Merge pull request #37 from MikaelSlevinsky/feat-adc

Feat tdc

Author: MikaelSlevinsky

.appveyor.yml

@@ -9,24 +9,38 @@ install:
   - set PATH=%PATH%;c:\tools\vcpkg
   - set PATH=%PATH%;C:\mingw-w64\x86_64-7.2.0-posix-seh-rt_v5-rev1\mingw64\bin\
   - set CC=gcc && set BLAS=OPENBLAS
-  - vcpkg install openblas:x64-windows fftw3:x64-windows
+  - vcpkg install openblas:x64-windows fftw3:x64-windows mpir:x64-windows mpfr:x64-windows
   - mklink c:\projects\fasttransforms\src\cblas.h c:\tools\vcpkg\installed\x64-windows\include\openblas\cblas.h
   - mklink c:\projects\fasttransforms\src\openblas_config.h c:\tools\vcpkg\installed\x64-windows\include\openblas\openblas_config.h
   - mklink c:\projects\fasttransforms\blas.dll c:\tools\vcpkg\installed\x64-windows\bin\openblas.dll
   - mklink c:\projects\fasttransforms\src\fftw3.h c:\tools\vcpkg\installed\x64-windows\include\fftw3.h
   - mklink c:\projects\fasttransforms\fftw3.dll c:\tools\vcpkg\installed\x64-windows\bin\fftw3.dll
+  - mklink c:\projects\fasttransforms\src\mpfr.h c:\tools\vcpkg\installed\x64-windows\include\mpfr.h
+  - mklink c:\projects\fasttransforms\src\gmp.h c:\tools\vcpkg\installed\x64-windows\include\gmp.h
+  - mklink c:\projects\fasttransforms\src\mpir.h c:\tools\vcpkg\installed\x64-windows\include\mpir.h
+  - mklink c:\projects\fasttransforms\mpir.dll c:\tools\vcpkg\installed\x64-windows\bin\mpir.dll
+  - mklink c:\projects\fasttransforms\mpfr.dll c:\tools\vcpkg\installed\x64-windows\bin\mpfr.dll
   - set PATH=c:\tools\vcpkg\installed\x64-windows\bin;%PATH%
 
+matrix:
+  allow_failures:
+    - image: Visual Studio 2015
+
 build_script:
   - gcc --version
   - mingw32-make lib
   - mingw32-make wintests
   - mingw32-make examples
 
 test_script:
+  - set OMP_NUM_THREADS=4
   - test_transforms.exe
   - test_rotations.exe
-  - set OMP_NUM_THREADS=4
+  - test_tridiagonal.exe
+  - test_hierarchical.exe
+  - test_banded.exe
+  - test_dprk.exe
+  - test_tdc.exe
   - test_drivers.exe 3 3 0
   - additiontheorem.exe
   - calculus.exe

.gitattributes

@@ -6,27 +6,27 @@ matrix:
     - os: osx
       osx_image: xcode10
       compiler: gcc-4.9
-      env: MATRIX_EVAL="brew update && brew install gcc@4.9 fftw doxygen && export CC=gcc-4.9 && export BLAS=APPLEBLAS"
+      env: MATRIX_EVAL="brew update && brew install gcc@4.9 fftw mpfr doxygen && export CC=gcc-4.9 && export BLAS=APPLEBLAS"
     # Recommended system requirements
     - os: osx
       osx_image: xcode10
       compiler: gcc-7
-      env: MATRIX_EVAL="brew update && brew install gcc@7 fftw doxygen && export CC=gcc-7 && export BLAS=APPLEBLAS"
+      env: MATRIX_EVAL="brew update && brew install gcc@7 fftw mpfr doxygen && export CC=gcc-7 && export BLAS=APPLEBLAS"
     # Minimum system requirements
     - os: linux
       compiler: gcc-4.9
       addons:
         apt:
           sources: ubuntu-toolchain-r-test
-          packages: ['gcc-4.9', 'libblas-dev', 'libopenblas-base', 'libfftw3-dev', 'doxygen']
+          packages: ['gcc-4.9', 'libblas-dev', 'libopenblas-base', 'libfftw3-dev', 'libmpfr-dev', 'doxygen']
       env: MATRIX_EVAL="export CC=gcc-4.9 && export BLAS=OPENBLAS && export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:."
     # Recommended system requirements
     - os: linux
       compiler: gcc-7
       addons:
         apt:
           sources: ubuntu-toolchain-r-test
-          packages: ['gcc-7', 'libblas-dev', 'libopenblas-base', 'libfftw3-dev', 'doxygen']
+          packages: ['gcc-7', 'libblas-dev', 'libopenblas-base', 'libfftw3-dev', 'libmpfr-dev', 'doxygen']
       env: MATRIX_EVAL="export CC=gcc-7 && export BLAS=OPENBLAS && export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:."
 
