
  			 mkdir build
 			 cd build
-			 cmake -GNinja -DCMAKE_INSTALL_PREFIX=../install -DSLEEF_SHOW_CONFIG=1 -DBUILD_SHARED_LIBS=FALSE -DENFORCE_TESTER3=TRUE -DFORCE_AAVPCS=On -DENABLE_GNUABI=On -DBUILD_QUAD=TRUE -DDISABLE_SVE=TRUE -DBUILD_INLINE_HEADERS=TRUE -DENABLE_CUDA=TRUE -DENFORCE_CUDA=TRUE ..
+			 cmake -GNinja -DCMAKE_INSTALL_PREFIX=../install -DSLEEF_SHOW_CONFIG=1 -DBUILD_SHARED_LIBS=FALSE -DENFORCE_TESTER3=TRUE -DFORCE_AAVPCS=On -DENABLE_GNUABI=On -DBUILD_QUAD=TRUE -DBUILD_DFT=TRUE -DDISABLE_SVE=TRUE -DBUILD_INLINE_HEADERS=TRUE -DENABLE_CUDA=TRUE -DENFORCE_CUDA=TRUE ..
 			 ninja
 			 export OMP_WAIT_POLICY=passive
 		         export CTEST_OUTPUT_ON_FAILURE=TRUE
@@ -78,7 +78,7 @@ pipeline {
 			 docker exec jenkins tar xf /tmp/builddir.tgz -C /build
 			 docker exec jenkins rm -f /tmp/builddir.tgz
 			 rm -f /tmp/builddir.tgz
-			 docker exec jenkins bash -c "set -ev;export OMP_WAIT_POLICY=passive;cd /build;rm -rf build;mkdir build;cd build;export CC=gcc;export PATH=/opt/bin:$PATH;cmake -GNinja -DBUILD_QUAD=TRUE -DENFORCE_TESTER3=TRUE ..;ninja;ctest -j `nproc`"
+			 docker exec jenkins bash -c "set -ev;export OMP_WAIT_POLICY=passive;cd /build;rm -rf build;mkdir build;cd build;export CC=gcc;export PATH=/opt/bin:$PATH;cmake -GNinja -DBUILD_QUAD=TRUE -DBUILD_DFT=TRUE -DENFORCE_TESTER3=TRUE ..;ninja;ctest -j `nproc`"
 			 docker stop jenkins
 			 '''
             	     }
@@ -99,7 +99,7 @@ pipeline {
 			 docker exec jenkins tar xf /tmp/builddir.tgz -C /build
 			 docker exec jenkins rm -f /tmp/builddir.tgz
 			 rm -f /tmp/builddir.tgz
-			 docker exec jenkins bash -c "set -ev;export OMP_WAIT_POLICY=passive;cd /build;rm -rf build;mkdir build;cd build;export CC=clang;export PATH=/opt/bin:$PATH;cmake -GNinja -DBUILD_QUAD=TRUE -DENFORCE_TESTER3=TRUE ..;ninja;ctest -j `nproc`"
+			 docker exec jenkins bash -c "set -ev;export OMP_WAIT_POLICY=passive;cd /build;rm -rf build;mkdir build;cd build;export CC=clang;export PATH=/opt/bin:$PATH;cmake -GNinja -DBUILD_QUAD=TRUE -DBUILD_DFT=TRUE -DENFORCE_TESTER3=TRUE ..;ninja;ctest -j `nproc`"
 			 docker stop jenkins
 			 '''
             	     }
@@ -114,7 +114,7 @@ pipeline {
 			 rm -rf build
  			 mkdir build
 			 cd build
-			 cmake -GNinja -DCMAKE_INSTALL_PREFIX=../install -DSLEEF_SHOW_CONFIG=1 -DENFORCE_TESTER3=TRUE -DBUILD_QUAD=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_AVX2=TRUE -DENFORCE_AVX512F=TRUE ..
