[SCM] multiprecision integer library forked from gmp annotated tag, mpir-2.6.0, created. mpir-2.6.0
gladman
brg at gladman.plus.com
Thu May 30 17:51:24 UTC 2013
The annotated tag, mpir-2.6.0 has been created
at 39d5dfd0edb4a0c553c8c4a26471b9b9e9ee9ff0 (tag)
tagging a373abbc93aa7a65f8d7139bce801356db07316b (commit)
tagged by William Hart
on Mon Nov 26 00:36:14 2012 +0000
- Shortlog ------------------------------------------------------------
MPIR 2.6.0
(no author) (305):
checkin
Testing SVN access policy
Added text of LGPL and GPL v3.
Moved LGPL and GPL v2/2.1 from COPYING and COPYING.LIB to separate files.
Moved LGPL and GPL v3 into COPYING.LIB and COPYING.
Changed documentation (mpir.texi) to read GPL and LGPL v3.
Temporarily committing these so they can be converted to yasm format.
Added divrem_1 and divrem_2 x86_64 assembly code from GMP, replaced
Replaced nhgcd2 with the GMP hgcd2.c file, removing the matrix1_vector fns and
Correct minor error in Windows build
Fixed a bug in mpz/mul.c which caused it to always call the mul code
New tuning values for core2/penryn.
Convert divem_euclidean_qr_1.asm to yasm format.
Convert divrem_2.asm to yasm format.
Added a clarification to the copyright.
1. Update g2y.py to cope with GMP assembler
Hand tuned SQR thresholds for core2/penryn.
Fixed some SQR_THRESHOLD typos in gmp-impl.h.
Add HAVE_NATIVE defines for Windows code
correct bug in new assembler code
Fixed a bug in toom7_sqr_n.
Fixed some bugs in mpn_sqr_n to do with SQR_THRESHOLDS.
adjust windows tune slightly
Add mp_bitcnt_t type to MPIR
1. Add missed mp_bitcnt_t changes
correct t-headers.cc
Restored the t-headers test.
Another minor change to core2 tuning values. Seems to slightly improve
Add another missed mp_bitcnt_t change
Moved dcpi1_dc_div_q.c from GMP to dc_div_q.c (modified trivially).
Fixed a segfault.
Added sb_div_q based on dc_div_q code.
Added sb_div_q, sb_divappr_q from GMP and tdiv_3by2 and invert_1 from GMP
Fixed some typos.
Restored some of original GMP code.
Removed and error.
Added sb_div_qr and dc_div_qr from GMP sources. Fixed some
Autoconf and prototype updates for previously added code.
Fixed tests so they pass.
More fiddling with tuning.
Fixed a bug in the sb_divappr_q test code.
Fixed a bug I accidentally introduced into the code.
Added tdiv_qr.c from GMP.
Add new division code to Windows builds
Add clears and inits functions for mpf, mpq and mpz types
Added asymptotically fast division code.
Fixed some typos.
Add three new divide functions to Windows builds.
Fixed some segfaults.
Added a test for inv_divappr_q_n and made it pass.
Added test code for inv_divappr_q_n.
Fixed approximate quotient code.
Added code for balanced and unbalanced division with remainder using a
Added configure and prototype infor for new inv_div_qr(_n) functions and
Turned on new inv_div_qr division code.
Turned on inv_div_qr code in the tdiv_qr function.
add latest division code to Windows build
Added GMP rootrem code and added crossover from old code, which is faster.
Made some minor changes to rootrem code.
Ran autoconf, connnected up rootrem code and changed tuning value.
Fiddled with tuning value.
Added GMP mpz root code.
Corrected small root case.
Whoops, correct a typo.
Fixed a missing mpir.h.
Removed mpz_nthroot.
Ran automake.
Removed more refs to nthroot.
Minor cleanup.
Removed an unnecessary line.
1. remove C99 out of order declarations in rootrem.c
add new (test inv_div_appr_q_n) to Windows build
correct bug in rootrem.c
Improve correction in rootrem.c
remove debugging stub from rootrem test code (accidentally left in place)
Added toom8 files.
Added toom8 files to configure.
Added a missing function to configure.
Added missing function and made some corrections.
Finished putting too8 mul in.
tune for windows nehalem
Fixed multiplications being too unbalanced for Toom8h.
Turned on toom8 squaring code.
Added a missing mpir.h and some proxy tuning values for core2/penryn
1. Add toom8 code to Windows build
Added speed, tune and try code for toom8_mul/sqr and tuning for K102.
Added tuning info for core2/penryn.
1. Remove duplicate logic defines from gmp-impl.h
Update Windows tuning files (only nehalem done though)
Add logic functions in C (logops.c) to the Windows builds that need them (32-bit & generic C)
Attempt to tune some of the division functions.
Ran automake.
Hopefully fixed tuning of division routines.
Added more tuning for division functions.
Added some tuning for core2/penryn and some additional functions to be
Added further functions to be tuned.
Added some more sensible starting points.
Retuned for core2/penryn.
remove use of of GMP function not in MPIR (mpn_bdiv_dbm1)
Fixed a tuning bug and tuned for K102.
Tuned yet again for core2/penryn.
Added a new test for sb_div_q.
Ran automake and corrected test code.
Added a test for sb_div_qr.c.
Ran automake.
Switched sb_div_qr test to test the right thing! Added test for dc_div_q.
Ran automake and fixed some bugs.
Added test for dc_div_qr.
Ran automake and fixed a bug in the test.
Added test for dc_divappr_q.
Ran automake and fixed the test.
Added test code for dc_div_qr_n.
Ran automake and fixed the test.
Added test for inv_divappr_q.
Fixed a bug in inv_divappr_q and added a test for inv_divappr_q.
Fixed a bug in inv_div_qr.
Fixed more bugs.
Add division tests to the Windows builds
Minor removal of comment.
Corrected some bugs in the division code.
Added some extra checks to prevent infinite loops and did some minor
Retune for Core2/penryn.
commit 0ad3dfc64c2bbf0d005de3e02fe0658cd2eebb38
Fixed some incorrect things in the division code.
Added an itch function.
Improve memory usage of dc_divappr_q_n.
Add mpn_sqr, mpn_neg and mpn_com for GMP5 compatibility
add new C file for compliment
Add GMP5 mpn_get_str and mpn_set_str to MPIR
Added test for inv_div_q.
Ran automake.
Added test for inv_div_qr.
Ran automake.
Added test for inv_div_qr_n.
Ran automake and fixed a typo.
Added test code for tdiv_qr.
Ran automake and fixed the test.
Added mulmid_basecase assembly code from David's patch.
Added mulmid_basecase assembly code and fixed a few build issues.
Switched over to the GMP dc_divappr_q_n function.
Switched ove to GMP dc_divappr_q_n code.
Fixed a typo and reverted a change.
Reverted back to David's dc_divappr_q_n code.
Add new division tests to Windows build
Add mulmid_basecase assembler to Windows x64 builds
Update Python gas to yasm converter to remove some poor conversions
Change MPIR v3+ tuning for mpn_get_str and mpn_set_str to match GMP5
Add extra file needed for tuning mpn_set_str
Added missing mpir.h.
Added set_strp.c to Makefile and ran automake.
Added sb_bdiv_q and sb_bdiv_qr from GMP.
Added prototypes and ran autoconf.
Added test code for sb_bdiv_q (which does not pass)!!
Fixed bug in test code for sb_div_q and fixed comment in code for
Added test code for sb_bdiv_qr.
A tiny typographical correction.
Add new division code to the Windows build
Changed the interface of the sb_bdiv functions and moved David Harvey's dc_bdiv
Made dc_bdiv_q_n compile.
Added test code for dc_bdiv_q_n.
Added dc_bdiv_q/qr functions from GMP.
Made dc_bdiv functions build.
Added test for dc_bdiv_qr_n.
Added test code for dc_bdiv_qr.
Added test code for dc_bdiv_q.
Added tuning code for dc_bdiv functions.
Partly fixed tuning code for dc_bdiv fns.
Finished tuning code for dc_bdiv fns and tuned for K102.
Tuned for core2/penryn.
add latest division code and tests to the Windows build
Added new mpn_divexact function.
Fixed some bugs in divexact.c
Fixed more bugs in divexact and ran autoconf.
Added GMP mpz_divexact code to MPIR.
add divexact.c to Windows builds
Fixed some more bugs.
Reverted changes to division cutoffs.
Fixed K102 tuning values.
Removed new_bdiv_q which has been superceded.
A small typograhical correction.
Attempt at speeding up inv_div_qr_n (probably doesn't pass test yet).
Fixed the bugs in my speedup of the inv_div_qr_n code.
Made some corrections to the tuning code.
Core2/penryn tuning.
Added GMP itanium assembly code.
Added tuning for ia64 (tune did not complete).
Added mpir.h.
Added mpir.h.
remove use of __GMP_BITS_PER_MP_LIMB (obsolete define) from t-constants.c and Windows build
remove duplicate mpn_com_n define and fix 32-bit assertion bug in mul.c
removed spurious param for toomX_sqr for try.exe
improve g2y.py - the gas to yasm translator in Python
Further update to Python translator
Further revamp of Python gas to yasm translator
Add GMP compatibility defines and gmp_errno to MPIR
remove mpirbench from tarball trac ticket 275
enable applenopic for more p6 type cpus
add core to apple darwin apllenopic selection
Made some improvements to the arithmetic in toom4.
Slight speedup of toom4 code.
Better tuning values and commit a modified file.
Lowered the iterations on two well tested functions.
Removed toom7 code.
Temporarily switched to sb_div_qr so that sb_divrem_mn can be removed.
Removed sb_divrem_mn
Remove dc_divrem_n.
Rearrange tuning code and retune for core2/penryn.
Tuning for K10 and fixed a tuning bug.
Removed old sb_divrem_mn and added copyright to sb files I've touched.
Some commits appear to have been missed.
Tuning for ia64.
Add latest changes to the Windows builds
update benchmpn
update benchmpn
Update tune build on Windows
correct bugs in building tune on Windows
replace troublesome TMP_SALLOC_LIMBS uses with TMP_ALLOC_LIMBS
corrected bug in mpz_probable_prime_p , forgot to reset t
Correct Windows build of tune
up the trial div limit in mpz_probable_prime
correct mpz_probable_prime bug , never tested last condition
Fixed some bugs in the division code.
Fixed some test code so that asserts wouldn't fail.
Fixed some broken asserts and test code.
Added some checks for nocarry.
More ASSERT_NOCARRYs.
More ASSERT_NOCARRYs.
More asserts.
Another assert.
Another assert.
Made a small change to inv_divappr_q_n.
changed order of the tuning params for mullow,high,mulmod so that they are after fft tuning
Another change to asymptotically fast code.
Another small change to asymptotically fast division code.
Corrected some asserts.
More test code fiddling and asserts.
Fixed some bugs in the division code.
Removed a rubbish assert.
New asm ia64 mul_2 addmul_4
add divrem_2 to the fat structure
correct minor Windows build configuration issue
asm ia64 logic functions
further minor change to Windows build
mpz_probable_prime_p update
ia64 addmul_4 adjustments for icc
add -fpic to itanium icc build
set TOOM8 square minimum size in gmp-impl.h to 58
Reverted some changes to the toom4 interpolation.
Added some further default cutoffs and fixed a bug in mpz/mul.c.
Went back to new toom4_mul_n.c.
Changed to new cutoff names in test code.
params for atom
mpn/x86_64/k8/k10/gmp-mparam.h from cuda1
mpn/x86/pentium4/sse2/gmp-mparam.h from cicero
mpn/x86_64/core2/gmp-mparam.h from eno
mpn/ia64/gmp-mparam.h from cleo
mpn/x86/core2/gmp-mparam.h from menas
mpn/x86_64/atom/gmp-mparam.h from atom
mpn/powerpc32/gmp-mparam.h from varro
mpn/sparc32/v9/gmp-mparam.h from T2
mpn/x86/k7/mmx/k8/gmp-mparam.h from flavius
mpn/sparc64/gmp-mparam.h from T2
mpn/x86_64/core2/penryn/gmp-mparam.h from mod
mpn/x86_64/k8/k10/k102/gmp-mparam.h from lena
mpn/powerpc64/gmp-mparam.h from varro
Tune Windows Core2 build (using Linux/GCC values)
mpn/x86/k7/gmp-mparam.h from box4
mpn/x86_64/k8/gmp-mparam.h from box2
copy old fft tables into the updated gmp-mparam.h
Copy some tuning parameters from Linux/GCC to Windows
Updated library and version numbers.
Tuning for netburst.
Changed gmp version number to 5.0.1.
Tuning for ARM.
Swapped over to the new doc license and did a major revision of all the
Added some clarifying remarks regarding licensing to the Windows
Added mpz_inits/clears to the linux build and docs.
Added mpz_nthroot back in with tests code and correctly added inits and
Added all the mp_bitcnt_t stuff to docs.
Added mpf/mpq_inits/clears to the linux build, library and docs.
Made some more doc changes, esp. to out_str and get_str functions.
