1. Extend the support of VLEN to 65536. The data type in struct tw_in…

…str of kernel/ifftw.h should be extended synchronously, to support longer integer.

2. Include vtw.h for VTW1, VTW2, and VTWS, to be compatible with the coding style of ARM SVE. Although the vtw.h is auto-generated in the version of ARM SVE, for the integrity and correctness, the vtw.h will be submitted in this version.

Makefile.am

@@ -112,7 +112,19 @@ rdft/simd/rvv256/librdft_rvv256_codelets.la	\
 dft/simd/rvv512/libdft_rvv512_codelets.la      \
 rdft/simd/rvv512/librdft_rvv512_codelets.la	\
 dft/simd/rvv1024/libdft_rvv1024_codelets.la      \
-rdft/simd/rvv1024/librdft_rvv1024_codelets.la
+rdft/simd/rvv1024/librdft_rvv1024_codelets.la      \
+dft/simd/rvv2048/libdft_rvv2048_codelets.la      \
+rdft/simd/rvv2048/librdft_rvv2048_codelets.la      \
+dft/simd/rvv4096/libdft_rvv4096_codelets.la      \
+rdft/simd/rvv4096/librdft_rvv4096_codelets.la      \
+dft/simd/rvv8192/libdft_rvv8192_codelets.la      \
+rdft/simd/rvv8192/librdft_rvv8192_codelets.la      \
+dft/simd/rvv16384/libdft_rvv16384_codelets.la      \
+rdft/simd/rvv16384/librdft_rvv16384_codelets.la    \
+dft/simd/rvv32768/libdft_rvv32768_codelets.la      \
+rdft/simd/rvv32768/librdft_rvv32768_codelets.la    \
+dft/simd/rvv65536/libdft_rvv65536_codelets.la      \
+rdft/simd/rvv65536/librdft_rvv65536_codelets.la
 endif
 
 if THREADS

api/version.c

@@ -85,7 +85,7 @@ const char X(version)[] = PACKAGE "-" PACKAGE_VERSION
    "-generic_simd256"
 #endif
 
-#if defined(HAVE_R5V)
+#if defined(HAVE_RVV)
    "-rvv"
 #endif
 

configure.ac

@@ -780,6 +780,12 @@ AC_CONFIG_FILES([
    dft/simd/rvv256/Makefile
    dft/simd/rvv512/Makefile
    dft/simd/rvv1024/Makefile
+   dft/simd/rvv2048/Makefile
+   dft/simd/rvv4096/Makefile
+   dft/simd/rvv8192/Makefile
+   dft/simd/rvv16384/Makefile
+   dft/simd/rvv32768/Makefile
+   dft/simd/rvv65536/Makefile
 
    rdft/Makefile
    rdft/scalar/Makefile
@@ -804,6 +810,12 @@ AC_CONFIG_FILES([
    rdft/simd/rvv256/Makefile
    rdft/simd/rvv512/Makefile
    rdft/simd/rvv1024/Makefile
+   rdft/simd/rvv2048/Makefile
+   rdft/simd/rvv4096/Makefile
+   rdft/simd/rvv8192/Makefile
+   rdft/simd/rvv16384/Makefile
+   rdft/simd/rvv32768/Makefile
+   rdft/simd/rvv65536/Makefile
 
    reodft/Makefile
 

dft/codelet-dft.h

@@ -112,5 +112,11 @@ extern const solvtab X(solvtab_dft_rvv128);
 extern const solvtab X(solvtab_dft_rvv256);
 extern const solvtab X(solvtab_dft_rvv512);
 extern const solvtab X(solvtab_dft_rvv1024);
+extern const solvtab X(solvtab_dft_rvv2048);
+extern const solvtab X(solvtab_dft_rvv4096);
+extern const solvtab X(solvtab_dft_rvv8192);
+extern const solvtab X(solvtab_dft_rvv16384);
+extern const solvtab X(solvtab_dft_rvv32768);
+extern const solvtab X(solvtab_dft_rvv65536);
 
 #endif				/* __DFT_CODELET_H__ */

dft/conf.c

@@ -95,5 +95,17 @@ void X(dft_conf_standard)(planner *p)
           X(solvtab_exec)(X(solvtab_dft_rvv512), p);
      if (X(have_simd_rvv)(1024))
           X(solvtab_exec)(X(solvtab_dft_rvv1024), p);
+     if (X(have_simd_rvv)(2048))
+          X(solvtab_exec)(X(solvtab_dft_rvv2048), p);
+     if (X(have_simd_rvv)(4096))
+          X(solvtab_exec)(X(solvtab_dft_rvv4096), p);
+     if (X(have_simd_rvv)(8192))
+          X(solvtab_exec)(X(solvtab_dft_rvv8192), p);
+     if (X(have_simd_rvv)(16384))
+          X(solvtab_exec)(X(solvtab_dft_rvv16384), p);
+     if (X(have_simd_rvv)(32768))
+          X(solvtab_exec)(X(solvtab_dft_rvv32768), p);
+     if (X(have_simd_rvv)(65536))
+          X(solvtab_exec)(X(solvtab_dft_rvv65536), 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 generic-simd128 generic-simd256 rvv128 rvv256 rvv512 rvv1024
+SUBDIRS = common sse2 avx avx-128-fma avx2 avx2-128 avx512 kcvi altivec vsx neon generic-simd128 generic-simd256 rvv128 rvv256 rvv512 rvv1024 rvv2048 rvv4096 rvv8192 rvv16384 rvv32768 rvv65536
 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/rvv16384/Makefile.am