+			 cmake -GNinja -DCMAKE_INSTALL_PREFIX=../install -DSLEEF_SHOW_CONFIG=1 -DENFORCE_TESTER3=TRUE -DBUILD_QUAD=TRUE -DBUILD_DFT=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_AVX2=TRUE -DENFORCE_AVX512F=TRUE ..
 			 ninja
 			 export OMP_WAIT_POLICY=passive
 		         export CTEST_OUTPUT_ON_FAILURE=TRUE
@@ -134,7 +134,7 @@ pipeline {
 			 rm -rf build
  			 mkdir build
 			 cd build
-			 cmake -GNinja -DCMAKE_INSTALL_PREFIX=../install -DSLEEF_SHOW_CONFIG=1 -DOPENSSL_ROOT_DIR=/usr/local/opt/openssl -DENFORCE_TESTER3=TRUE -DBUILD_INLINE_HEADERS=TRUE -DBUILD_QUAD=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_FMA4=TRUE -DENFORCE_AVX2=TRUE -DENFORCE_AVX512F=TRUE ..
+			 cmake -GNinja -DCMAKE_INSTALL_PREFIX=../install -DSLEEF_SHOW_CONFIG=1 -DOPENSSL_ROOT_DIR=/usr/local/opt/openssl -DENFORCE_TESTER3=TRUE -DBUILD_INLINE_HEADERS=TRUE -DBUILD_QUAD=TRUE -DBUILD_DFT=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_FMA4=TRUE -DENFORCE_AVX2=TRUE -DENFORCE_AVX512F=TRUE ..
 			 ninja
 			 export OMP_WAIT_POLICY=passive
 		         export CTEST_OUTPUT_ON_FAILURE=TRUE
@@ -154,7 +154,7 @@ pipeline {
 			 rm -rf build
  			 mkdir build
 			 cd build
-			 cmake -GNinja -DCMAKE_INSTALL_PREFIX=../install -DSLEEF_SHOW_CONFIG=1 -DBUILD_SHARED_LIBS=FALSE -DOPENSSL_ROOT_DIR=/usr/local/opt/openssl -DENFORCE_TESTER3=TRUE -DBUILD_INLINE_HEADERS=TRUE -DBUILD_QUAD=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_FMA4=TRUE -DENFORCE_AVX2=TRUE -DENFORCE_AVX512F=TRUE ..
+			 cmake -GNinja -DCMAKE_INSTALL_PREFIX=../install -DSLEEF_SHOW_CONFIG=1 -DBUILD_SHARED_LIBS=FALSE -DOPENSSL_ROOT_DIR=/usr/local/opt/openssl -DENFORCE_TESTER3=TRUE -DBUILD_INLINE_HEADERS=TRUE -DBUILD_QUAD=TRUE -DBUILD_DFT=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_FMA4=TRUE -DENFORCE_AVX2=TRUE -DENFORCE_AVX512F=TRUE ..
 			 ninja
 			 export OMP_WAIT_POLICY=passive
 		         export CTEST_OUTPUT_ON_FAILURE=TRUE
@@ -172,7 +172,7 @@ pipeline {
 			 rm -rf build
  			 mkdir build
 			 cd build
-			 cmake -GNinja -DCMAKE_INSTALL_PREFIX=../install -DSLEEF_SHOW_CONFIG=1 -DENFORCE_TESTER3=TRUE -DBUILD_INLINE_HEADERS=TRUE -DBUILD_QUAD=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_FMA4=TRUE -DENFORCE_AVX2=TRUE -DENFORCE_AVX512F=TRUE ..
+			 cmake -GNinja -DCMAKE_INSTALL_PREFIX=../install -DSLEEF_SHOW_CONFIG=1 -DENFORCE_TESTER3=TRUE -DBUILD_INLINE_HEADERS=TRUE -DBUILD_QUAD=TRUE -DBUILD_DFT=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_FMA4=TRUE -DENFORCE_AVX2=TRUE -DENFORCE_AVX512F=TRUE ..