Added numerous missing things to mpn docs.
Added note that the factorial computation uses an algorithm of
Added actual reference to book or Schoenage et. al.
Updated the docs to include a note about Solaris SPARC.
Add missing C file and repair Windows builds
Aupdated the algorithms section (to some degree).
Replaced some GMP's with MPIR's.
Updated contributors section of docs.
Added a couple of papers.
Change handling of VERSION in Windows (and the setversion script)
Updated NEWS, AUTHORS and ChangeLog.
Replaced some gmp.h's with mpir.h's.
Correct win32 build error
add gpl/lgpl legal text to make dist
Added Eiffel interface and minor correction in contributors.
Added tuning for i7 (thanks to Gonzalo Tornaria).
Added note in AUTHORS.
retune atom params
Add nehalem tuning for Windows
Added some #defines in for deprecated and removed functions.
Changed one of the defines for something better.
update setversion script
remove Windows issue with VERSION definition
Testing new setup
brgladman (19):
1. longlong.h change to add MSVC intrinsics
Add MSVC build files
minor MSVC build fix
minor MSVC build fix
Apply GMP 4.2.1 patches to MPIR
Correct wrongly applied patch
Improve AMD64 sgr_basecase
AMD64 assembler code tweak
Rename output files to their MPIR names
commit 075a98ed18614ea76660db8f558a2562f0c48576
Update x64 assembler to use prologue/epilogue macros
commit 3ea0d1f0fc3401ddda2a08ed21af3076e2ac670e
Minor Windows assembler update
commit c6b1b8daaf71c7e0f95248aef57f08f845b0f630
Correct VC++ project dependency settings for core2 builds
Update VC++ build to add new version numbers
Correct VC++ derivation of gmp.h and mpir.h
minor VC++ build update
Update Windows build for Jason's gcd additions
dmharvey (4):
create separate trunk and branches directories
documentation changes (see #79)
remove some auto-generated files from yasm tree -- see #52
rewording (trac #78)
gladman (520):
add windows assembler include file
Enable intrinsic functions in Windows build projects
Update intrinsic settings in Windows build projects
New repository commit test
update gmp.h generation for Windows
Update some text in Windows x64 assembler files (no code changes).
Update Python test program for tests on Windows
Tidy up the Windows build projects and revert a broken gmp-mparam.h file on experimental branch
Further tidy up of AMD64 Windows build and added g2y.py for converting JM's assembler files into YASM format
improved support for assembler translation to YASM format on Windows
Improve Prologue/Epilogue macros in Windows assembler
correct minor errors in windows build
Update copyright notices in Windows assembler code
Minor Windows build update to explicitly set the compile language for *.cc files as C++
commit 43160f61a6c17bd9f91a747c87e7450523c4970c
Bring Windows up to date on new trunk
Update windows assembler to remove alignment padding problems
Major Windows Commit:
commit 694af0fc3498eddee88d777378af3888b0cbb8d1
Windows tuning update
Correct Bug in 0.9.0
Commit revised windows assembler code
commit 9cdf5f371ae69811c4a21dddba30eb470ca25c08
update g2y.py for 3 op instructions
g2y.py (gas to yasm) enhancements
add missing Windows AMD64 assembler file to SVN
Major tidy up of Windows x86_64 assembler code
correct core2 Windows assembler error
update Windows build projects
1. Minor windows build update
improvement of Windows Core2 assembler
revert try.c that I broke
1. Tune Windows builds
1. Add TRY to Windows build
correct 32 to 64 bit sign extension in Windows assembler code
add new speed designator to command line
avoid VC++ optimisation issue in speed/tune
correct copyright notice spelling errors only in Windows assembler (caused by rsi -> rdx substitution)
Two minor Windows assembler changes plus code formatting to remove tabs and trailing white space
change Core2 assembler code for mpn_add_n and mpn_sub_n.asm
commit b4e75f80f076f3884264815d0a206cc808172a5a
Update VC++ readme and project files to cover the AMD K10
commit ee52e625c58a9e0d28628a9adb2b906f4cf821b2
minor update to runtests.py
1. Update Windows version of getrusage
update an orphan Windows assembler that got missed first time round
1. Add Windows code for latest trunk revisions
Update Windows build system
Update Windows build for latest changes
Adjust Windows assembler files to assist planned automation of aspects of the Windows build
Revised TOOM4 and TOOM7 thresholds
Update Windows build for latest code
Add missing files for tune on Windows
Add Windows assembler files to trunk
commit 255c8c255ca730ac7cd18cc1fa4821450e660375
Windows Core2 tune
minor cosmetic update to g2y.py
Commit missing asm files
commit missing Windows assembler file to trunk
commit a missing windows test
Change Windows test projects to provide full build in release mode
Make two 'ease of use' changes to the Windows build
Adjust Windows build slightly
commit d18fab116faa011d529dc77970b40d5d3d24df84
Remove C99 features in mul_fft.c
update g2y.py to allow command line parameters
1. Update g2y.py, the GAS to YASM Python script
Add Windows build files to bench in trunk
1. Update Windows Powershell scripts in mpirbench to refer to MPIR rather than GMP
correct erroneous Windows addition to gmp-h.in
Update Windows K8 build to add new assembler
Further update to Windows K8 build
1. Add new/changed Core2 assembler files to the Windows build
Update MPIR version number to 1.2.0
1. revert t-fat and remove fat test from Windows testing
commit 2768eeaaf07e1bb7db80e1f5faa1e5a7613ba8c5
Adjust Windows build to remove assembler that is slower than VC++ intrinsics (copy, move, set)
1. Fix minor bug in Windows build
1. Improve g2y.py
Update windows build for latest assembler
commit dea89e476083a971c20ee25619b4e8612f9203fc
commit 2bf84f6a747273c1c669a06bd07364516b0dea79
commit 0d90e4faa825b9c034d867fe72894e32617eee53
commit b96d7f466bc3de9e58b5e3485dbeb0b3a7111053
1. Minor (cosmetic) change of a Windows aspect of speed.h
1. Windows Core2 tuning with Linux FFT tuning values
Record Windows Core2 FFT tuning values but don't use them
Update VC++ build instructions
remove C99 stuff
Update Windows build for recent added and deleted C source code files (assembler not done yet)
Update Windows build for 32-bit versions
Add new assembler files - new (l/r)shift2 on k8, modified (l/r)shift on Core2 - and sb_divappr_q.c to the Windows build
Correct Windows assembler for lshift2.asm in x86_64w\amd64
commit 558c10cba31ef8b285a7ab3270b09244eff1d541
add missing test project files
commit b1bc208b963d83e0e5f9bd7e1551b8184f32cb7a
Take out old (obsolete) copy operation in MPIR tests on Windows
Bring Windows build up to date
Add mpn_add_nc and mpn_sub_nc to the Windows Core2 assembler
Move latest rshift assembler code for k8 to Windows AMD64
Update Windows builds for latest changes
Add new Core2 assembler to Windows build
Correct latest Windows assembler for mpn_copyi
add mpz_nextprobprime.c to WIndows builds
remove bug in g2y.py
further change to g2y.py
1. Add new x64 assembler functions to the Windows build
1. update Core2 x64 build
commit 70a4a3823fb541c4b06d87b9a6171a80dd53f1cc
Update Core2 x64 windows assembler
Add Jason's new root code on Windows
Add new files to the Windows builds
On Windows DLLs export mpn_cls_tab unconditionally for GMP compatibility
Update Windows build for recent additions
Adjust Jason's VC++ build batch file for CPP -> CXX name changes
correct the currently unused code path in next_likely_prime.c (tested and now working)
Add new assembler, C and test files to Windows builds
commit f3b7ed1f2ea96194c488b9077bd04b282a5aae67
1. Add generic C mod_1_<n> files to the Windows build
Update Windows timing code
Adjust windows timing routines
commit 8d6ef07ef6e57164873ab9c5603400751a1a012f
Add mod_1_<n> assembler to Windows AMD64 builds
Python tests update
Add new k8 assembler to Windows builds
Add mod_1_3 assembler to Windows
Core2 assembler (copies) for Windows
remove bad exit behaviour in Python test script in interactive mode
Add latest division code to the Windows builds
commit 448c60fd7757be371c3dbbee271ba3e6b47c1057
correct windows build error in a DLL build
fix minor pre-build warning on Windows.
update VC++ readme.txt and Core2 tuning
correct minor issues in windows build files
Commit Windows Project Files to sync the repository
Update Windows build of MPIR tests
Update Windows MPIR build for tests
commit 044ee0f7f5e1b035c46d14f1802fef621b148433
commit 769271f9e45f4c7c15e5be0699e1b9dcfa3ad9e9
Correct test dependency
Bring Windows naming into line with that on Linux
commit b086474ddffeaf87099073ece4dbbf9b802dbbf8
1. Add Jason's new assembler code to the Windows builds
Add Windows k10 build projects
commit 5093673876342b4e4e1b712a8be2864634c74572
commit 766a5fb1dc193d35935830ea0372a87ca4c71006
commit e3f6129a64e350edf988a04370acf8fe7185d656
update Windows readme.txt file
Add Windows build for Intel nehalem
commit 11c4a695723ef988a2ec4c8b4fa20f1acbc71a4a
1. Add a new test to Windows build
Take out Windows line endings
Adjust likely_prime test for Windows
remove bug I added to t-mpz_likely_prime_p.c
add missing line to sqrt in likely_prime_p.c
1. Add mpn.invert test to Windows
1. update run-test.py for Windows build
correct bug I introduced into invert.c
1. Change Windows memory management to use alloca()
1. Set mp_size_t & mp_exp_t as signed 64-bit integers on Windows x64
1. repair Windows build script
Correct a typo in the win32 version of count_leading_zeroes
Add missed C file to Windows builds
Update Windows build to cope with huge numbers
Small improvements in Windows build projects
1 Add Windows extern inline define to mpir.h (but don't use it)
Add mpn_zero to Windows builds
Add new GCD test to Windows build
Add Core2 and i3/5/7 (nehalem) tuning for GCD to Windows builds
Update Windows tuning values
1. Add mpn_copyi and mpn_copyd source code files in the mpn\generic directory
Add Visual Studio 2010 builds
update Jason's configure.bat for Visual Studio 2010
update Jason's make.bat for Visual Studio 2010 (not quite working yet)
update to assist Jason's command line build for Visual Studio 2010
remove bugs from Windows command line build files
Amazing - Visual Studio 2010 has stdint.h - update builds accordingly
1. Correct Windows build error for logical operations (VS 2008 and VS 2010)
minor (cosmetic) change to Windows VS 2010 build
remove unwanted file from VS 2010 build
Remove need for logops.c on Windows
minor tweak to logops_n.c
Correct setversion to set the Windows version number string (and current gmp-h.in)
1. Correct assembler build rules and target directories (cosmetic)
Allow YASM path to be set by an environment variable (YASMPATH)
Further change to vsyasm.props
Add x64 configuration to Windows prebuild steps
commit 274e99715531e7140ce60502f768133a51765e7a
Improve Python detection in Jason's Windows batch build
Remove serious bug in Windows K8/K10 assembler code
correct k10 build on Windows
minor update to VC++ readme files
Add Python programs to support VC++ 2010 Express x64 builds
update VC++ readme to cover VC++ 2010 Express
correct Python files for VC++ Express conversion
Update the Python gas to yasm translator for Visual Studio 2010
correction of mpn_store on Windows
1. Update Windows k10 build to match Linux
remove prebuild steps on Windows
commit 9fc76411c8170e4a362ed37d82de1f8175f98ef1
a small change to g2y.py tot test a commit
Remove three of five prebuild steps from the Windows vc9 build
1. Remove all WIndows prebuild steps
tidy up documentation of this function
remove unused include directories in Windows build
remove prebuild steps for Windows VC9 builds
simplify the mpir.h generator batch file gen_mpir_h.bat
move fib_table.c and mp_bases.c into mpn\generic and add them to Windows builds
Remove pre-build steps in Visual Studio 2008 Windows build
1. Change symbol mpn_divexact_bybM1of to mpn_divexact_fobm1 in *.c, *.h, *.in and config.* files
Minor correction to function renaming changes
Change "Bm1of" to "fobm1" in text files (hopefully fairly safe)
commit 700ead092cbcfd301e030be4fb90ef8fbcaf6da2
Unify Windows build configuration for tests
Add _GMP_DECLSPEC to symbols in gmp-impl.h needed for the tests
Update the Windows tests to allow MPIR DLL builds to be tested
commit 6be1fd6bc63cf998b41999a5e27f9d60140767d6
Make cosmetic change to Windows builds
Further simplification of Windows builds
Further simplification of Windows build
Further simplification of Windows build
Update the VC++ readme.txt
Move Windows output directories to mpir\lib and mpir\dll
Move Windows output directories to mpir\lib and mpir\dll
Update the VC++ readme.txt
remove old library output directories
Ensure configuration of Windows tests match last MPIR build configuration
Serious reversion to cancel last 24 hours work as all win32 builds failed
Commit changes to allow Windows DLLs to be tested
1. Add config.h file generation for Windows
Another try at an improved Windows build for Visual Studio 2010
Update the Visual Studio 2010 readme.txt to outline the revised build process
Minor change to build of tests on Windows
minor tweak to batch file
minor tweak to batch file
remove ngcd.c from Windows builds
1. remove single epilogue restriction in Windows assembler code
commit e352970a1a15170bdaa1c1c97bad5ba84a9ccd89
correct a typo in two Python files
commit cba24f002b8fb8d05781c3f30905f9c72223416e
Make Changes to Windows builds for recent file name changes
Update Windows build for tune
Add new files to Visual Studio builds
Revise Visual Studio 2008 builds for recent changes (untested)
Change namespace back to old name in vc9
Revert some name changes in VS 2010 build
Change Windows assembler include file organisation (VS 2010)
Change Windows assembler include file organisation (VS 2008)
minor cosmetic correction
Minor Windows build change
Change Windows build to make nmake easier to use
Further adjustments to batch files to cope with an NMAKE based build
cosmetic change to postbuild batch file on Windows
split x86_64w 'aors_err' assembler files into add, sub and inc files
Update Visual Studio 2008 build for aors_err.. changes
cosmetic update
Correct a configuration error when tests on Windows are built in debug mode
Simplify Windows builds of MPIR tests by removing redundant compiler/linker settings
Remove redundant compiler options in some Windows builds
Reinstate the misc.locale test in the Visual Studio 2010 build only
Correct old YASM RULES file
Re-commit 'lost' changes
Commit missing mparam.h file
Commit trivial change to see if revision 3194 is then OK.