 before_install: eval "${MATRIX_EVAL}"
@@ -36,6 +36,11 @@ script:
   - make tests
   - ./test_transforms
   - ./test_rotations
+  - ./test_tridiagonal
+  - ./test_hierarchical
+  - ./test_banded
+  - ./test_dprk
+  - ./test_tdc
   - OMP_NUM_THREADS=4 ./test_drivers 3 3 0
   - OMP_NUM_THREADS=4 ./test_fftw 3 3 0
   - make examples

.travis.yml

@@ -1,6 +1,6 @@
 LIB = fasttransforms
 LIBDIR = .
-LIBFLAGS = -shared -lm -fPIC -fopenmp
+LIBFLAGS = -shared -lm -fPIC
 
 ifeq ($(OS), Windows_NT)
     UNAME := Windows
@@ -14,7 +14,7 @@ else
 	endif
 endif
 
-OBJ = src/transforms.c src/rotations.c src/permute.c src/drivers.c src/fftw.c
+OBJ = src/transforms.c src/rotations.c src/permute.c src/tdc.c src/drivers.c src/fftw.c
 
 #CC = gcc-8
 #CC = /usr/local/opt/llvm/bin/clang
@@ -24,7 +24,7 @@ OBJ = src/transforms.c src/rotations.c src/permute.c src/drivers.c src/fftw.c
 
 #AVX512 = -mavx512f
 
-CFLAGS = -std=gnu99 -Ofast $(AVX512) -march=native -mtune=native -I./src
+CFLAGS = -std=gnu99 -Ofast $(AVX512) -march=native -mtune=native -mno-vzeroupper -I./src
 
 ifeq ($(CC),/usr/local/opt/llvm/bin/clang)
     CFLAGS += -I/usr/local/opt/llvm/include
@@ -49,9 +49,11 @@ ifeq ($(UNAME), Windows)
     LDFLAGS += -L.
 endif
 
-LDLIBS += -lm -lblas -lfftw3
+LDLIBS += -lm -lquadmath -lmpfr -fopenmp -lblas -lfftw3
 ifeq ($(UNAME), Linux)
-    LDLIBS += -lfftw3_threads
+    LDLIBS += -lfftw3_threads -lgmp
 else ifeq ($(UNAME), Darwin)
-    LDLIBS += -lfftw3_threads
+    LDLIBS += -lfftw3_threads -lgmp
+else ifeq ($(OS), Windows_NT)
+    LDLIBS += -lmpir
 endif

Make.inc

@@ -22,13 +22,23 @@ tests:
 	$(CC) src/ftutilities.c test/test_transforms.c $(CFLAGS) -L$(LIBDIR) -l$(LIB) $(LDFLAGS) $(LDLIBS) -o test_transforms
 	$(CC) src/ftutilities.c test/test_permute.c $(CFLAGS) -L$(LIBDIR) -l$(LIB) $(LDFLAGS) $(LDLIBS) -o test_permute
 	$(CC) src/ftutilities.c test/test_rotations.c $(CFLAGS) -L$(LIBDIR) -l$(LIB) $(LDFLAGS) $(LDLIBS) -o test_rotations
+	$(CC) src/ftutilities.c test/test_tridiagonal.c $(CFLAGS) -L$(LIBDIR) -l$(LIB) $(LDFLAGS) $(LDLIBS) -o test_tridiagonal
+	$(CC) src/ftutilities.c test/test_hierarchical.c $(CFLAGS) -L$(LIBDIR) -l$(LIB) $(LDFLAGS) $(LDLIBS) -o test_hierarchical
+	$(CC) src/ftutilities.c test/test_banded.c $(CFLAGS) -L$(LIBDIR) -l$(LIB) $(LDFLAGS) $(LDLIBS) -o test_banded
+	$(CC) src/ftutilities.c test/test_dprk.c $(CFLAGS) -L$(LIBDIR) -l$(LIB) $(LDFLAGS) $(LDLIBS) -o test_dprk
+	$(CC) src/ftutilities.c test/test_tdc.c $(CFLAGS) -L$(LIBDIR) -l$(LIB) $(LDFLAGS) $(LDLIBS) -o test_tdc
 	$(CC) src/ftutilities.c test/test_drivers.c $(CFLAGS) -L$(LIBDIR) -l$(LIB) $(LDFLAGS) $(LDLIBS) -o test_drivers
 	$(CC) src/ftutilities.c test/test_fftw.c $(CFLAGS) -L$(LIBDIR) -l$(LIB) $(LDFLAGS) $(LDLIBS) -o test_fftw
 