 			 ninja
 			 export OMP_WAIT_POLICY=passive
 		         export CTEST_OUTPUT_ON_FAILURE=TRUE
@@ -192,7 +192,7 @@ pipeline {
 			 rm -rf build
  			 mkdir build
 			 cd build
-			 cmake -GNinja -DCMAKE_INSTALL_PREFIX=../install -DSLEEF_SHOW_CONFIG=1 -DENFORCE_TESTER3=TRUE -DBUILD_QUAD=TRUE -DBUILD_INLINE_HEADERS=TRUE -DBUILD_SHARED_LIBS=FALSE -DENABLE_LTO=TRUE -DDISABLE_FMA4=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_AVX2=TRUE -DENFORCE_AVX512F=TRUE ..
+			 cmake -GNinja -DCMAKE_INSTALL_PREFIX=../install -DSLEEF_SHOW_CONFIG=1 -DENFORCE_TESTER3=TRUE -DBUILD_QUAD=TRUE -DBUILD_DFT=TRUE -DBUILD_INLINE_HEADERS=TRUE -DBUILD_SHARED_LIBS=FALSE -DENABLE_LTO=TRUE -DDISABLE_FMA4=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_AVX2=TRUE -DENFORCE_AVX512F=TRUE ..
 			 ninja
 			 export OMP_WAIT_POLICY=passive
 		         export CTEST_OUTPUT_ON_FAILURE=TRUE
@@ -212,7 +212,7 @@ pipeline {
 			 rm -rf build
  			 mkdir build
 			 cd build
-			 cmake -GNinja -DCMAKE_INSTALL_PREFIX=../install -DSLEEF_SHOW_CONFIG=1 -DENFORCE_TESTER3=TRUE -DBUILD_QUAD=TRUE -DBUILD_INLINE_HEADERS=TRUE -DBUILD_SHARED_LIBS=FALSE -DENABLE_LTO=TRUE -DLLVM_AR_COMMAND=llvm-ar-10 -DDISABLE_FMA4=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_AVX2=TRUE -DENFORCE_AVX512F=TRUE ..
+			 cmake -GNinja -DCMAKE_INSTALL_PREFIX=../install -DSLEEF_SHOW_CONFIG=1 -DENFORCE_TESTER3=TRUE -DBUILD_QUAD=TRUE -DBUILD_DFT=TRUE -DBUILD_INLINE_HEADERS=TRUE -DBUILD_SHARED_LIBS=FALSE -DENABLE_LTO=TRUE -DLLVM_AR_COMMAND=llvm-ar-10 -DDISABLE_FMA4=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_AVX2=TRUE -DENFORCE_AVX512F=TRUE ..
 			 ninja
 			 export OMP_WAIT_POLICY=passive
 		         export CTEST_OUTPUT_ON_FAILURE=TRUE
@@ -237,7 +237,7 @@ pipeline {
 			 docker exec jenkins tar xf /tmp/builddir.tgz -C /build
 			 docker exec jenkins rm -f /tmp/builddir.tgz
 			 rm -f /tmp/builddir.tgz
-			 docker exec jenkins bash -c "set -ev;export OMP_WAIT_POLICY=passive;cd /build;rm -rf build;mkdir build;cd build;export CC=gcc;cmake -GNinja -DBUILD_QUAD=TRUE -DBUILD_INLINE_HEADERS=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_FMA4=TRUE -DENFORCE_AVX2=TRUE -DENFORCE_AVX512F=TRUE ..;ninja;ctest -j `nproc`"