Add conditional definition of _WIN64 for Windows x64 (for safety)
Update _PROTO definition to recognise MSVC
remove dangling reference in VS 2008 build
add Jason new code to Windows build
add latest assembler for nehalem to Windows
add new assembler for and and nior to Windows
Add new assembler for store on k8/k10 to Windows build
make redc changes for Windows
more redc stuff for Windows
try again
Ho redc changes in Visual Studio 2008 (untested)
more
minor update to windows build for redc changes
minor build change to output which connfiguration is being built
1. Add Jason's fantastic new mul assembler to Windows
add misssing DLL export definition in gmp-impl.h
typo in source code
Minor typos in source code
add nehalem mul_1 to windows
add latest mul_2 assembler to Windows nnehalem builld
add latest mul_1 assembler to Windows build
Add single file changes to Visual Studio 2010 build
1. Review all Windows Visual Studio 2010 builds
Remove outdated Windows builds for Visual Studio 2008
get the Windows batch build 'closer' to working order
improve pre and post build steps for Windows build
minor cosmetic change to Windows build
Minor Windows build change
Add some new defines for config.h now that VC++ has stdint.h
1. add latest assembler code to the Windows build.
minor change to assembler code for mpn_not
Tweak Windows assembler code for mod functions
Windows assembler correction
update slow assembler code on WIndows
correct typo
correct some names in a Windows batch file (no functionality change)
commit a3d493377745dca61c343d15e0da940211c34ab3
correct cosmetic typo
Add an alias used by MPFR (snprintf -> _snprintf)
commit f5322ec3a2b82a934328fa1e310d86f631943ce3
commit 7dde675da1051ede8e9ee98ab2fcc53e7dca1b4e
minor text change
correct errors in new (u)intmax functions and add two more tests
change mechanism for detecting stdint.h
cosmetic tidy up of Visual Studio build files
Add tests for mmpz_get/set_ux functions
commit 2489506c00d880a377662a442380f16a01361b17
commit 12ea51f4c7100be1d6706b06d436666c109f687e
Tidy up Visual Studio builds (cosmetic)
make changes for building longlong.h
correct name of longlong.h in mpn\x86_64w\ to longlong_inc.h
correct incorrect parameter type for mpz_cdiv_q_2exp
add bswap limb definitions to MSVC longlong_inc.h versions
define __GMP_BITCNT_MAX (and use it in mpz_popcount)
correct errors introduced when mp_bitcnt_t was added
correct windows byte swap intrinsics
remove obsolete define for _WIN64_ABI
Add a version string for the Microsoft compiler
Correct functions returning mp_bitcnt_t type to return correct error return values on Windows x64.
add karaadd and karasub assembler code to k8 and k10 Windows builds
add latest assembler code to nehalem and core2 builds
change labels from global to local in latest Windows assembler code
delete obsolete file
1. rename umul.asm and umul.asm to align with the symbols they export
add latest nehalem code to Windows build
add double and half assembler to the Windows builds
correct error that only impacted on building for Windows with MSBUILD
correct error that only impacted on building for Windows with MSBUILD
update latest assembler for nehalem on Windows
Update latest core2 assembler on Windows
add mpn_sumdiff_n for nehalem on Windows
Add sumdiff_n for core2 and penryn and mod_1_1 for penryn on Windows
minor tidy up
remove HAVE_HOST stuff in build.vc10
correct FP changes for Windows
Add experimental Python VC++ build file creator
correct typo
commit e425019812245642461047b173aa7c078261b65e
commit 1cff1804928068bf3a511bd3688ac7da03bf961c
bring into full sync with my current code
remove unnecessary files when new GCD is used
update tuning for new GCD on Windows and add new tests
update all VC++ builds for new GCD code
correct minor build error in VC++ builds
Add missed tests for new GCD code
commit new GCD code to trunk
1. Correct errors in Visual Studio builds
small improvement when running tests on Windows
Add 64-bit integer conversions to mpirxx.h
Windows x64 build configuration change (_WIN64 must now be defined for all x64 builds)
update Python Windows build generator
Copy latest tuning to Windows builds
bring new code in mpz.next_likely_prime into use and add a test
change gmp_h.in to reflect mpz_next_likely_prime API change
change nextprime.c to use new gmpz_next_likely_prime API
copy Jasons tuning to Windows
copy Jason's latest tuning to Windows
1. revert mpz_nexxt_likely_prime to previous API
Update old Miller Rabin code and mark it obsolete
change a local function name to avoid potential name conflicts; add copyright
Create a branch for version with choice of integer types in ui/si functions
populate new branch
correct some build errors
correct some build errors
More build corrections
More build corrections
More build corrections
minor build correction
Major build change to prepare for Visual Studio 11
minor build correction
add some comments to the Python Visual Studio build generator
correct mpq_cmp_ui declaration error found by David Cleaver
move Visual Studio lib and dll output directories from mpir\build.vc10 to mpir
correct python build generator
correct error in python build generator
commit 2d66eb20c5e3878d61e348346172333869273ca9
correct errors in the integration of the new FFT code
Further corrections to FFT code integration
Further corrections to FFT code integration
Further corrections to FFT code integrationcommit FFT test source code
commit FFT test source code
commit FFT test source code
commit FFT test source code
commit Visual Studio builds for the FFT tests
minor change in printf output
update the FFT multiplication wrapper
put FFT mpn_ files back in FFT directory
change some names in the FFT code
FFT name changes and tuning
correct rename error and change errors outputs to output limbs correctly
add square tests
commit changes for efficient squaring
add FFT tuning code
commit 7464db57445fddf7ce31c081def78e1a49d610dc
nehalem desktop FFT tuning
sync with trunk
commit eca7796ff87d273f8c1a82d8b15477642a60ec2c
1. correct FFT code
commit latest FFT changes
Final integration of Bills new FFT code
use HAVE_NATIVE defines in FFT code
add missing file to SVN
correct a bug in the new FFT code
Bring FFT calling interfaces into line with GMP/MPIR conventions
correct call to old FFT from new code (for timing old FFT only)
set zero for output when bits = 0
minor updates to FFT code
Change gmp_ui/gmp_si types to mpir_ui/mpir_si
correct bug in WIndows assembler code for karasub
minor documentation update for karasub assembler on Windows
karasub documentation correction
correct the Windows versions of karasub assembler
improve documentation in karasub assembler
small change to Windows karasub asseembler
correction to karasub assembler documentation
adopt Jason's six loop solution for carries in karasub assembler
Correct the Windows kara_sub assembler implementation.
correct minor Windows Visual Studio build error
commit c3282496908b9934fee9cb307b259054a6352443
update win32 builds
correct python configuration generator
prepare for new Python build system
commit 406c8e8b24a026fb16f6a6aef0fad0bda1403600
correct prebuild step
new karasub C code and a Windows postbuild tweak
minor windows build correction
minor build change
minor configuration change
add bug fixes from trunk
update Windows Python build generator
add latest trunk changes to exp branch
minor changes to add comments for symbols defined in macros
minor changes to add comments for symbols defined in macros
adjust VC++ build for some latest changes
Turn off stdint.h if Visual Studio earlier than 2010
Turn off stdint.h if Visual Studio earlier than 2010
add redc_2 file to the Visual Studio builds
correct minor error in VC++ ttune build
minor change in mpir_config.py
commit 6158abe9bbd9075f38d8e4902bc0b0f96fb71b43
add latest gmp-impl.h changes
correct 32-bit build error reporteed by Case
minor VC++ build generator change
commit 53d235597bc911e64fb2f1dd652e585565cbb576
rebuild all Visual Studio builds to remove win32 build errors
correct bug introduceed in mpir_config.py
correct bug in run-tests.py
commit 98529cded9f13c8d098ceb70c920e9052039ad70
Bring the Windows prebuild and postbuild files up to latest versions (from exp branch)
fix aa few typos
adjust a few casts
minor windows build corrections
adjust python build generator
Add latest bugfix to FFT code
improve the Windows build generator
improve the Windows build generator
improve the Windows build generator
improve the Windows build generator
further improvements to Windows build generator
add signal/signal.h to Windows config file
correct Windows p0 build and add signal.h to Windows config file
correct Windows p0 build and add signal.h to Windows config file
remove a few duplicate PROLOGUE lines
remove some directories that are not needed
Add recent bug fixes to the FFT code
complete new FFT and strip out old one
convert mpn_mulmod_2expp1 to mpn_mulmod_Bexpp1 and remove test of the original version
minor changes in Windows build
add minor changes from trunk to Windows x64 assembler files
add some trunk changes to exp branch
update Windows build files
commit changes to remove conflicts after merge from trunk
merge mpir-exp branch into trunk
replace a few missed longs with mpir_si/mpir_ui types
remove windows builds no longer supported by default
A further update to the Visual Studio build
add HAVE_LITTLE_ENDIAN define to Windows configuration files
change some longs to mpir_si in tests
Add macros for __GNU_MP_RELEASE and __MPIR_RELEASE
Update to batch file build for MSVC provided by Case
move Windows command line build files
rename Visual Studio test projects to cut name lengths of fft projects
change 'truncate' in file names AND symbols with 'trunc' (except in mpn_invert_truncate)
revert invert_trunc back to invert_truncate
update Visual Studio build files for file name changes
change mpn_invert_truncate to mpn_invert_trunc for naming compatibility
remove some Visual Studio build files wrongly added to the repository
add some missing win32 assembler files
change the Visual Studio build generator to provide for building cfg.h files on their own
change the Visual Studio build generator to provide for building cfg.h files on their own
add build of cfg.h files to the Windows command line build (requires Python)
minor update to Windows readme file
correct bug in the Windows build generator
copy more headers to Windows output directories
Improve resilience of the Visual Studio build for test and tune projects
revert gmp-impl.h
Extend the speed and try programs on Windows to work with MPIR DLL libraries.
change Visual Studio build generator to allow multiple builds to be specified
change Visual Studio build generator to allow multiple builds to be specified
remove multiple definition warning in win32 build
add two defines (for Windows) now needed by MPFR
Clariify the vc10 readme.txt on the need for VSYASM, not YASM
adjust _MSC_VER guard in mpirxx.h
adjust _MSC_VER guard in mpirxx.h
remove config.h from tests that do not appear to require this
reduce dependence on config.h in tests
put config.h include back in misc.c
correct minor error in the Windows build for tuning
move new *nix tuning values into windows
move new *nix tuning values into windows
change 'unsigned long' to 'mpir_ui' in function declaration
add section on the handling of long integers on Windows x64
remove redundant FFT tune Visual Studio project
remove redundant FFT tune Visual Studio project
add cxx files to windows command line build for DLLs
export a few more DLL symbols on Windows (needed by try).
