Merging of RISC-V V1.0 support.

Makefile.am

@@ -128,6 +128,25 @@ GENERIC_SIMD256_LIBS = dft/simd/generic-simd256/libdft_generic_simd256_codelets.
 rdft/simd/generic-simd256/librdft_generic_simd256_codelets.la
 endif
 
+if HAVE_R5V
+R5V_LIBS = dft/simd/r5v128/libdft_r5v128_codelets.la      \
+rdft/simd/r5v128/librdft_r5v128_codelets.la	\
+dft/simd/r5v256/libdft_r5v256_codelets.la      \
+rdft/simd/r5v256/librdft_r5v256_codelets.la	\
+dft/simd/r5v512/libdft_r5v512_codelets.la      \
+rdft/simd/r5v512/librdft_r5v512_codelets.la	\
+dft/simd/r5v1024/libdft_r5v1024_codelets.la      \
+rdft/simd/r5v1024/librdft_r5v1024_codelets.la	\
+dft/simd/r5v2048/libdft_r5v2048_codelets.la      \
+rdft/simd/r5v2048/librdft_r5v2048_codelets.la	\
+dft/simd/r5v4096/libdft_r5v4096_codelets.la      \
+rdft/simd/r5v4096/librdft_r5v4096_codelets.la   \
+dft/simd/r5v8192/libdft_r5v8192_codelets.la      \
+rdft/simd/r5v8192/librdft_r5v8192_codelets.la   \
+dft/simd/r5v16384/libdft_r5v16384_codelets.la      \
+rdft/simd/r5v16384/librdft_r5v16384_codelets.la
+endif
+
 if THREADS
 if COMBINED_THREADS
 COMBINED_THREADLIBS=threads/libfftw3@PREC_SUFFIX@_threads.la
@@ -152,7 +171,7 @@ libfftw3@PREC_SUFFIX@_la_LIBADD =			\
         $(AVX2_LIBS) $(ALTIVEC_LIBS) \
         $(VSX_LIBS) $(NEON_LIBS) $(SVE_LIBS) $(KCVI_LIBS) $(AVX512_LIBS) \
         $(LSX_LIBS) $(LASX_LIBS) \
-        $(GENERIC_SIMD128_LIBS) $(GENERIC_SIMD256_LIBS) \
+        $(GENERIC_SIMD128_LIBS) $(GENERIC_SIMD256_LIBS) $(R5V_LIBS) \
 	$(COMBINED_THREADLIBS)
 
 if QUAD

api/version.c

@@ -97,4 +97,8 @@ const char X(version)[] = PACKAGE "-" PACKAGE_VERSION
    "-generic_simd256"
 #endif
 
+#if defined(HAVE_R5V)
+   "-r5v"
+#endif
+
 ;

configure.ac

@@ -253,6 +253,12 @@ if test "$have_sve" = "yes"; then
 fi
 AM_CONDITIONAL(HAVE_SVE, test "$have_sve" = "yes")
 
+AC_ARG_ENABLE(r5v, [AC_HELP_STRING([--enable-r5v],[enable RISC-V V optimizations])], have_r5v=$enableval, have_r5v=no)
+if test "$have_r5v" = "yes"; then
+        AC_DEFINE(HAVE_R5V,1,[Define to enable RISC-V V optimizations.])
+fi
+AM_CONDITIONAL(HAVE_R5V, test "$have_r5v" = "yes")
+
 dnl FIXME:
 dnl AC_ARG_ENABLE(mips-ps, [AS_HELP_STRING([--enable-mips-ps],[enable MIPS pair-single optimizations])], have_mips_ps=$enableval, have_mips_ps=no)
 dnl if test "$have_mips_ps" = "yes"; then
@@ -825,6 +831,14 @@ AC_CONFIG_FILES([
    dft/simd/lasx/Makefile
    dft/simd/generic-simd128/Makefile
    dft/simd/generic-simd256/Makefile
+   dft/simd/r5v128/Makefile
+   dft/simd/r5v256/Makefile
+   dft/simd/r5v512/Makefile
+   dft/simd/r5v1024/Makefile
+   dft/simd/r5v2048/Makefile
+   dft/simd/r5v4096/Makefile
+   dft/simd/r5v8192/Makefile
+   dft/simd/r5v16384/Makefile
 
