Package: atlas / 3.10.3-8

Metadata

Package Version Patches format
atlas 3.10.3-8 3.0 (quilt)

Patch series

view the series file
Patch File delta Description
rename lapack atlas.patch | (download)

CONFIG/src/SpewMakeInc.c | 2 1 + 1 - 0 !
Make.top | 4 2 + 2 - 0 !
makes/Make.lib | 24 12 + 12 - 0 !
3 files changed, 15 insertions(+), 15 deletions(-)

 rename atlas incomplete liblapack.a to liblapack_atlas.a
 The liblapack.a created by ATLAS build system is incomplete: it only contains
 a small subset of the full LAPACK API (only those optimized by ATLAS).
 Hence rename this library to liblapack_atlas.a to avoid a name conflict with
 the full liblapack.a.
 In particular, note that the "-Ss flapack" option of the configure script does
 not work as expected as of ATLAS 3.10.3.
mips.patch | (download)

CONFIG/src/SpewMakeInc.c | 2 1 + 1 - 0 !
CONFIG/src/probe_comp.c | 2 1 + 1 - 0 !
2 files changed, 2 insertions(+), 2 deletions(-)

 on mips{,el}, do not pass -mabi=n32 flag.
 The mips{,el} ports use the O32 ABI.
 See https://www.linux-mips.org/wiki/MIPS_ABI_History for more details on ABIs.
Bug: https://sourceforge.net/p/math-atlas/support-requests/1053/
kfreebsd.patch | (download)

CONFIG/src/probe_OS.c | 11 6 + 5 - 0 !
1 file changed, 6 insertions(+), 5 deletions(-)

 ensure that kfreebsd-* is detected as freebsd
 The "GNU" test is moved at the end, otherwise kfreebsd is recognized as Linux.
 This is necessary in order to have correct linker flags.
generic.patch | (download)

CONFIG/include/atlconf.h | 7 4 + 3 - 0 !
1 file changed, 4 insertions(+), 3 deletions(-)

 add a generic machine type
 In practice this type will have the same effect than the UNKNOWN machine type,
 except that it does not try to autodetect the machine CPU.
 This is useful when trying to build a generic package on an arch for which
 ATLAS does not define a generic machine type. For example, on a armel buildd
 with ARMv7 CPU, ATLAS would detect ARMv7 if provided the UNKNOWN type.
Bug-Debian: https://bugs.debian.org/719355
ppc64el abiv2.patch | (download)

tune/blas/gemm/CASES/ATL_dmm4x4x2pf_av.c | 7 7 + 0 - 0 !
tune/blas/gemm/CASES/ATL_dmm4x4x32_ppc.c | 9 8 + 1 - 0 !
tune/blas/gemm/CASES/ATL_dmm4x4x80_ppc.c | 20 18 + 2 - 0 !
3 files changed, 33 insertions(+), 3 deletions(-)

 elfv2 abi changes for ppc64el
 For more details, see:
 https://bugzilla.redhat.com/show_bug.cgi?id=1080073#c40
Bug-Debian: https://bugs.debian.org/766695
ppc64el ifdef files with lvx.patch | (download)

tune/blas/gemm/CASES/ATL_cmm4x4x128_av.c | 7 7 + 0 - 0 !
tune/blas/gemm/CASES/ATL_dmm4x4x80_ppc.c | 6 6 + 0 - 0 !
tune/blas/gemm/CASES/ATL_dmm8x4x2_vsx.c | 6 6 + 0 - 0 !
tune/blas/gemm/CASES/ATL_smm4x4x128_av.c | 5 5 + 0 - 0 !
tune/blas/gemm/CASES/ATL_smm8x2x4_av.c | 6 6 + 0 - 0 !
5 files changed, 30 insertions(+)