@@ -0,0 +1,12 @@
+SIMD_HEADER=simd-support/simd-rvv16384.h
+
+include $(top_srcdir)/dft/simd/codlist.mk
+include $(top_srcdir)/dft/simd/simd.mk
+
+if HAVE_RVV
+
+BUILT_SOURCES = $(EXTRA_DIST)
+noinst_LTLIBRARIES = libdft_rvv16384_codelets.la
+libdft_rvv16384_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif

dft/simd/rvv2048/Makefile.am

@@ -0,0 +1,12 @@
+SIMD_HEADER=simd-support/simd-rvv2048.h
+
+include $(top_srcdir)/dft/simd/codlist.mk
+include $(top_srcdir)/dft/simd/simd.mk
+
+if HAVE_RVV
+
+BUILT_SOURCES = $(EXTRA_DIST)
+noinst_LTLIBRARIES = libdft_rvv2048_codelets.la
+libdft_rvv2048_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif

dft/simd/rvv32768/Makefile.am

@@ -0,0 +1,12 @@
+SIMD_HEADER=simd-support/simd-rvv32768.h
+
+include $(top_srcdir)/dft/simd/codlist.mk
+include $(top_srcdir)/dft/simd/simd.mk
+
+if HAVE_RVV
+
+BUILT_SOURCES = $(EXTRA_DIST)
+noinst_LTLIBRARIES = libdft_rvv32768_codelets.la
+libdft_rvv32768_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif

dft/simd/rvv4096/Makefile.am

@@ -0,0 +1,12 @@
+SIMD_HEADER=simd-support/simd-rvv4096.h
+
+include $(top_srcdir)/dft/simd/codlist.mk
+include $(top_srcdir)/dft/simd/simd.mk
+
+if HAVE_RVV
+
+BUILT_SOURCES = $(EXTRA_DIST)
+noinst_LTLIBRARIES = libdft_rvv4096_codelets.la
+libdft_rvv4096_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif

dft/simd/rvv65536/Makefile.am

@@ -0,0 +1,12 @@
+SIMD_HEADER=simd-support/simd-rvv65536.h
+
+include $(top_srcdir)/dft/simd/codlist.mk
+include $(top_srcdir)/dft/simd/simd.mk
+
+if HAVE_RVV
+
+BUILT_SOURCES = $(EXTRA_DIST)
+noinst_LTLIBRARIES = libdft_rvv65536_codelets.la
+libdft_rvv65536_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif

dft/simd/rvv8192/Makefile.am

@@ -0,0 +1,12 @@
+SIMD_HEADER=simd-support/simd-rvv8192.h
+
+include $(top_srcdir)/dft/simd/codlist.mk
+include $(top_srcdir)/dft/simd/simd.mk
+
+if HAVE_RVV
+
+BUILT_SOURCES = $(EXTRA_DIST)
+noinst_LTLIBRARIES = libdft_rvv8192_codelets.la
+libdft_rvv8192_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif

kernel/ifftw.h

@@ -877,7 +877,7 @@ 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

@@ -151,6 +151,12 @@ extern const solvtab X(solvtab_rdft_rvv128);
 extern const solvtab X(solvtab_rdft_rvv256);
 extern const solvtab X(solvtab_rdft_rvv512);
 extern const solvtab X(solvtab_rdft_rvv1024);
+extern const solvtab X(solvtab_rdft_rvv2048);
+extern const solvtab X(solvtab_rdft_rvv4096);
+extern const solvtab X(solvtab_rdft_rvv8192);
+extern const solvtab X(solvtab_rdft_rvv16384);
+extern const solvtab X(solvtab_rdft_rvv32768);
+extern const solvtab X(solvtab_rdft_rvv65536);
 
 /* real-input & output DFT-like codelets (DHT, etc.) */
 typedef struct kr2r_desc_s kr2r_desc;