+			 docker exec jenkins bash -c "set -ev;export OMP_WAIT_POLICY=passive;cd /build;rm -rf build;mkdir build;cd build;export CC=gcc;cmake -GNinja -DBUILD_QUAD=TRUE -DBUILD_DFT=TRUE -DBUILD_INLINE_HEADERS=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_FMA4=TRUE -DENFORCE_AVX2=TRUE -DENFORCE_AVX512F=TRUE ..;ninja;ctest -j `nproc`"
 			 docker stop jenkins
 			 '''
             	     }
@@ -258,7 +258,7 @@ pipeline {
 			 docker exec jenkins tar xf /tmp/builddir.tgz -C /build
 			 docker exec jenkins rm -f /tmp/builddir.tgz
 			 rm -f /tmp/builddir.tgz
-			 docker exec jenkins bash -c "set -ev;export OMP_WAIT_POLICY=passive;cd /build;rm -rf build;mkdir build;cd build;export CC=clang;cmake -GNinja -DBUILD_QUAD=TRUE -DBUILD_INLINE_HEADERS=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_FMA4=TRUE -DENFORCE_AVX2=TRUE -DENFORCE_AVX512F=TRUE ..;ninja;ctest -j `nproc`"
+			 docker exec jenkins bash -c "set -ev;export OMP_WAIT_POLICY=passive;cd /build;rm -rf build;mkdir build;cd build;export CC=clang;cmake -GNinja -DBUILD_QUAD=TRUE -DBUILD_DFT=TRUE -DBUILD_INLINE_HEADERS=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_FMA4=TRUE -DENFORCE_AVX2=TRUE -DENFORCE_AVX512F=TRUE ..;ninja;ctest -j `nproc`"
 			 docker stop jenkins
 			 '''
             	     }
@@ -281,7 +281,7 @@ pipeline {
 			 rm -rf build
  			 mkdir build
 			 cd build
-			 /usr/local/bin/cmake -GNinja -DBUILD_QUAD=TRUE -DBUILD_INLINE_HEADERS=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_FMA4=TRUE -DENFORCE_AVX2=TRUE ..
+			 /usr/local/bin/cmake -GNinja -DBUILD_QUAD=TRUE -DBUILD_DFT=TRUE -DBUILD_INLINE_HEADERS=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_FMA4=TRUE -DENFORCE_AVX2=TRUE ..
 			 ninja
 			 '''
             	     }
@@ -296,9 +296,9 @@ pipeline {
 			 rm -rf build
  			 mkdir build
 			 cd build
-			 cmake -GNinja -DBUILD_QUAD=TRUE -DBUILD_INLINE_HEADERS=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_FMA4=TRUE -DENFORCE_AVX2=TRUE -DENFORCE_AVX512F=TRUE ..
+			 cmake -GNinja -DBUILD_QUAD=TRUE -DBUILD_DFT=TRUE -DBUILD_INLINE_HEADERS=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_FMA4=TRUE -DENFORCE_AVX2=TRUE -DENFORCE_AVX512F=TRUE ..
 			 ninja
-			 cmake -GNinja -DBUILD_QUAD=TRUE -DBUILD_INLINE_HEADERS=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_FMA4=TRUE -DENFORCE_AVX2=TRUE -DENFORCE_AVX512F=TRUE ..
+			 cmake -GNinja -DBUILD_QUAD=TRUE -DBUILD_DFT=TRUE -DBUILD_INLINE_HEADERS=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_FMA4=TRUE -DENFORCE_AVX2=TRUE -DENFORCE_AVX512F=TRUE ..
 			 ninja
 		         ctest -j `nproc`
 			 '''