jasonmartin (9):
Added GLOBAL_FUNC assembly macro
Adding file add_n.as in core2 assembly code
Adding file sub_n.as in core2 assembly code
Adding file addmul_1.as in core2 assembly code
Adding file submul_1.as in core2 assembly code
Changed core2 Assembly code to LGPL license and made some cosemtic changes
Converted core2call to yasm/Intel syntax
Fixed bug in yasm_mac.inc to allow global functions to have labels without __g prefix
Incorporated Niels Mollers lgpl 2.1 gcd code into trunk
jasonmoxham (885):
result after running aclocal,autoconf,automake,autoheader and rm -r autom4te.cache/
mpz_perfect_power_p corrections and new test program
A few more tests for mpz_perfect_power_p
reverted to rev 1593
cp AUTHORS,README to build.vc9/gnu.license/ and run autoconf,automake
for file in $(find -name \*.c ) ; do sed -e "s/#include \"gmp\.h\"/#include \"mpir.h\"/g" $file > temp ; mv temp $file ; done
for file in $(find -name \*.cc) ; do sed -e "s/#include \"gmpxx\.h\"/#include \"mpirxx.h\"/g" $file > temp ; mv temp $file ; done
changed gmp.h to mpir.h for a few odd cases left
changed gmpxx.h to mpirxx.h for a few odd cases left
for file in $(find -name Makefile.am) ; do sed -e "s/libgmp\.a/libmpir.a/g" $file > temp ; mv temp $file ; done
changed libgmp*.* for a few odd cases left
changed libgmp*.* for a few odd cases left
change libgmp_la to libmpir_la and libgmpxx_la to libmpirxx_la
autostuff and change name of file gmpxx.h to mpirxx.h
update configure
update info docs
merged k8-branch into trunk , autotools , few handbits , windows bits just copyed over
added include longlong to diveby3.c
todo
Copy AMD asm code to core2 and change inc/dec to add/sub
new fns lshift1,rshift1,addadd,addsub,sumdiff,divexactbyff,redc_basecase not visible to user only internally
set default speed/time precision from 10000 to 1000000
fixed fat build failure for dvexact_byff
add redc_basecase and sumdiff_n to the fat structure
update gmp-mparam for linux K8,K10,core2
trunk fat dummy functions mul/sqr basecase for x86_64 , and move amd64 copy asm to x86_64
trunk fat dummy functions redc_basecase,sumdiff_n for x86_64 and x86
update core2 cpu detection for fat
removed # comments from x86_64 assembler as OSX dont like em
movq to movd fix for broken old masm on OSX
merged buildtest branch into trunk for building outside the src tree , the command I used was svn merge -r 1643:1669 ../branches/buildtest/ in my local copy of trunk
buildtest merge copyi,copyd were missed as I moved them
removed dos crlf from linux asm files , update configure to recognize GLOBAL_FUNC for HAVE_NATIVE_functions
added lshift1 rshift1 divexact_byff addadd addsub to tests/try.c as tests
tests for redc , not yet complete
move amd specific copy back to amd dir , and write new fat fallback copy fn
cant spell
added include files
remove define test for copyi/d
remove old TODO , remove spare yasm_mac.inc link
remove crlf from k10 asm files
delete amd copyi.as and copyd.as
added redc_basecase tests to tests/try.c
cast mpn_addsub_b try tests to int
update try.c for 3 src functions
new configure option --enable-gmplink to install symlinks of gmp->mpir
correct path for yasm_mac.inc in tune directory
autoreconf
tr upper lower case quoting nightmare
fix for mpz_urandomm reuse error
Nehalem cpuid update
Atom cpuid update
autotools with ylwrap link
autotools with the missing ylwrap link
lahf and nocona hack... aaarrrrgggghhhh
copy add/sub from mpir-0.9/mpn/x86_64/amd/ to mpn/x86_64/core2/ for the nocona with no lahf
remove crlf from old add/sub_n and remove yasm macros from GLOBAL_FUNC names
removed space
update Makefile.am so that make dist works
make dist works
make dist now touches demos/calc/files
remove .svn files from make dist from build.vc9 directory
autoreconf
make dist to force chmod 755 on config.guess and friends
merged x86_64 cpuid branch into trunk with svn merge -r 1755:1779 ../branches/x86_64_cpuid/ run in my local copy of trunk
new directory and target make bench
change configfsf.sub to match already-canonicalized machine name *-pc
x86_64 mpn_subadd_n plus tests,tune
New assembler x86_64 mpn_mul_2
merge div-branch into trunk with svn merge -r 1782:1816 ../branches/x86_64-division/ run on my local trunk
autoreconf
added toom4_mul_n.c to tuning makefile so that threshold can be changed
update k10,nehalem,core2,penryn gmp-mparam.h
update gmp-impl.h for toom4 varible threshold
nehalem,k8 tune params
update gmp-mparam for k10,core2,penryn
remove un-needed case in mul_basecase.as for x86_64
run old autotools
autoreconf and update tune/Makefile.am for toom7
remove dead links for --enable-gmplink
remove bench from build
copy some core2 asm to netburst
copy more core2 asm function to netburst
change gmplink to gmpcompat and update docs etc
new x86_64 addmul_2.asm :note convert to yasm , update netburst gmp-mparam.h
add header
update gmp-mparam's
removed divrem_hensel_1 , save it for a rainy day :)
make dist-hook to change yasm/Makefile.in to stop install/check
put back bench into trunk
update docs
make dist-hook to add dummy alias make install-gmpcompat
change gmp to mpir in docs
removed divrem_hensel.asm
duplicate x86_64 mul_2.as to overcome fat issues
add divrem_euclidean_qr_1 , divexact_byBm1of to fat structure
fat bits
div update
divexact_by Bm1of try tests
divexact_by Bm1of try tests
autotools
converted addmul_2 to yasm
assertion correction in divrem_euclidean
update bits
update speed options -r and -d to accept any column
define gnu_source in configure.in
config.guess update for pathscale cc
updated speed to accept column sums
speed update
mpn_mul_2 doesn't do size=1
correct divexact_byBm1of speed measuring
New asm functions for mpn_copyi mpn_copyd for k8,k10,core2,penryn,nehalem
New asm functions mpn_add_n mpn_sub_n for Core2/penryn/nehalem
New functions mpn_sumdiff for core2/penryn/nehalem , or rather faster to do separate add and sub
New asm functions mpn_addadd_n mpn_addsub_n mpn_subadd_n for K8/K10
New asm functions mpn_rsh1add_n mpn_rsh1sub_n for K8/K10/Core2/penryn/nehalem
New asm function mpn_divexact_byff for K8/K10/Core2/penryn/nehalem
New asm function mpn_sublsh1_n for K8/K10
New asm function mpn_mul_basecase for K8/K10/Core2/Penryn/Nehalem
mul_basecase tweeks
New generic functions/macros mpn_lshift2 mpn_rshift2 and tests/speed etc
New generic mpn_store and tests/tune etc
New asm functions mpn_store MPN_ZERO for k8/k10/nehalem
Add new function generic mpn_divrem_euclidean_qr_2
divrem bits
divrem updates
divrem update
mul_basecase for GAS, so I know what the yasm conversion is of
mul_basecase to yasm
divrem_2 generic correction
New asm function mpn_divrem_euclidean_qr_2 for X86_64
divrem_2 speedup
Convert new divrem to yasm format
nehalem mparam update
k7 gmp-mparam.h
32bit pentium4 gmp-mparam
ia64 param update
update fat for divrem_2
int to mp_size_t fix
int to mp_size_t fix
sparc64 tuning
autoreconf
Sparc64 fft params
powerpc64 varro fft tuning
remove gnu_source redefinition warning
update config.guess to remove dummy file under cygwin/mingw
fix trac ticket 216 alias for getrusage not used properly on mingw
fixed configure to ignore mingw broken mprotect , run autotools
fixed trac 212 , missing entrie in fat structure
Configure and make mechnism for windows from windows branch
fix for windows make.bat so it can detect python in 64bit mode
make clean windows updates
add exit codes to windows batch files
Copyright header stuff
dos line endings
whoops managed to delete config.guess.c , add it back in
Update speed to accept columns from other files
corrected ratios/fastest (#) in speed when taking values from a file
New intel x86_64 assembler code for left/right shift
New K8 asm functions mpn_lshift2 mpn_rshift2
change gmp.h to mpir.h
New generic mullow and mulhigh
tuning for mullow and mulhigh
mullow mulhigh tests
remove old yasm
replace yasm with latest svn yasm rev 2220
yasm/config/config.sub fix , line 1076 , Make sure to match an already-canonicalized machine name. add *-pc to *-unknown | *-pc)
run yasm/autogen , make distclean , and add the files produced to svn
These files are also ment ot be part of yasm , they were made by autogen , but I deleted them thinking they were part of a failed make distclean
remove yasm symlinks
and replace with actual files
recognize bsd cpuid report of amd64 as a x86_64
fix BSD install problem with gmp-compat and cxx , and remove need for make install-gmpcompat , i thinkl
remove install-gmpcompat option from makefile
update manual to reflect make install-gmpcompat not needed
cast pointer fix trac issue 33
fix trac ticket 146 , ie clean up after a fat build
Update make check to test static versions , when we build both , of popham and fat
delete old yasm
replace with yasm svn 2227 , delete yasm .svn stuff , run yasm's autogen , replace symbolic links in config directory with real files , run yasm's ./configure && make && make check && make distclean (this does add a few files we need ) , rm autom4te.cache , add to svn
yasm/config/config.sub fix , line 1076 , Make sure to match an already-canonicalized machine name. add *-pc to *-unknown | *-pc)
t-mulmid mingw32 update
fix trac 234
fixed trac 176 version number have 3 parts
script for setting version numbers etc , trac ticket 174
when building both static and shared librarys , we can make "try" and "static_try"
fix trac 225 mpf_eq is broken
New generic function mpn_neg_n and tests
mpf_eq tests
Add Robert Gerbicz mpz factorial code + tuning threshold
New generic mpn_mulmod_2expp1 and tests
New generic mpn_mulmod_2expm1 and tests and tuning thresholds
fix cxx mpf_eq tests
mpz_fac include path when tuning
mpn benchmark
delete old functions from manual
Tweek K8 mpn_rshift
remove obsolete fns mpz_random mpz_random2 and all references to them
removed obsolete fns mpn_divrem mpn_divmod and all referances to them
delete obsolete fn gmp_randinit and all referances to it
remove gmp ver 1 src compatatible defines
remove some referances to the old functions
doc update
New K8 asm function mpn_lshiftc
autoreconf
have-native only bits
remove all referances to mpbsd
remove mpbsd from docs
remove ansitoknr stuff
New core2/penryn mpn_store
New asm function nehalem mpn_store
New asm functions mpn_copyi for core2/penryn
New path for K8/K10 in 32bit mode
New K8 32bit add/submul_1
New K8 32bit asm function mul_basecase
Obsoleting mpz_nextprime
replacement function mpz_next_probable_prime
Obsoleted mpz_probab_prime_p
add mpn_add_err1_n to speed
New K8/K10 asm function add_err1
New K8/K10 asm function mpn_sub_err1_n
windows batch build clean up make help
python script to run tests option for no enter
remove suprious error message from make clean windows batch buil
Obsoleted mpz_array_init
New path for core2 in 32bit mode , and select best asm function from what we allready have
correct syntax
save popham for later
put 32bit k8 in fat path
add generic new function HAVE_NATIVE function mpn_addlsh_n mpn_sublsh_n mpn_inclsh_n mpn_declsh_n
add new function core2/penryn mpn_addlsh_n
try tests for addlsh etc
core2/penryn new addlsh faster than old addlsh1 , so delete
macros for mpn_inclsh,mpn_addlsh1 in terms of mpn_addlsh etc
tweek macros
tweek macros
macro tweek
update try to cope with macros
New asm function nehalem mpn_addlsh_n , delete old mpn_addlsh1_n
New asm functions K8/K10 mpn_addlsh_n mpn_sublsh_n and carry-in varients
New path for Phenom II ie family 16 model 4 , we call it K10_2
should be underscore not -
and update acinclude.m4 for new cpu
New macro/function for AMD mpn_inclsh_n
didn't like it as a macro , so new amd asm functions mpn_inclsh mpn_declsh
mixed up gas and yasm syntax
add Fermat number prime test in benchmark directory
add Mersenne prime test to bench directory
New asm functions for nehalem mpn_add_err1_n mpn_sub_err1_n
add mpn_add_err1 etc to try
correct overlap requirements for add/sub_err1/2_n
update overlap requirement for try for add_err etc
add mpn_add/sublsh_nc to try.exe
New mpn_rootrem
Add new function mpz_nthroot
Add tuning for new mpn_rootrem threshold
C89 changes
revamp tests for mpz_nthroot
New functions mpz_probable_prime_p mpz_practical_prime
Select best asm functions from existing for Atom cpu 64bit
New C functions mpn_mod_1_k etc
add divrem_eucldean to tuniing makefile
change name of mpz_practical_prime to mpz_likely_prime
change name of mpz_next_probable_prime to mpz_next_likely_prime
add macro describing min size needed to avoid allocation for mpz_add
update fat to cope with K10_2 and core2,k8 etc on 32bit
rename K10_2 to K102 as autotools doesnt like - and fat mechanism doesn't like _
whoops uppercase instead of lower case
add macro describing min size needed to avoid allocation for mpz_add,sub,neg,abs
reinstate mpn_divrem, but leave obsoleted
expose mulmod_2exp+-1 to the User , no docs, adjust fermat,mersenne to compile without gmp-impl,longlong
more min size alloc macros
in refmpn.c make sure we use refmpn functions not just mpn functions
refmpn functions for add/sub_err1/2
split out trial division from both probable prime tests and creat new function
update setversion
add tests for trial div
added more tests for roots , corrected existing rootrem error
removed some obsolete error checking
remove obsolete global varible gmp_errno
remove an old gmp version 1 source only compat typedef for MP_INT
remove an old gmp version 1 source only compat typedef for MP_RAT
replace all MP_RAT,MP_INT gmp version with the new 1994 stuff , mpz_ptr mpq_ptr , about timesvn status! , change version numbers to 1.3.0
Obsolete mpf_random2 mpn_random mpn_random2
Add new function without global state mpf_rrandomb
replace all the old mpf_random2 with the new mpf_rrandomb
remove very slight non-uniformness from mpz_urandomm
remove nonsense test from random numbers
Add new functions mpn_urandomb mpn_urandomm mpn_rrandom mpn_randomb
replace most occurances of mpn-random2 with mpn_rrandom
repalce most occarances of mpn_random with mpn_randomb
remove most of the old global random stuff and replace with local
add prototypes
replace most occurances of RANDS with local versions
update mpnbench
tweek bench
tweek bench
fix mpf printing exponent should be long not int bug
tweek benchmpn
add 32bit path for nehalem
remove empty cyrix cpuid
add via nano cpuid and code path
New generic divrem_hensel's functions and tests,tunes
New AMD rsh_divrem_hensel_qr_1 asm fn
New asm functions for AMD divrem_hensel_qr_1 divrem_hensel_r_1
New generic functions for two limb inverse of divrem_hensel_qr_1 and tunes,tests
rename divrem_hensel amd asm to match
new AMD asm function mpn_divrem_hensel_qr_1_2
limit try test of divrem_1_2 to 2limbs
split out mpn_mod_1_? from divrem_euclidean_r and add New asm function for AMD for mod_1_1
autotools and a few tweeks
tweek mod_1_1 amd asm to full speed
New amd asm function mpn_mod_1_2
add mod_1_? to mpnbench
add mingw random/rand workaround
update windows make.bat so that make check returns error code
New AMD asm function mpn_rsh_divrem_hensel_qr_1_2
autotools
add generic mpn_rsh_divrem_hensel_qr_1_?