    rdft/Makefile
    rdft/scalar/Makefile
@@ -852,6 +866,14 @@ AC_CONFIG_FILES([
    rdft/simd/lasx/Makefile
    rdft/simd/generic-simd128/Makefile
    rdft/simd/generic-simd256/Makefile
+   rdft/simd/r5v128/Makefile
+   rdft/simd/r5v256/Makefile
+   rdft/simd/r5v512/Makefile
+   rdft/simd/r5v1024/Makefile
+   rdft/simd/r5v2048/Makefile
+   rdft/simd/r5v4096/Makefile
+   rdft/simd/r5v8192/Makefile
+   rdft/simd/r5v16384/Makefile
 
    reodft/Makefile
 

dft/codelet-dft.h

@@ -115,5 +115,12 @@ extern const solvtab X(solvtab_dft_lsx);
 extern const solvtab X(solvtab_dft_lasx);
 extern const solvtab X(solvtab_dft_generic_simd128);
 extern const solvtab X(solvtab_dft_generic_simd256);
-
+extern const solvtab X(solvtab_dft_r5v128);
+extern const solvtab X(solvtab_dft_r5v256);
+extern const solvtab X(solvtab_dft_r5v512);
+extern const solvtab X(solvtab_dft_r5v1024);
+extern const solvtab X(solvtab_dft_r5v2048);
+extern const solvtab X(solvtab_dft_r5v4096);
+extern const solvtab X(solvtab_dft_r5v8192);
+extern const solvtab X(solvtab_dft_r5v16384);
 #endif				/* __DFT_CODELET_H__ */

dft/conf.c

@@ -105,4 +105,22 @@ void X(dft_conf_standard)(planner *p)
 #if HAVE_GENERIC_SIMD256
      X(solvtab_exec)(X(solvtab_dft_generic_simd256), p);
 #endif
+#if HAVE_R5V
+     if (X(have_simd_r5v)(128))
+          X(solvtab_exec)(X(solvtab_dft_r5v128), p);
+     if (X(have_simd_r5v)(256))
+          X(solvtab_exec)(X(solvtab_dft_r5v256), p);
+     if (X(have_simd_r5v)(512))
+          X(solvtab_exec)(X(solvtab_dft_r5v512), p);
+     if (X(have_simd_r5v)(1024))
+          X(solvtab_exec)(X(solvtab_dft_r5v1024), p);
+     if (X(have_simd_r5v)(2048))
+          X(solvtab_exec)(X(solvtab_dft_r5v2048), p);
+     if (X(have_simd_r5v)(4096))
+	  X(solvtab_exec)(X(solvtab_dft_r5v4096), p);
+     if (X(have_simd_r5v)(8192))
+	  X(solvtab_exec)(X(solvtab_dft_r5v8192), p);
+     if (X(have_simd_r5v)(16384))
+	  X(solvtab_exec)(X(solvtab_dft_r5v16384), p);
+#endif
 }

dft/simd/Makefile.am

@@ -1,4 +1,4 @@
 AM_CPPFLAGS = -I $(top_srcdir)
-SUBDIRS = common sse2 avx avx-128-fma avx2 avx2-128 avx512 kcvi altivec vsx neon sve128 sve256 sve512 sve1024 sve2048 lsx lasx generic-simd128 generic-simd256
+SUBDIRS = common sse2 avx avx-128-fma avx2 avx2-128 avx512 kcvi altivec vsx neon sve128 sve256 sve512 sve1024 sve2048 lsx lasx generic-simd128 generic-simd256 r5v128 r5v256 r5v512 r5v1024 r5v2048 r5v4096 r5v8192 r5v16384
 EXTRA_DIST = n1b.h n1f.h n2b.h n2f.h n2s.h q1b.h q1f.h t1b.h t1bu.h	\
 t1f.h t1fu.h t2b.h t2f.h t3b.h t3f.h ts.h codlist.mk simd.mk