appveyor.yml

@@ -20,17 +20,17 @@ build_script:
   - if "%DO_TEST%" == "TRUE" echo PATH c:\Cygwin64\bin;c:\Cygwin64\usr\bin;%CD%\build-cygwin\bin;%PATH% > q.bat
   - if "%DO_TEST%" == "TRUE" powershell -Command "(gc q.bat) -replace ' ;', ';' | Out-File -encoding ASCII p.bat"
   - if "%DO_TEST%" == "TRUE" call p.bat
-  - if "%DO_TEST%" == "TRUE" "C:\\Cygwin64\\bin\\bash" -c 'mkdir build-mingw;cd build-mingw;CC=x86_64-w64-mingw32-gcc cmake -G Ninja .. -DRUNNING_ON_APPVEYOR=TRUE -DBUILD_SHARED_LIBS=FALSE -DBUILD_QUAD=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE;ninja'
+  - if "%DO_TEST%" == "TRUE" "C:\\Cygwin64\\bin\\bash" -c 'mkdir build-mingw;cd build-mingw;CC=x86_64-w64-mingw32-gcc cmake -G Ninja .. -DRUNNING_ON_APPVEYOR=TRUE -DBUILD_SHARED_LIBS=FALSE -DBUILD_QUAD=TRUE -DBUILD_DFT=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE;ninja'
   - if "%DO_TEST%" == "TRUE" cd "%BUILDFOLDER%"
-  - if "%DO_TEST%" == "TRUE" "C:\\Cygwin64\\bin\\bash" -c 'mkdir build-cygwin;cd build-cygwin;cmake -G Ninja -DRUNNING_ON_APPVEYOR=TRUE -DBUILD_QUAD=TRUE -DBUILD_INLINE_HEADERS=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE ..;ninja'
+  - if "%DO_TEST%" == "TRUE" "C:\\Cygwin64\\bin\\bash" -c 'mkdir build-cygwin;cd build-cygwin;cmake -G Ninja -DRUNNING_ON_APPVEYOR=TRUE -DBUILD_QUAD=TRUE -DBUILD_DFT=TRUE -DBUILD_INLINE_HEADERS=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE ..;ninja'
   - if "%DO_TEST%" == "TRUE" cd "%BUILDFOLDER%"
   - if "%DO_TEST%" == "TRUE" del /Q /F build-cygwin\bin\iut*
   - if "%DO_TEST%" == "TRUE" echo PATH %ORGPATH%;c:\Cygwin64\bin;c:\Cygwin64\usr\bin;%CD%\build-cygwin\bin;%CD%\build\bin > q.bat
   - if "%DO_TEST%" == "TRUE" powershell -Command "(gc q.bat) -replace ' ;', ';' | Out-File -encoding ASCII p.bat"
   - if "%DO_TEST%" == "TRUE" call p.bat
   - mkdir build
   - cd build
-  - cmake -G"Visual Studio 16 2019" .. -DRUNNING_ON_APPVEYOR=TRUE -DCMAKE_INSTALL_PREFIX=install -DSLEEF_SHOW_CONFIG=1 -DSLEEF_SHOW_ERROR_LOG=1 -DENFORCE_TESTER3=TRUE -DBUILD_QUAD=TRUE -DBUILD_INLINE_HEADERS=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_FMA4=TRUE -DENFORCE_AVX2=TRUE -DENFORCE_AVX512F=TRUE %ENV_BUILD_STATIC%
+  - cmake -G"Visual Studio 16 2019" .. -DRUNNING_ON_APPVEYOR=TRUE -DCMAKE_INSTALL_PREFIX=install -DSLEEF_SHOW_CONFIG=1 -DSLEEF_SHOW_ERROR_LOG=1 -DENFORCE_TESTER3=TRUE -DBUILD_QUAD=TRUE -DBUILD_DFT=TRUE -DBUILD_INLINE_HEADERS=TRUE -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_FMA4=TRUE -DENFORCE_AVX2=TRUE -DENFORCE_AVX512F=TRUE %ENV_BUILD_STATIC%
   - cmake --build . --target install --config Release
   - if "%DO_TEST%" == "TRUE" (ctest --output-on-failure -j 16 -C Release)
   - cd "%BUILDFOLDER%"
@@ -39,7 +39,7 @@ build_script:
   - call p.bat
   - mkdir build-clang
   - cd build-clang
-  - cmake -G Ninja .. -DRUNNING_ON_APPVEYOR=TRUE -DCMAKE_C_COMPILER:PATH="C:\Program Files\LLVM\bin\clang.exe" -DCMAKE_INSTALL_PREFIX=install -DSLEEF_SHOW_CONFIG=1 -DSLEEF_SHOW_ERROR_LOG=1 -DENFORCE_TESTER3=TRUE -DBUILD_INLINE_HEADERS=TRUE -DBUILD_QUAD=TRUE  -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_FMA4=TRUE -DENFORCE_AVX2=TRUE -DENFORCE_AVX512F=TRUE %ENV_BUILD_STATIC%
+  - cmake -G Ninja .. -DRUNNING_ON_APPVEYOR=TRUE -DCMAKE_C_COMPILER:PATH="C:\Program Files\LLVM\bin\clang.exe" -DCMAKE_INSTALL_PREFIX=install -DSLEEF_SHOW_CONFIG=1 -DSLEEF_SHOW_ERROR_LOG=1 -DENFORCE_TESTER3=TRUE -DBUILD_INLINE_HEADERS=TRUE -DBUILD_QUAD=TRUE -DBUILD_DFT=TRUE  -DENFORCE_SSE2=TRUE -DENFORCE_SSE4=TRUE -DENFORCE_AVX=TRUE -DENFORCE_FMA4=TRUE -DENFORCE_AVX2=TRUE -DENFORCE_AVX512F=TRUE %ENV_BUILD_STATIC%
   - ninja
 test_script:
   - if "%DO_TEST%" == "TRUE" (ctest --output-on-failure -j 16 -C Release)