add tuning for mpn_rsh_divrem_hensel_qr_1_?
add carry limb into the existing rsh_divrem_hensel
Add threshold for euclid/hensel divrem_1 switchover
New AMD asm function mpn_mod_1_3
remove old dead addsub
remove old copy,zero from fft
add divrem_1 to the tuning
back out the divrem_1 tuning change
New nehalem asm functions mod_1_? rsh_divrem_hensel_qr_1_2 divrem_hensel_qr_1_2
New atom asm functions mod_1_? rsh_divrem_hensel_qr_1_1 divrem_hensel_qr_1_1
New core2/penryn asm fns mod_1_? divrem_hensel_qr_1_2 rsh_divrem_hensel_qr_1_2
correct tuning for mod_1_k
correct/tweek divrem euclid/hensel threshold tuning
nehalem 64 params
nehalem 32bit params
set proper soname and number
k102 params
k8 params
core2 params
whoops , restore mpir.info
sparc 32 bit param
sparc64 params's
k8 32bit params
k10 params
k7 params
atom params
add mising fat entry for 32bit nano
add back in old fft tuning values , better than nothing
core2 32bit params
pentium4 32bit params
ia64 params
varro powerpc64/mode64 64bit params
more error checking in windows build script
adjust sparc32/v9=t2,ABI=32 params for mulhigh
fix broken sparc32/v9 add/sub code
put more realistic trial division limit in next_likely_prime
change asm #comment to C comment
change movq to movd for old masm assembler
some more masm? movq/movd mixups
more news
new directory for core2-apple-darwin 32bit with no PIC code in it
cygwin now builds with enable-auto-import this fixes gcc-4 c++ issues
rollback the cygwin --enable-auto-import changes
trac 94 fix extended for penryn and all darwin versions
add nano to x86_64 types
testing junk
testing junk2
testing jnk
convert addlsh from gas to yasm format
new nehalem asm logic mpn fn's
New nehalem asm mpn_com
New nehalem asm mpn_copyi mpn_copyd
remove # comments
New atom asm funtion mpn-copyi
New K8 asm funtions mpn_lshift2,3,4,5,6
New K8 asm functions mpn_and,ior,xor
new K8 asm functions mpn_popcount mpn_hamdist
New core2/penryn asm functions popcount hamdist
k7 params
atom 64bit params
removed all old gcdext_threshold that were set to zero
update cpuid for AMD model 8 , a K102 a six core beast, istanbul
minor doc update
set default flags for gcc to -O2 on itanium as gcc-4.5.0 barfs on -O3
try again for itanium flags
ia64 params for gcd function on cleo.skynet
fix for solaris not recogizing function in *.as files as sed script had [[:space:]] instead of [:space;]
copyi,copyd now alway availible
change all mpn_sqr_n to mpn_sqr
params for powerpc32 and powerpc64 from varro.skynet
update docs to reflect gcdext changes
update makefile.am for build.vc10 directory
update authors in the manual
nicer format for long line of authors
tweek makefile.am for autotools
threshold fix for toom8h unequall sizes
fix bug in gcdext test code when un<vn
update sparc32/v9 params from ABI=32 t2 with manual adjust form mulhigh
add ylwrap form boxen.skynet
remove asmembler assert from mode1o.asm in ia64 , icc doesn't like it
change license in changelog from 2.1 to 3.0
removed old referance to mp.h from readme
update manual for MSVC 2008 and 2010
make info manual
release mpir-2.1.1
removed support for a29k cpu
removed support for clipper cpu
removed support for i960 and i960mx cpu
removed support for m88k m88110 cpu
removed support for ns32k cpu
removed support for pyr cpu
removed support for z8000 and z8000x cpu
autotools
removed configure info for non-existent path for fujitsu cpu
cleanup setversion script
removed non-relavant lisp directory from mpn
remove demos directory from library
keep record of difference between our yasm and yasm-svn , so its easy to upgrade yasm at any time
delete old yasm sv rev 2227
new yasm svn rev 2334
remove dumbmp.c and pre-build gen-* and just include the files they generated
revert rev 3035 changes
whoops missed dumbmp.c
remove support for gmicro , i860 , ibm032 or romp or 032 , uxp or xp fujitsu cpu's
remove unused/tested x86_64 asm
new K8only dir for asm that is faster on K8 but NOT K10
add in westmere cpu properly
remove all refs to the old gcd,gcdext,find_a etc
remove support for POWER1,2 ie RIOS1,2 cpu's
remove pw32 djgpp os2 unicos osf tru64 operating system support
remove pre-build file mpz_fac_ui.h
move setversion and yasm.diff to the devel directory
remove pre-build file fib_table.h
remove pre-build file mp_bases.h
remove cray and vax arch's
update info
put gen-psr in the devel dir
trac ticket 295 , removed file and function mpn_ngcd
autotools
remove prebuild file mp_bases.c
a few more remove prebuild file mp_bases.c
autotools
remove prebuild for fib_table
add in dummy function to temp workaround newer autotools
autotools
remove pre-build file gen-psqr.c dumbmp.c perfsqr.h
autotools
move definition of CNST_LIMB forward in the gmp-impl.h file
remove configure tests for header files invent.h sys/attributes.h sys/iograph.h sys/syssgi.h as they are only needed for IRIX
cleanup
autotools
correct fat entry , got wrong case
add missing asm functions for nano,netburst as fat requires asm (not C) functions for any function in the fat structure
upgrade autotools to 2.2 with the command libtoolize --force --copy
autoreconf
enable parallel-tests
autoreconf
new generic function mpn_not
remove configure tests for functions sigaction sigstack sigalstack as we dont use them at all
remove configure test for readline , was only used in the demos
clearup
remove old macro __umulsidi3 , not used anywhere
change dive_1.* to divexact_1.*
replace diveby3.* with divexact_by3c.*
replace divebyff.* with divexact_byff.*
replace divebyfobm1.* with divexact_byfobm1.* and rename consistently
replace mode1o.* with modexact_1c_odd.*
replace pre_mod_1.* with preinv_mod_1.*
replace pre_divrem_1.* with preinv_divrem_1.*
whoops , missed a pre_divrem_1
change file names in mpn jacbase perfsqr divis to match the function names , split out mpn/dc_div_qr_n from mpn/dc_div_qr and same for bdiv version
update build.vc9 make.bat batch file prebuild step
gmp_randinit_lc_2exp definintion and declaration dont match , fix it
adjust size of bitcnt_t for and WIN64 platform
when WIN64 is set define longlong limb
new x86_64 cpuid that works with both calling conventions
path selection for mingw64
for mingw64 update object format
change ln -s with mingw
mingw build to use x86_64w/yasm_mac copied to yasm_macwin
update configfsf.guess to latest
update configfsf.sub with latest , NOTE did not do the *-unkown | *-pc | *-apple) change
put back in the *-pc | *-apple to configfsf.sub
ifdef out t-locale test for mingw64 and msvc
new dir for atom under mingw64 , files taken from the relavant k8,core2 dirs
bug correction to mpn_divexact
update configure for new atom mingw64 path
new code path for netburst on mingw64
replace multifunction aors_err1,2 with add,sub separate src code
copy latest version numbers from 2.1.2 into setversion , so we dont forget where we are
update test for mpf_get_d to test for -ve numbers for mpir-1.3 , 2.1 and trunk
update tests for mpf_get_d_2exp to test for -ve numbers for mpir 1.3 , 2.1 and trunk
correct -ve number bug in mpf_get_d_2exp for mpir 1.3 , 2.1 and trunk
hack config.guess to distinguish between mingw32 and mingw64
adapt configure to handle the new *-w64-mingw32 machine triple
update configfsf.sub to recognize -w64 as a valid name
exclude mingw64 from the t-locale test
copy x86_64/fat to x86_64w/fat
update fat_entry for windows ABI
add fat paths for mingw64 to configure
make asm to as changes for mingw64 via the makefile rules instead , as fat builds are very difficult to do in configure otherwise
get rid of the old configure rename stuff for mingw64
well the makefile wont work due to the diffences between libtool .o and .lo and yasm .o , so reverting that change
for mingw64 renaming hack the configure
whoops , change name triple mingw64 to mingw*
combine post processing step of m4 to yasm with the second part of the renaming
put a pair of quotes whereever we have a dnl like dnl , so that autoconf does not think the rest of the line is a comment and blank it out
correct script as the word define is a reserved word in autotools
yasm include format requires quotes
add missing empty directorys for cpu types to x86_64w , they are needed to fill in the fat structure properly
path selection in configure must be the same for x86_64 and x86_64w , this is for the fat structure
fat path selection must be the same for x86_64 and x86_64w
make sure configure creates the CPUVEC_SETUP_* as x86_64 not x86_64w
put it in the right place , acinclude.m4
exclude fat_entry.asm from suffix changes
another spot for the fat_entry exception
autotools
exclude fat_entry without whatever path it may be under
exclude preinv macro so that mingw64 shared builds work
x86_64 cpuid always availible , simplify detection code
very preliminary msvc command line build
unique names for mingw64 asm code
hack configure to get around EXCLUDE_PREINV bug for mingw64
define is a reserved word in autotools so put some quotes inside it
enable noexecstack for gcc and x86/x64
gcc on mingw64 doesnt like noexecstack , so exclude this case
copy all param's from x86_64 to x86_64w
update trunk version numbers ready for RC1 release
update setverion script to use mktemp rather than tempfile as eno/fedora doesn't like it
run setversion script on eno for RC1
remove spurious referance to the old mpn/power directory
autoreconf
add yasm_macwin.* files to the distro ie for mingw64
add mangy cours model numbers =9 to the cpuid
exclude locale c++ test on mingw64
refmpn_not should not try to return a value
change default distribution from .tar.gz to tar.bz2
set minimum version of automake from 1.8 to 1.11
autoreconf
add intel model 37 westmere mobile chip , no AES to the cpuid.c
change some %u to %d in printf in the file test/memory.c
fix for race condition in a parallel build for make install
bump version numbers
config.sub update to accept *-w64 as done in yasm
update NEWS
change spacing on AUTHORS
setversion to 2.2.0
add sandybridge to cpu's
new asm code for K10 popcount hamdist
new K8 asm function popcount hamdist
update asm def file
New core2/penryn asm function mpn_popcount
New asm function mpn_hamdist for nehalem
New nehalem asm function popcount
New K8 asm logic functions mpn_nand_n mpn_nior_n
windows conversion script doestn't like sums in code
faster core2/penryn mpn_hamdist by using the K8 version
New K8 asm function mpn_store
add model 44 westmere hexa-core
remove C++ style comments from gmp.h
change name of redc_basecase to redc_1 and export
correct wrong order of params in redc_1 definition
changed order of params for redc_1 for all the C code , just asm and try.c to do
change calling convention on the asm code in x86_64 for redc_1
adjust try.c for redc_1 changes
New core2/penryn mul_1 asm function
change redc params for fat build
make sure we have unique names for FRAME_PROC in redc
update Makefile.am with all versions
New nehalem mul_1 asm function
redc param change in tuning/speed code
new nehalem mpn_mul_2 asm function
devel/regen script to fill in Makefile.am's for mpz mpf mpq printf scanf
new nehalem mpn_mul_1 , this time for real....