dft/simd/r5v1024/Makefile.am

@@ -0,0 +1,12 @@
+SIMD_HEADER=simd-support/simd-r5v1024.h
+
+include $(top_srcdir)/dft/simd/codlist.mk
+include $(top_srcdir)/dft/simd/simd.mk
+
+if HAVE_R5V
+
+BUILT_SOURCES = $(EXTRA_DIST)
+noinst_LTLIBRARIES = libdft_r5v1024_codelets.la
+libdft_r5v1024_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif

dft/simd/r5v128/Makefile.am

@@ -0,0 +1,12 @@
+SIMD_HEADER=simd-support/simd-r5v128.h
+
+include $(top_srcdir)/dft/simd/codlist.mk
+include $(top_srcdir)/dft/simd/simd.mk
+
+if HAVE_R5V
+
+BUILT_SOURCES = $(EXTRA_DIST)
+noinst_LTLIBRARIES = libdft_r5v128_codelets.la
+libdft_r5v128_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif

dft/simd/r5v16384/Makefile.am

@@ -0,0 +1,12 @@
+SIMD_HEADER=simd-support/simd-r5v16384.h
+
+include $(top_srcdir)/dft/simd/codlist.mk
+include $(top_srcdir)/dft/simd/simd.mk
+
+if HAVE_R5V
+
+BUILT_SOURCES = $(EXTRA_DIST)
+noinst_LTLIBRARIES = libdft_r5v16384_codelets.la
+libdft_r5v16384_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif

dft/simd/r5v2048/Makefile.am

@@ -0,0 +1,12 @@
+SIMD_HEADER=simd-support/simd-r5v2048.h
+
+include $(top_srcdir)/dft/simd/codlist.mk
+include $(top_srcdir)/dft/simd/simd.mk
+
+if HAVE_R5V
+
+BUILT_SOURCES = $(EXTRA_DIST)
+noinst_LTLIBRARIES = libdft_r5v2048_codelets.la
+libdft_r5v2048_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif

dft/simd/r5v256/Makefile.am

@@ -0,0 +1,12 @@
+SIMD_HEADER=simd-support/simd-r5v256.h
+
+include $(top_srcdir)/dft/simd/codlist.mk
+include $(top_srcdir)/dft/simd/simd.mk
+
+if HAVE_R5V
+
+BUILT_SOURCES = $(EXTRA_DIST)
+noinst_LTLIBRARIES = libdft_r5v256_codelets.la
+libdft_r5v256_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif

dft/simd/r5v4096/Makefile.am

@@ -0,0 +1,12 @@
+SIMD_HEADER=simd-support/simd-r5v4096.h
+
+include $(top_srcdir)/dft/simd/codlist.mk
+include $(top_srcdir)/dft/simd/simd.mk
+
+if HAVE_R5V
+
+BUILT_SOURCES = $(EXTRA_DIST)
+noinst_LTLIBRARIES = libdft_r5v4096_codelets.la
+libdft_r5v4096_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif

dft/simd/r5v512/Makefile.am

@@ -0,0 +1,12 @@
+SIMD_HEADER=simd-support/simd-r5v512.h
+
+include $(top_srcdir)/dft/simd/codlist.mk
+include $(top_srcdir)/dft/simd/simd.mk
+
+if HAVE_R5V
+
+BUILT_SOURCES = $(EXTRA_DIST)
+noinst_LTLIBRARIES = libdft_r5v512_codelets.la
+libdft_r5v512_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif

dft/simd/r5v8192/Makefile.am

@@ -0,0 +1,12 @@
+SIMD_HEADER=simd-support/simd-r5v8192.h
+
+include $(top_srcdir)/dft/simd/codlist.mk
+include $(top_srcdir)/dft/simd/simd.mk
+
+if HAVE_R5V
+
+BUILT_SOURCES = $(EXTRA_DIST)
+noinst_LTLIBRARIES = libdft_r5v8192_codelets.la
+libdft_r5v8192_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif

kernel/ifftw.h

@@ -109,7 +109,8 @@ extern void X(extract_reim)(int sign, R *c, R **r, R **i);
       defined(HAVE_ALTIVEC) || defined(HAVE_VSX) || \
       defined(HAVE_MIPS_PS) || \
       defined(HAVE_LSX) || defined(HAVE_LASX) || \
-      defined(HAVE_GENERIC_SIMD128) || defined(HAVE_GENERIC_SIMD256)
+      defined(HAVE_GENERIC_SIMD128) || defined(HAVE_GENERIC_SIMD256) || \
+      defined(HAVE_R5V)
 #define HAVE_SIMD 1
 #else
 #define HAVE_SIMD 0
@@ -127,6 +128,7 @@ extern int X(have_simd_neon)(void);
 extern int X(have_simd_lsx)(void);
 extern int X(have_simd_lasx)(void);
 extern int X(have_simd_sve)(int minwidth);
+extern int X(have_simd_r5v)(int);
 
 /* forward declarations */
 typedef struct problem_s problem;
@@ -884,7 +886,7 @@ enum { TW_COS = 0, TW_SIN = 1, TW_CEXP = 2, TW_NEXT = 3,
 
 typedef struct {
      unsigned char op;
-     signed char v;
+     short v;
      short i;
 } tw_instr;
 

rdft/codelet-rdft.h

@@ -154,6 +154,14 @@ extern const solvtab X(solvtab_rdft_lsx);
 extern const solvtab X(solvtab_rdft_lasx);
 extern const solvtab X(solvtab_rdft_generic_simd128);
 extern const solvtab X(solvtab_rdft_generic_simd256);
+extern const solvtab X(solvtab_rdft_r5v128);
+extern const solvtab X(solvtab_rdft_r5v256);
+extern const solvtab X(solvtab_rdft_r5v512);
+extern const solvtab X(solvtab_rdft_r5v1024);
+extern const solvtab X(solvtab_rdft_r5v2048);
+extern const solvtab X(solvtab_rdft_r5v4096);
+extern const solvtab X(solvtab_rdft_r5v8192);
+extern const solvtab X(solvtab_rdft_r5v16384);
 
 /* real-input & output DFT-like codelets (DHT, etc.) */
 typedef struct kr2r_desc_s kr2r_desc;

rdft/conf.c

@@ -122,4 +122,22 @@ void X(rdft_conf_standard)(planner *p)
 #if HAVE_GENERIC_SIMD256
      X(solvtab_exec)(X(solvtab_rdft_generic_simd256), p);
 #endif
+#if HAVE_R5V
+     if (X(have_simd_r5v)(128))
+          X(solvtab_exec)(X(solvtab_rdft_r5v128), p);
+     if (X(have_simd_r5v)(256))
+          X(solvtab_exec)(X(solvtab_rdft_r5v256), p);
+     if (X(have_simd_r5v)(512))
+          X(solvtab_exec)(X(solvtab_rdft_r5v512), p);
+     if (X(have_simd_r5v)(1024))
+          X(solvtab_exec)(X(solvtab_rdft_r5v1024), p);
+     if (X(have_simd_r5v)(2048))
+          X(solvtab_exec)(X(solvtab_rdft_r5v2048), p);
+     if (X(have_simd_r5v)(4096))
+	  X(solvtab_exec)(X(solvtab_rdft_r5v4096), p);
+     if (X(have_simd_r5v)(8192))
+	  X(solvtab_exec)(X(solvtab_rdft_r5v8192), p);
+     if (X(have_simd_r5v)(16384))
+	  X(solvtab_exec)(X(solvtab_rdft_r5v16384), p);
+#endif
 }