src/common/commonfuncs.h

@@ -116,7 +116,7 @@ static INLINE CONST VECTOR_CC tdi_t tdisettdi_tdi_vd3_vi(vdouble3 v, vint i) {
 #if defined(ENABLE_MAIN)
 // Functions for debugging
 #include <stdio.h>
-#include <wchar.t>
+#include <wchar.h>
 
 static void printvmask(char *mes, vmask g) {
   uint64_t u[VECTLENDP];
@@ -163,7 +163,7 @@ static void printvint64(char *mes, vint64 vi) {
   uint64_t u[VECTLENDP*2];
   vstoreu_v_p_vd((double *)u, vreinterpret_vd_vm(vreinterpret_vm_vi64(vi)));
   printf("%s ", mes);
-  for(int i=0;i<VECTLENDP;i++) printf("%016x : ", (unsigned long)u[i]);
+  for(int i=0;i<VECTLENDP;i++) printf("%016lx : ", (unsigned long)u[i]);
   printf("\n");
 }
 
@@ -359,6 +359,11 @@ static INLINE CONST vmask vilogbk_vm_vd(vdouble d) {
 
 // vmask2 functions
 
+static INLINE CONST vopmask veq64_vo_vm2_vm2(vmask2 x, vmask2 y) {
+  return vand_vo_vo_vo(veq64_vo_vm_vm(vm2getx_vm_vm2(x), vm2getx_vm_vm2(y)),
+		       veq64_vo_vm_vm(vm2gety_vm_vm2(x), vm2gety_vm_vm2(y)));
+}
+
 static INLINE CONST vmask2 vsel_vm2_vo_vm2_vm2(vopmask o, vmask2 x, vmask2 y) {
   return vm2setxy_vm2_vm_vm(vsel_vm_vo64_vm_vm(o, vm2getx_vm_vm2(x), vm2getx_vm_vm2(y)), vsel_vm_vo64_vm_vm(o, vm2gety_vm_vm2(x), vm2gety_vm_vm2(y)));
 }