 skip optimizations for big-endian powerpc.
 Some of the existing optimized files/cases for PowerPC
 contain assembly instructions with implicit big-endian
 behavior - thus incorrect for the little-endian mode -
 incurring tests failures during the build (i.e., FTBFS).
 This is being worked on; this is the workaround for now.
 .
 Note: this is an alternative for Michael Normand's patch
 'atlas.3.10.2-ppc64le_do_not_use_files_with_lvx.patch',
 in order not to touch file lists/indexes, but only C 
 source code, so the C preprocessor can avoid code per-arch.
 .
 For more details, see:
 https://bugzilla.redhat.com/show_bug.cgi?id=1080073#c40
Bug-Debian: https://bugs.debian.org/766695
powerpc dcbt.patch | (download)

include/atlas_prefetch.h | 6 3 + 3 - 0 !
tune/blas/gemm/CASES/ATL_cmm4x4x128_av.c | 8 4 + 4 - 0 !
tune/blas/gemm/CASES/ATL_dmm4x4x32_ppc.c | 4 2 + 2 - 0 !
tune/blas/gemm/CASES/ATL_dmm4x4x80_ppc.c | 24 12 + 12 - 0 !
tune/blas/gemm/CASES/ATL_dmm8x4x2_vsx.c | 2 1 + 1 - 0 !
tune/blas/gemm/CASES/ATL_smm4x4x128_av.c | 20 10 + 10 - 0 !
6 files changed, 32 insertions(+), 32 deletions(-)

 fix ftbfs on powerpc with recent binutils
 A recent modification in binutils (see
 https://sourceware.org/ml/binutils/2015-04/msg00332.html) introduced a change
 in the way the ppc assembly instructions "dcbt" and "dcbtst" are handled.
 When compiling on a generic ppc processor, the 3-arguments form of those
 instructions is no longer accepted, and one must instead use a 2-arguments
 form (ignoring the TH argument).
 .
 Incidentally, the binutils change also fixed a bug in the way those
 instructions were handled. On a generic ppc processor, they used to be
 interpreted in the so-called "embedded" form ("dcbt TH, RA, RB", only used on
 some embedded machines), while they should have been interpreted in the
 so-called "server" form ("dcbt RA, RB, TH"). The ATLAS assembly directive were
 apparently written in "server" form, and were therefore mis-assemblied.
 .
 Ideally, this patch should be improved in order to use the TH argument on
 machines that support it (only needed in atlas_prefetch.h; other instances use
 the default zero value for TH).
fix typos.patch | (download)

TexDoc/atlas_devel.tex | 2 1 + 1 - 0 !
bin/stattime.c | 2 1 + 1 - 0 !
results/atl2tvec.c | 2 1 + 1 - 0 !
results/cattvecs.c | 2 1 + 1 - 0 !
results/mergetvecs.c | 2 1 + 1 - 0 !
results/perctvecs.c | 2 1 + 1 - 0 !
results/reducetvec.c | 2 1 + 1 - 0 !
results/stattime.c | 2 1 + 1 - 0 !
results/tvec2plp.c | 2 1 + 1 - 0 !
tune/blas/gemv/mvktime.c | 2 1 + 1 - 0 !
tune/blas/ger/r1ktime.c | 2 1 + 1 - 0 !
tune/blas/ger/r2ktime.c | 2 1 + 1 - 0 !
tune/blas/ger/s1nxtune.c | 2 1 + 1 - 0 !
tune/blas/ger/s2nxtune.c | 2 1 + 1 - 0 !
14 files changed, 14 insertions(+), 14 deletions(-)

 fix typos
missing cflags.patch | (download)

makes/Make.mmtune | 2 1 + 1 - 0 !
1 file changed, 1 insertion(+), 1 deletion(-)

 add missing cflags when building xmmgen_sse
ppc64 endianness.patch | (download)

CONFIG/src/SpewMakeInc.c | 2 1 + 1 - 0 !
1 file changed, 1 insertion(+), 1 deletion(-)

 fix endianness detection on ppc64*
 The __ORDER_LITTLE_ENDIAN__ macro is defined even on big endian systems.