rdft/conf.c

@@ -103,13 +103,25 @@ void X(rdft_conf_standard)(planner *p)
      X(solvtab_exec)(X(solvtab_rdft_generic_simd256), p);
 #endif
 #if HAVE_RVV
-	 if (X(have_simd_rvv)(128))
+     if (X(have_simd_rvv)(128))
           X(solvtab_exec)(X(solvtab_rdft_rvv128), p);
      if (X(have_simd_rvv)(256))
           X(solvtab_exec)(X(solvtab_rdft_rvv256), p);
      if (X(have_simd_rvv)(512))
           X(solvtab_exec)(X(solvtab_rdft_rvv512), p);
      if (X(have_simd_rvv)(1024))
           X(solvtab_exec)(X(solvtab_rdft_rvv1024), p);
+     if (X(have_simd_rvv)(2048))
+          X(solvtab_exec)(X(solvtab_rdft_rvv2048), p);
+     if (X(have_simd_rvv)(4096))
+          X(solvtab_exec)(X(solvtab_rdft_rvv4096), p);
+     if (X(have_simd_rvv)(8192))
+          X(solvtab_exec)(X(solvtab_rdft_rvv8192), p);
+     if (X(have_simd_rvv)(16384))
+          X(solvtab_exec)(X(solvtab_rdft_rvv16384), p);
+     if (X(have_simd_rvv)(32768))
+          X(solvtab_exec)(X(solvtab_rdft_rvv32768), p);
+     if (X(have_simd_rvv)(65536))
+          X(solvtab_exec)(X(solvtab_rdft_rvv65536), 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 generic-simd128 generic-simd256 rvv128 rvv256 rvv512 rvv1024
+SUBDIRS = common sse2 avx avx-128-fma avx2 avx2-128 avx512 kcvi altivec vsx neon generic-simd128 generic-simd256 rvv128 rvv256 rvv512 rvv1024 rvv2048 rvv4096 rvv8192 rvv16384 rvv32768 rvv65536
 EXTRA_DIST = hc2cbv.h hc2cfv.h codlist.mk simd.mk

rdft/simd/rvv16384/Makefile.am

@@ -0,0 +1,12 @@
+SIMD_HEADER=simd-support/simd-rvv16384.h
+
+include $(top_srcdir)/rdft/simd/codlist.mk
+include $(top_srcdir)/rdft/simd/simd.mk
+
+if HAVE_RVV
+
+noinst_LTLIBRARIES = librdft_rvv16384_codelets.la
+BUILT_SOURCES = $(EXTRA_DIST)
+librdft_rvv16384_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif

rdft/simd/rvv2048/Makefile.am

@@ -0,0 +1,12 @@
+SIMD_HEADER=simd-support/simd-rvv2048.h
+
+include $(top_srcdir)/rdft/simd/codlist.mk
+include $(top_srcdir)/rdft/simd/simd.mk
+
+if HAVE_RVV
+
+noinst_LTLIBRARIES = librdft_rvv2048_codelets.la
+BUILT_SOURCES = $(EXTRA_DIST)
+librdft_rvv2048_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif

rdft/simd/rvv32768/Makefile.am

@@ -0,0 +1,12 @@
+SIMD_HEADER=simd-support/simd-rvv32768.h
+
+include $(top_srcdir)/rdft/simd/codlist.mk
+include $(top_srcdir)/rdft/simd/simd.mk
+
+if HAVE_RVV
+
+noinst_LTLIBRARIES = librdft_rvv32768_codelets.la
+BUILT_SOURCES = $(EXTRA_DIST)
+librdft_rvv32768_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif

rdft/simd/rvv4096/Makefile.am

@@ -0,0 +1,12 @@
+SIMD_HEADER=simd-support/simd-rvv4096.h
+
+include $(top_srcdir)/rdft/simd/codlist.mk
+include $(top_srcdir)/rdft/simd/simd.mk
+
+if HAVE_RVV
+
+noinst_LTLIBRARIES = librdft_rvv4096_codelets.la
+BUILT_SOURCES = $(EXTRA_DIST)
+librdft_rvv4096_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif

rdft/simd/rvv65536/Makefile.am

@@ -0,0 +1,12 @@
+SIMD_HEADER=simd-support/simd-rvv65536.h
+
+include $(top_srcdir)/rdft/simd/codlist.mk
+include $(top_srcdir)/rdft/simd/simd.mk
+
+if HAVE_RVV
+
+noinst_LTLIBRARIES = librdft_rvv65536_codelets.la
+BUILT_SOURCES = $(EXTRA_DIST)
+librdft_rvv65536_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif

rdft/simd/rvv8192/Makefile.am

@@ -0,0 +1,12 @@
+SIMD_HEADER=simd-support/simd-rvv8192.h
+
+include $(top_srcdir)/rdft/simd/codlist.mk
+include $(top_srcdir)/rdft/simd/simd.mk
+
+if HAVE_RVV
+
+noinst_LTLIBRARIES = librdft_rvv8192_codelets.la
+BUILT_SOURCES = $(EXTRA_DIST)
+librdft_rvv8192_codelets_la_SOURCES = $(BUILT_SOURCES)
+
+endif

simd-support/Makefile.am

@@ -12,4 +12,5 @@ kcvi.c simd-kcvi.h \
 altivec.c simd-altivec.h vsx.c simd-vsx.h \
 neon.c simd-neon.h \
 simd-generic128.h simd-generic256.h \
-rvv.c simd-rvv.h simd-rvv128.h simd-rvv256.h simd-rvv512.h simd-rvv1024.h
+rvv.c simd-rvv.h simd-rvv128.h simd-rvv256.h simd-rvv512.h simd-rvv1024.h  \
+simd-rvv2048.h simd-rvv4096.h simd-rvv8192.h simd-rvv16384.h simd-rvv32768.h simd-rvv65536.h vtw.h