eno/fedora doesn't like tempfile use mktemp for benchmpn
split-up popham into popcount and hamdist
remove all traces of aors_err?_n.* , we now use separate add and sub versions
remove all traces of rsh1aors_n.* , we now use separate add and sub versions
remove all traces of aorslsh1_n.* , we now use separate add and sub versions
remove all traces of lorrshift.* , we now use separate add and sub versions
remove all traces of aors_n.* , we now use separate add and sub versions
remove all traces of aorsmul_1.* , we now use separate add and sub versions
remove all traces of logops.* , we now use separate add and sub versions
remove mulfunc from x86/pentium/logic*.asm
remove MULFUNC_PROLOGUE from most asm files , only carry varients left
remove some MPBSD stuff left lying around
remove define TARGET_REGISTER_STARVED as never used
remove some old comments
add AMD K102 model numbers 5,6,10
add AMD family 18=12h , a K102 with hardware divider
add AMD family 17=11h , fusion of K8 and GPU
add AMD family 20=14h , fusion of bobcat and GPU
add new Intel cpu models 31 46 47 and change model 30 from westmere to nehalem
New AMD addmul_1 , was 450bytes now 407bytes
update longlong for mips asm restrictions on later gcc's
update AMD mul_1 and submul_1 , just like the new addmul_1
remove spurious error in non-fat build on mingw64 systems
add in missing break statment in cpuid.c
New nehalem mod_1_1
add memory check to t-mulmid
add mem tests to t-rand.c
New mod_1_1 for core2 and a slightly different one for penryn
correct wrong name in amd submul_1
New AMD mpn_mod_1_2
New mpn_mod_1_2 for Core2/Penryn
New AMD mpn_not asm function
corrected K8 not.asm
retune params for nehalem
retune core2 params
retune K8 params
retune penryn params
retune atom params
retune k10 params
retune netburst params
setversion to 2.3.0 RC1
update version numbers to 2.3.0-RC1
remove build.vc9 from makefile
autotools
fix noexecstack for 32bit ABI
generate mpir-2.3.0-rc2
restrict noexecstack on x86 to *-*-linux-gnu
update NEWS
mpir-2.3 branch
New asm directory for bobcat
autotools
select from existing dir's fastest functions for bobcat linux and windows
asm basecases for bobcat from k8 for linux and windows
bobcat params
delete old yasm
new yasm installed rev 2368
by default make dist does .tar.bz2 AND .tar.lzma
autoreconf -f with the latest autotools automake 1.11.1 autoconf 2.68 libtool 2.4
libtoolize -c
ran devel/regen to add new mpz functions to Makefile.am and run autoreconf
add existence and sizeof tests for intmax and uintmax
change SIZEOF_UINTMAX to SIZEOF_UINTMAX_T
add tests into autotools for new mpz fns
remove some old libmp stuff
add tests in mpz mpf mpq misc mpn dir to regen script
add all tests except c++ tests toe the regen script
get yasm to assemble intel format asm code
prelimnary MSVC command line build
msvc cl build tweek
MSVC cl make clean
MSVC cl build some make checks
MSVC cl build tweeks
add C++ to MSVC cl build
MSVC CL build make speed
MSVC CL build make try
MSVC CL build make speed
MSVC CL build tweeks
MSVC CL build make tune
tweek
tweek
add win dir to distro
noexecstack for gcc+linux+x86 or x64 only
add HAVE_STDINT_H guards for the include <stdint.h>
correct speed tuning error for old data file inclusion
MSVC CL Build configure.bat
MSVC cl build update
msvc cl build update
msvc cl build update
msvc cl build update
msvc cl build tweeks
msvc cl build tweeks
msvc cl build tweeks
msvc cl build tweeks
msvc cl build tweeks
msvc cl build tweeks
msvc cl build tweeks
disable yasm from fat builds , temporary
clean up fat stuff
remove cpuid_availible function from fat detection
remove old test for gcc-2.9 on alpha as it stops certain cflags ie -Werror
add test to reject gcc-4.3.2 on 64bit system
add test to reject gcc-4.3.2 on 64bit system , left some rubbish in it , whoops
add test to reject gcc-4.3.2 on 64bit system , make sure not 32bit
get it right
identify k103 as k103 , not as just another k102 , but no separate path for it yet
nano uses the k8 path for assembler
netburst and netburstlahf cpus now use the same directory
delete empty dirs
update cpuid.c to be a function rather than a code fragment
and for windows
adapt __gmpn_cpu so it can be faked
split fake cpu into two , build and fat
delete , for now k8's udiv.asm and umul.asm
nonexecstack for linux only
copy some k8 files to x64 dir so that fat build on unknown cpu's work
script to change fake cpu
update version to MPIR-2.3.1
NEWS
remove arc(obsolete in gcc 4.6) support from longlong
add gcc-4.6 compiler flags for atom,bobcat,corei7,avx
make sure fat dir has every file in fat structure
and for x64w
and for x64w
and for x64w
change fat order from x86,fat,.... to fat,x86,....
test/mpz/reuse.c msvc tweek
add in 32bit paths for msvc cl build
add bulldozer family 21 to cpuid , map it to k102 for the mo
split up longlong into arches
put license info in all the new longlong's
split out windows longlong parts
remove define UDIV_TIME as never used
remove define UMUL_TIME SMUL_TIME as never used
movq to movd in asm
define __GMP_CC as __MPIR_CC and same for flags
add in paths and the best asm we've got for sandybridge
add in bulldozer as a separate cpu name that maps to the k102 dir
add new sandybridge path to the fat system
add new bulldozer paths to the fat system
split longlong into arches , windows and possibly a bit of ifdef wrapping to do
final bits of splitting longlong.h
change makefile to reflect that longlong.h is now a built file not a source file
add bulldozer and sandybridge paths to the x86_64w fat path
change fat order in fat path in configure as well
update for name changes to longlong_inc.h
remove k6/k7 mod_1 FAT optimization
add missing fft params for x86 fat build
make x86 fat preinv_* general , no more bloody special cases.....
reverse x86 fat order for x86 only , it's not worth the effort to fix all the special cases in the division code
reverse previous change
put back modexact_1 special case x86 FAT code , have to wait until we change all the configure stuff as well, Ugg
include some extra ifdef around inline asm
remove all refs to NO_ASM
move asm code in gmp-impl into the arch specific dirs
remove s390 pa32 pa64 explicit code
remove old code
make install gmpxx.h bug
bump SOME version numbers so I can test out the new testing framework
bump gmp compat version to 5.0.2
win cl build longlong tweek
remove old asm x86w/p6/sqr_basecase.old.asm
add -D DLL to the msvc cl build for yasm
msvc cl build flags correction
more msvc cl build tweeks
remove win32 spurious warning
get spurious warning right
comment out un-needed header files
fixed useless warning with yasm 32msvc signed constant out of range
msvc cl build tweeks
msvc cl build tweeks
make sure tests mpz/t-set_sx.c etc have entry point
msvc cl build tweeks
msvc cl build to stop on error
change gmp.h and gmpxx.h from a make_install_hook to proper built sources
update docs for intmax types
tweek docs
sandybridge gmp-mparam.h
bump version numbers
bobcat gmp-mparams
update NEWS
unique symbols requires for miongw64 fat build
and rename symbol in sandybridge/mul_1
delete inc/declsh_n entry point from sandybridge add/submul_1 as in mingw64 fat build the symbols clash
doc update
correct c++ sonumber
correct c++ sonumber
crlf to lf for t-get_sx.c etc files
CRLF to LF on all c files except in build.vc10 dir
CRLF to LF script
CRLF to LF script
CRLF to LF for *.h
CRLF for all *.h and *.c files not in build.vc10 or yasm
update NEWS
mpir-2.4 branch
x86_64 fat build uses rip relative addressing
x86_64w fat build also uses rip relative addressing
switch all x86_64 asm code from gas/yasm to just yasm
karaadd karasub and toom22
kara tweek
kara tweeks
modify kara_sqr for use karasub
removr 2.3 branch
add new karaadd/sub to speed
karasub swap rcx,rdx save one inst
karasub fixed rcx simplifications
karasub change mirrored to karaadd
tweek karaadd/sub , add k10 versions
karaadd/sub for bobcat , same as K10
karaadd/sub for sandybridge , same as K10
karaadd/sub for atom , same as K8
update gmp-impl for karaadd
copy k8 and karaadd/sub to the other cpu arches linux and windows
remove thumb,m68k,sh cpu's
remove old toom22 code
karasub/add for nehalem
mpn_half mpn_double asm for K8
add half/double test code and correct return values
Use new half and double code
add karaadd/sub to benchmpn
New nehalem asm functions for addadd addsub subadd
New addadd addsub subadd for core2/penryn , note same as nehalem code
New mpn_sumdiff_n for nehalem
New asm mpn_sumdiff_n for core2
New asm fn mpn_sumdiff_n for penryn
write new toom eval for +-1 using addadd and sumdiff
replace old add_n_sub_n by sumdiff
cleanup
removed some alpha specific stuff
remove some HAVE_HOST_CPU_* and some x86 macro's in then x64 dirs
remove use of FORCE_DOUBLE ie 80bit floats which are slower now anyway
remove pointless differences between old cpus ie k6's
remove rest of special alpha code , very minor slowdowns
remove some old powerpc32 minor optimiztions
remove referances to some non existent paths
remove more have_host_cpu_*
disable yasm for *.asm and use gas as some functions are slightly faster
correct rounding mode tests and generalize, remove some more have_host's
removed all have_host_x86 etc except for build,vc10 dir
remove all references to GMP_IEEE_FLOATS , thats the only type we support anyway
remove obsolete x86_fldcw and x86_fstcw that change rounding modes
corrected wrong ret code
removed all use HAVE_HOST , except for timing on the powerpc
add new model 45 is sandybridge?
syntax error
remove ancient stuff from docs
bump version and so numbers ready for release
branch from trunk
configure for gcd
some tuning for the new gcd
tuning fo hgcd
tuning for sandybridge
nehalem tuning
update copyright for tune/common.c for fn speed_mpn_hgcd
sandybridge tuning
atom tuning
bobcat tuning
nehalem tuning
k8 tuning
netburst tuning
core2 tuning
pentium4 tuning
ia64 tuning
sparc64 tuning
sparc32/v9 tuning
k8/k10/k102 tuning and copy to k8/k10 tuning
core2/penryn tuning
powerpc64 tuneup
delete old yasm
new yasm installed version 1.2
update yasm update instructions
crlf removal
run in devel regen and setverion
manual fix sparc32/v9/mulhigh param to be >=3
manual fix powerpc64/gmp-mparam mulhigh param to be>=3
cl make check to exclude t-fat test
updates docs
NEWS
setversion
NEWS
generic addadd addsub and subadd
change backslash to forward slash in fft_tuning.h
update makefiles and configuree for fft dir
some more fft makefile stuff
some more
some more
some ullong defines in tests/t-constat
correct k8 karasub
correct generic karasub for carry/borrow propagation
gmp-5.0.3 to gmp-5.0.4 bug fixed
karasub for all x64
karasub for windows
correct pop order for karasub , add redc_2 , add generic addadd addsub subadd sumdiff
bump version numbers
news
remove mpir-2.4 branch
sumdiff try test unconditionally
correct assert in sumdiff
correct header file in redc_2
cpuid update
add new undocumented macros in gmp-impl for mpfr-3.1.0
comments on new macros
asm redc_1 for sandybridge and bobcat for x86_64 and x86_64w
add three more atom model numbers and ivybridge
do a proper test for bad gcc-4.3.2
make the error message clearer
autoreconf
add in fallbacks for gcc-4.3.2
bump version numbers
new cpuid's perhaps
add GLOBAL_FUNC to all x86_64w asm files
change global func to prologue
make sure x86_64w asm files are in dos format line feeds
martin (1):
Added a blank line to the README file to test SVN
martinjw (6):
Corrected my name in AUTHORS
Fixed several problems with Moller gcd patches
Fixed problem with 'make speed' using Moller gcd patches
Set svn:executable on install-sh
Testing access policy
Ignore
mvngu (17):
core2: tuning values for GCD_THRESHOLD and GCDEXT_THRESHOLD, averaged out for Linux and OS X
penryn: sage.math Linux: tuning values for GCD_THRESHOLD and GCDEXT_THRESHOLD
pentium4 32-bit: cicero.skynet Linux: tuning values for GCD_THRESHOLD and GCDEXT_THRESHOLD
k8: flavius.skynet Linux: tuning values for GCD_THRESHOLD and GCDEXT_THRESHOLD
ultrasparc3: mark.skynet and mark2.skynet Solaris: average tuning values for GCD_THRESHOLD and GCDEXT_THRESHOLD
netburst: sextus.skynet Linux: tuning values for GCD_THRESHOLD and GCDEXT_THRESHOLD
nehalem: taurus.skynet Linux: tuning values for GCD_THRESHOLD and GCDEXT_THRESHOLD
changes in Makefiles and configure scripts after running setversion for mpir-2.1.0-rc1
wrapping up MPIR 2.1.0-rc2
wrap up MPIR 2.1.0.final
update ChangeLog to reflect the release of MPIR 2.1.0
changes between MPIR 2.0.0 and MPIR 2.1.0
add Minh Van Nguyen to list of contributors
wrapping up MPIR 2.1.0, this time for real
acknowlege Minh Van Nguyen in MPIR documentation
set version for MPIR 2.1.1 bug fix release
set version for MPIR 2.1.1 bug fix release
wbhart (569):
Basic GMP files with a new core2 directory and amd_64 directory with Martin's and Gaudry's patches.