 wintests:
 	$(CC) src/ftutilities.c test/test_transforms.c $(CFLAGS) -L$(LIBDIR) -l$(LIB) $(LDFLAGS) $(LDLIBS) -o test_transforms
 	$(CC) src/ftutilities.c test/test_permute.c $(CFLAGS) -L$(LIBDIR) -l$(LIB) $(LDFLAGS) $(LDLIBS) -o test_permute
 	$(CC) src/ftutilities.c test/test_rotations.c $(CFLAGS) -L$(LIBDIR) -l$(LIB) $(LDFLAGS) $(LDLIBS) -o test_rotations
+	$(CC) src/ftutilities.c test/test_tridiagonal.c $(CFLAGS) -L$(LIBDIR) -l$(LIB) $(LDFLAGS) $(LDLIBS) -o test_tridiagonal
+	$(CC) src/ftutilities.c test/test_hierarchical.c $(CFLAGS) -L$(LIBDIR) -l$(LIB) $(LDFLAGS) $(LDLIBS) -o test_hierarchical
+	$(CC) src/ftutilities.c test/test_banded.c $(CFLAGS) -L$(LIBDIR) -l$(LIB) $(LDFLAGS) $(LDLIBS) -o test_banded
+	$(CC) src/ftutilities.c test/test_dprk.c $(CFLAGS) -L$(LIBDIR) -l$(LIB) $(LDFLAGS) $(LDLIBS) -o test_dprk
+	$(CC) src/ftutilities.c test/test_tdc.c $(CFLAGS) -L$(LIBDIR) -l$(LIB) $(LDFLAGS) $(LDLIBS) -o test_tdc
 	$(CC) src/ftutilities.c test/test_drivers.c $(CFLAGS) -L$(LIBDIR) -l$(LIB) $(LDFLAGS) $(LDLIBS) -o test_drivers
 
 examples:
@@ -38,13 +48,9 @@ examples:
 
 clean:
 	rm -f lib$(LIB).$(SLIB)
-	rm -f test_assembly.s
 	rm -f additiontheorem
+	rm -f calculus
 	rm -f holomorphic
-	rm -f test_transforms
-	rm -f test_permute
-	rm -f test_rotations
-	rm -f test_drivers
-	rm -f test_fftw
+	rm -f test_*
 
 .PHONY: all lib assembly examples tests clean

Makefile

@@ -155,7 +155,7 @@ int main(void) {
     printf("\n");
 
     printf("The 2-norm relative error in differentiating the Proriol series\n");
-    printf("for "MAGENTA("f(x,y)")" term-by-term and its sampled gradient is %4.2e.\n", sqrt(pow(norm_2arg(Fx, Gx, N*M), 2) + pow(norm_2arg(Fy, Gy, N*M), 2))/sqrt(pow(norm_1arg(Fx, N*M), 2) + pow(norm_1arg(Fy, N*M), 2)));
+    printf("for "MAGENTA("f(x,y)")" term-by-term and its sampled gradient is %4.2e.\n", sqrt(pow(ft_norm_2arg(Fx, Gx, N*M), 2) + pow(ft_norm_2arg(Fy, Gy, N*M), 2))/sqrt(pow(ft_norm_1arg(Fx, N*M), 2) + pow(ft_norm_1arg(Fy, N*M), 2)));
     printf("This error can be improved upon by increasing "MAGENTA("N")" and "MAGENTA("M")".\n");
 
     ft_destroy_harmonic_plan(P);

examples/calculus.c

@@ -81,7 +81,7 @@ int main(void) {
     printf(".\n\n");
     printf("Using an orthonormal basis, the integral of "MAGENTA("[f(x,y)]^2")" over the\n");
     printf("disk is approximately the square of the 2-norm of the coefficients, ");
-    printf(FMT, pow(norm_1arg(F, N*M), 2));
+    printf(FMT, pow(ft_norm_1arg(F, N*M), 2));
     printf(".\n");
 
     ft_destroy_harmonic_plan(P);