rdft/simd/Makefile.am

@@ -1,4 +1,4 @@
 
 AM_CPPFLAGS = -I $(top_srcdir)
-SUBDIRS = common sse2 avx avx-128-fma avx2 avx2-128 avx512 kcvi altivec vsx neon sve128 sve256 sve512 sve1024 sve2048 lsx lasx generic-simd128 generic-simd256
+SUBDIRS = common sse2 avx avx-128-fma avx2 avx2-128 avx512 kcvi altivec vsx neon sve128 sve256 sve512 sve1024 sve2048 lsx lasx generic-simd128 generic-simd256 r5v128 r5v256 r5v512 r5v1024 r5v2048 r5v4096 r5v8192 r5v16384
 EXTRA_DIST = hc2cbv.h hc2cfv.h codlist.mk simd.mk

rdft/simd/r5v1024/Makefile.am

@@ -0,0 +1,14 @@
+SIMD_HEADER=simd-support/simd-r5v1024.h
+
+include $(top_srcdir)/rdft/simd/codlist.mk
+include $(top_srcdir)/rdft/simd/simd.mk
+
+if HAVE_R5V
+
+noinst_LTLIBRARIES = librdft_r5v1024_codelets.la
+BUILT_SOURCES = $(EXTRA_DIST)
+librdft_r5v1024_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif
+
+

rdft/simd/r5v128/Makefile.am

@@ -0,0 +1,14 @@
+SIMD_HEADER=simd-support/simd-r5v128.h
+
+include $(top_srcdir)/rdft/simd/codlist.mk
+include $(top_srcdir)/rdft/simd/simd.mk
+
+if HAVE_R5V
+
+noinst_LTLIBRARIES = librdft_r5v128_codelets.la
+BUILT_SOURCES = $(EXTRA_DIST)
+librdft_r5v128_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif
+
+

rdft/simd/r5v16384/Makefile.am

@@ -0,0 +1,14 @@
+SIMD_HEADER=simd-support/simd-r5v16384.h
+
+include $(top_srcdir)/rdft/simd/codlist.mk
+include $(top_srcdir)/rdft/simd/simd.mk
+
+if HAVE_R5V
+
+noinst_LTLIBRARIES = librdft_r5v16384_codelets.la
+BUILT_SOURCES = $(EXTRA_DIST)
+librdft_r5v16384_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif
+
+

rdft/simd/r5v2048/Makefile.am

@@ -0,0 +1,14 @@
+SIMD_HEADER=simd-support/simd-r5v2048.h
+
+include $(top_srcdir)/rdft/simd/codlist.mk
+include $(top_srcdir)/rdft/simd/simd.mk
+
+if HAVE_R5V
+
+noinst_LTLIBRARIES = librdft_r5v2048_codelets.la
+BUILT_SOURCES = $(EXTRA_DIST)
+librdft_r5v2048_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif
+
+

rdft/simd/r5v256/Makefile.am

@@ -0,0 +1,14 @@
+SIMD_HEADER=simd-support/simd-r5v256.h
+
+include $(top_srcdir)/rdft/simd/codlist.mk
+include $(top_srcdir)/rdft/simd/simd.mk
+
+if HAVE_R5V
+
+noinst_LTLIBRARIES = librdft_r5v256_codelets.la
+BUILT_SOURCES = $(EXTRA_DIST)
+librdft_r5v256_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif
+
+

rdft/simd/r5v4096/Makefile.am

@@ -0,0 +1,14 @@
+SIMD_HEADER=simd-support/simd-r5v4096.h
+
+include $(top_srcdir)/rdft/simd/codlist.mk
+include $(top_srcdir)/rdft/simd/simd.mk
+
+if HAVE_R5V
+
+noinst_LTLIBRARIES = librdft_r5v4096_codelets.la
+BUILT_SOURCES = $(EXTRA_DIST)
+librdft_r5v4096_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif
+
+