Edited gmp.h to use static inline in c99 mode and changed permissions of
Added some notes about gmp.h and a devel directory to the doc directory.
SVN seems to screw this file up. Not sure whether this will fix it.
Try to fix ltmain.sh from linux.
Added ARM assembly support back in.
commit a369ff49df609c1aee7ce9c96167ba38d27b3cc2
Added all the assembly code back for all supported architectures.
Description of all the files in the main directory of the GMP source tree.
Roughly speaking mpir should now build on an AMD64. At the present moment the config.guess doesn't distinguish a Core 2 from an AMD64 and so the same code is probably built on both.
Ran autoconf, automake and aclocal to update all the autohell files.
Fixed some problems preventing it from building.
Small change to configure to help assembly of files not transitioned.
The gmp-mparam.h files seemed to be different between my two sets of files. These ones make the segfault go away in the flint test.
First attempt at making make check work.
Fixed a tab that had got converted to spaces.
Run automake to propagate changes.
Added execute privileges to m4-ccas.
Changed names so make check can work.
Will allow tune to build.
Run autoconf to propagate changes.
Fixed tune makefile to reflect changes to name of x86_64.asm.
Fixed the code to not use addressing via the GOT when PIC is not needed.
Made non-PIC code more efficient with lea.
Raised the UNROLL_COUNT so that tuneup will not segfault when it uses
Added some comments about the high value of UNROLL_COUNT used.
Set execute permissions on these scripts and applied Gonzalo's patch for
First attempt to do everything right by the core2.
Run autoconf to propagate changes to configure.in.
Modified so that core2 is passed through and recognised.
Changed execute permissions.
Corrected directory order for x86_64.
Try to get make check to work on core2.
Remove checking of calling conventions, etc, for now.
Rearrange files so build can occur on core2.
Attempt to fix assembler file names.
Run automake and autoconf to propagate changes.
Missed one file in the rename.
Fix for libtool to make the required files.
Explicitly give the output library name so libtool doesn't get confused.
Ensure sqr_basecase can have .as extension so make tune can work.
Rename this file so it can assemble with yasm.
Run automake and autoconf.
Remove this file.
Delete files not part of Jason Martin's patches.
Make MPIR_AS always defined.
Run autoconf.
Fix ltmain.sh for Darwin (hopefully).
New version of aclocal.m4.
Attempt to fix Darwin build.
Rebuild aclocal.m4
Attempt to get the object formats correct.
Rerun aclocal, autoconf, automake.
Set the object file correctly.
Unfortunately Yasm doesn't accept ..gotpcrel in Macho64 object format, which is needed on Darwin on core2.
Unfortunately Yasm doesn't seem to export global symbols properly on Darwin.
GPL Version 2 files.
Change these files to compile on Darwin.
Change x86_64.s to x86_64.asm.
Update makefile.
Run automake.
Get rid of transitioned.
Replace calling conventions.
Move calling conventions for core2 into right directory.
Run autoconf and automake.
Update makefile for core2call.
Add core2call in this directory too.
This is in the wrong place.
Slight speedup by getting alignment right.
Fixed bug in gcc calling conventions.
This file should be deleted.
Sped up add and sub when the loop unrolling code is used.
Fixed the speed issues with a static library vs Pierrick Gaudry's
These are the old versions of addmul and submul written by Brian
Removed some lines which were saving and restoring rsi unnecessarily.
Set execute permissions on config.sub
Added support for the pentiumD. Basically it is now recognised as some kind
Added model numbers 16 and 17 (Celeron and Core2) to config.guess.
Fixed the model numbers 22 and 23 which were previously entered as 16
Added missing semicolon, noted by iciti.
Attempt to check for gnu linker on Solaris.
Backed out changes to aclocal.m4 and fixed gmp.h to work correctly with c99 inline semantics when mpir is being compiled with gcc 4.3.0 or above.
Experimental fix for global symbol prefix for yasm assembly files.
Attempt to fix the global symbol prefix on core2.
Added macro for global symbols.
Temporarily removing yasm to update to newer version.
Added new version of yasm.
Added global extern.
Fixed typo in yasm_mac.inc
Removing yasm again due to something wrong with update.
Added new yasm.
Put macros instances in all yasm assembly files for global symbol
Fixed type on add_n.as
Fixed another typo in add_n.as.
Fixed bug in global symbol macro in acinclude.m4
This file appears to be generated by build process and is not needed.
Converted x86_64.asm to intel format.
Added global symbol to symbol.
Turned execute status on.
Experiment with execute permission.
Change execute permissions back to how they were.
Corrected documentation error for gmp_randinit_mt.
Attempt to get yasm to build unconditionally and automatically.
Try to fix order of directory recursion in make.
Run automake.
Cleaned the mode1o.as file up slightly.
Added make clean to yasmbuild's makefile.
Added blank check and install targets to the yasmbuild Makefile so these
Added check-recursive target to yasmbuild Makefile.
Attempt to fix c99 inline sematics again.
Fixed typos in gmp-h.in
Fixed yasm so that it only builds on x86, not by default.
Set native line endings for all .c, .h, as, .asm, .s, .in, .m4, .cc, am
Changed directory build order so yasm will build again.
Changed configure.in to set proper tuning flags for core2.
Added support for 64 bit PentiumD and Pentium4 which are canonicalised as i786.
Made family 6 model 22 celeron identify as a core 2.
A potential fix for building with cc on Core2 x86_64. This fix will not work until the Core 2 assembly patches are converted to intel format.
Yet another attempt to fix c99 inline stuff.
Reversed changes to gmp-h.in which did not work.
Added lines to gmp-h.in to try and fix problem on Apple GCC.
On apple gcc v 5465 and above c99 is enforced by -std=gnu99 the same as for gcc 4.3 and above. The problem is, one cannot use either extern inline or extern __inline__ __attribute__ ((__gnu_inline__)) or static inline for the inlined functions in gmp.h. The first can't be used since it isn't c99, the second can't be used since the attribute is not supported by apple's gcc and the third cannot be used since this makes the library functions also static inline which causes duplicates.
Added Jason's new intel format core2 code.
Fixed some documentation bugs reported by Chris Saunders to the gmp list.
Remove the -fast compiler option for Sun's Forte C compiler under solaris. It apparently selects invalid optimisations for GMP.
Ran autoconf, aclocal and automake.
Rebuilt gmp.info files.
Added core2, x86_64 as supported build targets and core2 as a supported CPU.
Edited version numbers and copyright info for mpir.
Did a fairly careful search and replace of GMP, GNU MP and swox for the relevant MPIR related alternatives.
A few minor corrections to the documenation.
Changed version numbers back to 4.2.1 in gmp-h.in.
Added PIC code for add_n and sub_n and did a basic cleanup of the files.
Removed addmul_1 and submul_1 from amd64 directory and added them to the generic x86_64 directory.
Moved intel format versions of add_n.as and sub_n.as to the /mpn/x86_64/amd64 directory to make way for intel format versions of original add_n.as and sub_n.as files from GMP 4.2.1.
These files were moved in error, so I'm moving them back. See the comment in trac #59.
Converted vanilla GMP 4.2.1 add_n, sub_n, addmul_1 and submul_1 code to intel syntax.
Reverted core2 patches since they need some more work before merging.
Put execute flag on install-sh.
Distinguish pentium4, prescott and nocona.
Dos to unix line endings.
Corrected error in config.guess.
Tuning for pentium4, prescott and nocona.
Added blank distclean so make distclean works.
Fixed distclean to also clean the yasm tree.
Fixed some of the documentation issues noted by David Harvey.
Patch to deal properly with 1 nail bits in mpz_set_si test.
Added a comment about the failure this file induces in make check.
Remove the .brg files which are not needed.
Add MPIR_VERSION macros.
Ran autoconf and makeinfo.
The .asm file is not needed now that it has been converted to yasm format.
No one wanted to retain this, so it bites the dust.
Removed macos from the list of directories in the makefile.
Updated documentation to reflect removal of macos port.
Change eol-style to native.
Ran automake and makeinfo.
Rename amd64call.asm and core2call.as and amd64check.c to x86_64call.as and x86_64check.c (merging the first two into a single file) and updated configure.in accordingly.
changed line endings to native style.
Ran automake.
Removed these .brg files which are no longer needed.
Removed the .asm files which are no longer needed.
Cleaned up /mpn/x86_64/*.as
Cleaned up /mpn/x86_64/amd64/*.as files.
Added comments to all the files /mpn/x86_64/*.as
Removed unused macros from yasm_mac.inc.
Corrected incorrect usage of G_EXTERN macro.
Whoops, these macros are needed! Putting them back.
Fixes duplicate label in sqr_basecase.as
Second attempt to fix label problem.
Replaced instances of swox addresses with link to mpir devel list.
Update eol-style for these files.
another eol-style change
Added some license information to files that didn't have it.
Updated license information for many files.
Ran autoconf and automake.
Changed eol-style for strip_fPIC.sh
Moved some of the developer specific files into the devel directory, updating the Makefile.am accordingly.
Ran automake and autoheader.
Ran makeinfo.
Updated gmp_mparam.h for x86_64/amd64
Finished editing gmp.texi and ran makeinfo.
Changed configure.in so odd nails values are not permitted. Ran
Added gmp-mparam.h parameters for Core2.
Apply all GMP issued patches for issues with GMP 4.2.1.
Apply GMP issued patch for mpf_set_str.
Apply GMP issued patch to the patch for mpf_set_str!!
Second part of patch to patch for mpf_set_str (fix to the base issue).
Merged SiCortex patches for MIPS.
Ran tuneup to generate gmp-mparam file for mips64.
Reran autoconf.
Regenerated calc and the gmp info files.
test
Test - blank line
Test - remove blank line
Added support for CPUID Family 6 Model 1D (Xeon 7460 -> Core 2).
Fix for c++ build issue (--tag missing).
Added support for Phenom in config.guess.
Added prototypes for mpn_bgcd, mpn_sgcd, mpn_rgcd and mpn_ngcd to
Converted line endings to Unix format.
Fixed numerous issues which prevented compilation with --enable-cxx on Sun CC.
Corrected an error I introduced.
Fixed a linkage bug which prevented Sun CC from compiling make check.
Fixed a minor typo.
Removed attributes and facets only for Sun CC instead of all non-GNU compilers.
Altered copyright notices to reflect edits.
Disabled nails altogether.
Updated docs to note that nails are not currently supported.
Fixed a std::FILE issue in gmp.h and a missing <cstdlib> in t-locale.cc
Added <csdio> when compiling with a C++ compiler instead of checking for _GMP_H_HAVE_FILE to fix the std::FILE issue.
Updated NEWS< ChangeLog and AUTHORS files.
Added prototype for mpn_basic_gcd.
Made changes to allow a fat build on x86_64.
These files seem to need regenerating. They need yacc to do so, but it is not available on all systems.
Changed the order of x86_64 and x86_64/fat in configure.in so it will
Made /mpn/x86_64/fat/gmp-mparam.h 64 bit instead of 32 bit.
Change extra_functions to extra_functions_64 for 64 bit fat build.
Rewrote fat.c to work with x86_64 processors. Made fat_entry.asm
Fixed cpuid.
Made fat binaries work for x86_64.
Ran autoconf.
Reinstated all the tests in t-fat.c.
Update NEWS.
Slight simplification of fat build stuff in configure and possible fix
Fixed a regular expression.
Ran autoconf.
Changed some GMP's to MPIR and a bug reporting address, found by Paul Zimmermann.
Updated README and AUTHORS in the Windows build. Removed some ansi2knr stuff which causes recent autotools to choke.
Ran aclocal, automake, autoconf.
Converted Jason Moxham's K10 code to yasm format.
Coverted left and right shift assembly functions of Jason Moxham to yasm
More of Jason Moxham's assembly code converted to yasm format.
More conversions of Jason Moxham's assembly code to yasm format.
Conversion of redc_basecase of Jason Moxham to yasm format.
More of Jason Moxham's code converted to yasm format.
More of Jason Moxham's assembly to yasm format.
More of Jason Moxham's code converted to yasm format.
Last of Jason Moxham's K8 assembly code converted to yasm format.
Jason Moxham's Core 2 assembly code to yasm format.
Changed alignb #,nop back to align # because it appears to make no
Added reference mpn implementations for missing try functions.