src/common/keywords.txt

@@ -471,8 +471,15 @@ vcast_vf128_tdx
 vcast_vf128_tdx_fast
 vcast_vf128_tdx_slow
 vcmp_vm_tdx_tdx
+vsignbit_vo_tdx
+visnan_vo_tdx
+veq_vo_tdx_tdx
+vneq_vo_tdx_tdx
 vge_vo_tdx_tdx
+vle_vo_tdx_tdx
 vgt_vo_tdx_tdx
+vlt_vo_tdx_tdx
+vcast_tdx_d
 vilogb_vm_tdx
 vneg_tdx_tdx
 vsel_tdx_vo64_tdx_tdx
@@ -635,6 +642,7 @@ vreinterpret_vi64_vm
 vreinterpret_vm_vi64
 vreinterpret_vm_vu64
 vreinterpret_vu64_vm
+veq64_vo_vm2_vm2
 vsel_vm2_vo_vm2_vm2
 vtruncate_vm_vd
 vugt64_vo_vm_vm

src/common/keywords_cuda.txt

@@ -469,8 +469,15 @@ vcast_vf128_tdx
 vcast_vf128_tdx_fast
 vcast_vf128_tdx_slow
 vcmp_vm_tdx_tdx
+vsignbit_vo_tdx
+visnan_vo_tdx
+veq_vo_tdx_tdx
+vneq_vo_tdx_tdx
 vge_vo_tdx_tdx
+vle_vo_tdx_tdx
 vgt_vo_tdx_tdx
+vlt_vo_tdx_tdx
+vcast_tdx_d
 vilogb_vm_tdx
 vneg_tdx_tdx
 vsel_tdx_vo64_tdx_tdx
@@ -633,6 +640,7 @@ vreinterpret_vi64_vm
 vreinterpret_vm_vi64
 vreinterpret_vm_vu64
 vreinterpret_vu64_vm
+veq64_vo_vm2_vm2
 vsel_vm2_vo_vm2_vm2
 vtruncate_vm_vd
 vugt64_vo_vm_vm

src/common/misc.h

@@ -50,6 +50,7 @@
 
 #define SLEEF_FLT_MIN 0x1p-126
 #define SLEEF_DBL_MIN 0x1p-1022
+#define SLEEF_INT_MAX 2147483647
 
 //
 

src/libm/sleefinline_header.h.org

@@ -9,7 +9,7 @@
 #endif
 
 #ifndef SLEEF_INLINE
-#define SLEEF_INLINE inline
+#define SLEEF_INLINE static inline
 #endif
 
 #ifndef SLEEF_CONST

src/libm/sleeflibm_header.h.org.in

@@ -131,6 +131,10 @@ typedef struct {
 #define Sleef_quad_DEFINED
 #if defined(__SIZEOF_FLOAT128__) || (defined(__linux__) && defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__))) || (defined(__PPC64__) && defined(__GNUC__) && !defined(__clang__) && __GNUC__ >= 8)
 typedef __float128 Sleef_quad;
+#define SLEEF_QUAD_C(x) (x ## Q)
+//#elif defined(__SIZEOF_LONG_DOUBLE__) && defined(__aarch64__)
+//typedef long double Sleef_quad;
+//#define SLEEF_QUAD_C(x) (x ## L)
 #else
 typedef struct { uint64_t x, y; } Sleef_quad;
 #endif

src/libm/sleefsimddp.c

@@ -285,7 +285,7 @@ EXPORT CONST VECTOR_CC vint xilogb(vdouble d) {
   vdouble e = vcast_vd_vi(vilogbk_vi_vd(vabs_vd_vd(d)));
   e = vsel_vd_vo_vd_vd(veq_vo_vd_vd(d, vcast_vd_d(0)), vcast_vd_d(SLEEF_FP_ILOGB0), e);
   e = vsel_vd_vo_vd_vd(visnan_vo_vd(d), vcast_vd_d(SLEEF_FP_ILOGBNAN), e);
-  e = vsel_vd_vo_vd_vd(visinf_vo_vd(d), vcast_vd_d(INT_MAX), e);
+  e = vsel_vd_vo_vd_vd(visinf_vo_vd(d), vcast_vd_d(SLEEF_INT_MAX), e);
   return vrint_vi_vd(e);
 }