Added new reference test for divexact_byff
Fix for c99 issue on Apple machines using FSF GCC.
Fix a bug I introduced.
Attempted fix for the upper case to lower case issue on Sun causing the fat build to fail on fulvia.
Fixed nocona build.
Try fixing nocona issue without breaking everything else!!
Gonzalo Tornaria's patch to make all 64 bit intel processors either core
Updated version to mpir-1.0.0.
Updated documentation with details about --gmp-compat and new host cpus.
Changed --gmp-compat to --enable-gmplink throughout.
Update AUTHORS, ChangeLog and NEWS files for release.
Turned execute and executable on for config*
Changed eol-style:native on two files.
Removed superfluous instructions from conversion to yasm format in
Fixed yasm -pc configure bug.
Added toom4 multiplication.
Temporarily ran autoconf and automake until someone runs it with an old
Removed cheat in mpz/mul.c.
Went back to my optimised mul.c as it was faster on Windows.
Tried to fix a tuning issue with toom 4.
Moved toom internal cutoff into gmp-impl.h.
Used mpn_divexact_byBm1of for division by 15.
Attempt to add toom cook 7.
Added missed reference to toom7 function.
Set library mode 1 in toom 7 file.
Added commented out Toom 7 sequence (hand generated).
Fixed failing t-locale test on Apple.
Corrected typo in docs.
Ran makeinfo.
Changed athlon to k8 in line with cpuid.c.
Added pathcc support for mips.
Ran autoconf.
Updated pathhcc_n32_cflags.
Ran autoconf.
This seems to fix pathscale support on MIPS64.
Added bogus dependency to work around an Apple ld bug.
Added __GMP_CC and __GMP_CFLAGS, __MPIR_CC and __MPIR_CFLAGS, and ran
Disabled yasm make check.
Fixed mpbsd and got rid of mocked up mpz version of tc4_divexact_ui and
Added support for testing three source operand functions.
Updated build targets for recent rearrangement of config.guess.
Ran makeinfo.
Added a file describing new mpn functions.
Update binary numbers.
Updated ChangeLog, AUTHORS and NEWS.
Updated documentation for install-gmpcompat.
Ran makeinfo on gmp.texi.
Removed TODO from Makefile.in.
Disable yasm tests.
Removed some broken asserts from toom code.
Credit Bodrato in the way he requested.
Added -apple to support very recent apple machines.
Added Zimmermann et al's FFT (after making a bug fix).
Fixed bug in mul_fft.c
Tried to clean up a little.
New toom4 and toom7 code.
Fixed a carry issue with tc*_addmul which created a requirement for extra memory in toom code.
Fixed some bugs in best_k code used by FFT.
Added new toom3 code.
Fixed toom4 and toom7 issues and added k8 tuning code.
Whoops I screwed up toom4 and toom7, putting them back now.
Added my copyright info.
Another slight speedup.
Added toom3_mul_n with better memory usage.
Reverted to earlier version of mul.c because the later version slows
Factored out mpn_toom3_sqr_n and mpn_toom3_mul_n and removed duplication
Added missing toom3 file.
Added missing #include "mpir.h".
Speed toom4 up by passing some arguments to the interpolate code in the
Went back to old mul.c (again) :-(
Added toom42 and code to handle unbalanced multiplication.
I put the wrong version of this in. Here it is back again.
Turned off unbalanced multiplications as they slow things down.
Fix speed regression in mul.c, switch unbalanced toom back on. Add
Fixed some buglets in toom4.
Fixed bugs in Toom3 code.
Added toom32 for unbalanced multiplications.
Went back to mpz/mul.c from mpir-1.1.
Added half gcd implementation based on the original ngcd implementation
Added toom4 squaring code - no tuning code yet!
Slight memory usage improvement for toom4 squaring code.
Added toom7 squaring and sped up multiplication slightly by better use
Added copyright notice.
Fixed some bugs related to tuning gcdext and added tuning for toom4 and
Reverted a change which slows things down on k8.
Added tuning values for core2 and fixed a minor bug in mul_n.c.
More tuning values, not that they make any difference.
Removed some compiler warnings for MSVC from mul_fft.c.
Fixed a bug in FFT squaring code.
Tuning parameters for penryn.
Added K10 tuning values.
K8 tuning values.
Tuning parameters for Core2.
Added workaround for gcc 4.1.2 on ia64 in toom4 and toom4 squaring code.
Added some generic x86_64 tuning values (just copied from K8).
More generic x86_64 tuning values and values for fat binaries (taken from k8 values).
Updated all version number, author, changelog info etc, for new release,
Ran aclocal, autoheader, autoconf, automake, configure.
Attempt to fix missing build.vc9 directory for make dist.
Ran automake.
Slightly relaxed the conditions used in make tune to prevent tuning malfunctions and added make tune values for netburst.
Added arm tuning parameters.
Added powerpc7455 tuning values.
Added tuning values for hppa2.0 (ABI=1.0).
Deleted all pa64 assembly code as it all appears to be broken.
Added Windos Core 2 tuning values (it's slower fith FFT-tuning values,
Added reference to the FFT paper of Gaudy, Kruppa and Zimmermann.
Added tuning values for netburst.
Moved new mul_basecase into netburst directory.
Test. Please ignore.
Added a line describing where FILES.txt can be found.
Added line describing where Configure.txt can be found in the source tree.
Merged test2.
Cleaned up some of the comments in gcdext to do with the "cofactors" u0 and u1.
Switched on lgcd for small gcd computations.
Added mpn_gcdinv_1 and turned on MPN_ZERO and mpn_store assembly code in
Improved the mpn_gcdinv_1 function for speed.
Implemented Lehmer nxgcd and turned it on.
Made some minor improvements to gcdext.c.
Slight speedup by allowing tdiv_qr to alias and reduction of memory
Added strassen multiplication and increased the gcd test sizes so that
Saved an unnecessary copy in lehmer gcd and fiddled with some parameters
First hack at toom4_mul (unbalanced toom4).
Ran autoconf and automake.
Fixed all the bugs I just implemented.
Fixed some missing definitions.
Added toom53 and fiddled with the toom4 cutoff on penryn.
Fiddled with the tuning crossover code in mul.c for the unbalanced
Merged in David Harvey's mulmid code - not actually used by anything
Attempt to add test code for mulmid and mulmid_n.
Added more mulmid test code including mulmid_basecase.
Added ordinary test code for mulmid (including mulmid_basecase, mulmid
Added in David Harvey's Hensel division code. It isn't being used by anything at this stage and has no test code.
Fixed a memory leak in lgcd.c.
Removed a broken assert.
Fixed some broken asserts.
Removed unused gcd functions, removed a static array of length 1000 and
Moved some gcd functions to the correct list in configure.in and added
Removed unused gcd files.
Moved some declarations to the beginning of a block.
Moved some declarations to the beginning of a block.
First very basic attempt at sb_divappr_q. Probably doesn't even compile yet.
Added sb_divappr_q to Makefile and configure and made it compile.
Added config stuff for invert.c and t-sb_divappr_q
Got invett and sb_divappr_q working (modulo some linker issue).
Fixed a missing mpir.h.
Added configure info for divide conquer division.
Added files for divide conquer division.
Fixed a segfault in t-sb_divappr_q.
Oops, this time the segfault in t-sb_divappr_q should be fixed.
Added autoconf stuff for mpn_tdiv_q.
Added fast mpn_tdiv_q and test code and used it in mpz_tdiv_q.
Fixed some bugs in the test code.
Fixed unusual bug in sb_divappr_q (I think).
Added prototypes for mpn_tdiv_q.
Extended the mpz/t-tdiv test and fixed a bug in mpn_tdiv_q.
Fixed a major bug in mpn_tdiv_q and extended some tests to run longer.
Added some rough justifications for doing what we do in sb_divappr_q.
Clarified some comments.
Fixed some bugs in mpn_tdiv_q and added a try test for mpn_tdiv_q.
Fixed more bugs in sb_divappr_q.
Fixed a carry bug in mpn_tdiv_q and added the failing cases to tests/mpz/t-tdiv.c.
Added some incomplete info to AUTHORS.
Added NEWS for release and committed changes made by make dist.
Updated ChangeLog and AUTHORS files.
Added copyright info to GCD and XGCD code.
Updated copyright info for mpz_tdiv_q.
More updates to license info for mpz functions.
Oops, wrong author added to license info.
Added note about Solaris 10 first release on SPARC.
Correct some spelling errors and other issues in docs.
Added some info to docs and corrected some out-of-date material.
Switch over to exact quotient when divisor gets large in mpn_tdiv_q.
Added some clarifying comments to the mulmid code.
attempt to push windows tuning update
This file had an LGPL v3 patch of TG's in it. I'm reverting it so
second try
correct minor windows build issues
Correct minor windows build issues
update mpf/set_str to correct bugs without using patches
remove mpf set_str.c bug
Further correction of set_str.c
update copyright on set_str.c
sadly, yet another bug fix for set_str.c
Changed eol-style to native.
correct minor windows build issues
This file had an LGPL v3 patch of TG's in it. I'm reverting it so
update mpf/set_str to correct bugs without using patches
remove mpf set_str.c bug
Further correction of set_str.c
sadly, yet another bug fix for set_str.c
Try again.
Try yet again.
Rewrte dc_divappr giving proof sketch and making it slightly different.
Removed a small bit of unnecessary code.
Added is_likely_prime_BPSW code and test.
Fixed the test code for mpz_likely_prime_p.
Restored Makefile.am.
Fixed piles of UL's which should be CNST_LIMB.
Corrected one of the constants I maladjusted.
Added a test for the sqrt code and made them work all the way up to
Added test for exact square roots,
Added test code for mpn_invert (currently fails for unknown reasons).
Added Paul's copyright.
Added my copyright and a small change to a print statement (not the only
Removed a mid block declaration.
Added my copyright to the file mpn/generic/invert.c.
Updated the documentation, including references, contributors, removed an empty
Made likely_prime do 10 R tests and checked for not too many pseudoprimes in
Changed AUTHORS, NEWS and ChangeLog files for new release.
Made t-invert a bit more comprehensive.
Added tuning params for K102. Added new fft code to main directory.
* Merged the inner FFT and IFFT's in the truncate MFA code with the pointwise
Turned negacyclic convolution back on after making it more efficient.
Made a change to hopefully support unaligned memory allocation as requested by
Attempt to run autoheader, autoconf, automake.
Added new family 6, model 30 CPU. Patch due to Gonzalo Tornaria.
Oops. Added Gonzalo's copyright.
Result of running aclocal, autoconf, automake.
Ran autoheader.
Corrected bug in Makefile.am
Fix limbp version numbers.
Gee, it's 2010 already. When did that happen!!
Regenerated Makefile.in for tests/mpn.
Added t-invert to tests/mpn.
Updated NEWS and Changelog for release.
Added missing reference.
Ran makeinfo.
Corrected reference.
Minor correction.
Make yasm recognise MPIR arguments.
Ran make dist.
Fixed a bug in mpn_gcdext reported by Dan Grayson.
Added copyright info to NEWS file.
Fixed an overrun in divexact.c.
Added copyright to file.
Added mpn_zero C file.
Ran autoconf, etc.
Fixed a missing mpir.h.
Fixed the xgcd normalisation issue and redid the tuning code for gcd and
Fixed the gcd tuning code and added core2/penryn tuning.
Added tuning for k102.
Added better default values for gcd and xgcd crossovers.
Fixed a missing case in inv_divappr_q.c
Fixed the same bug in dc_divappr_q.c
Fixed an issue with divexact.c.
Added missing mpn_mul_fft function (now mpn_mulmod_Bexpp1_fft) and used it.
Update configure for new fft functions, run autoconf, automake.
Added GLOBAL_FUNC ripping back as Linux doesn't build without it.
Removed mul_fft.c from tuneup.
Attempt to add fft files to tuneup.
Added typo to match incorrect filename in fft directory.
Fixed fft stuff in tuneup.
Fixed a bug in speed/common for dc_bdiv_qr_n.
Ran autoconf, automake, autoheader.
Whoops I reverted the fix by mistake. This puts it back.
Fixed a documentation error wrt -ve exponents in mpz_powm_ui.
Removed old FFT tuning params from gmp-mparam.h files.
Added 32/64 bit tuning values in fft_tuning.h.
Ran autoconf, autoheader, aclocal, automake.
Ran autoreconf -i.
Update line endings to local style.
Attempt to run MinGW32 version of autoreconf -i (same version as Linux!!)
Attempt to add autom4te.caches to build on MinGW32
Removing autom4te.cache's as these didn't fix the problem (and shouldn't).
Autoreconf -i after filename updates.
Added tests/fft to test directories.
autoconf -i after adding tests/fft.
Added fft/tune.
Autoreconf -i.
Corrected misspelled filename.
Fix compiler warning in fft_combine_limbs/bits.
Fixed integer overflow warning in gcdext.c.
Fixed missing includes in t-next_likely_prime.c and t-rand.c.
Bumped the library version number and the mpir release number to 2.6.0.
Ran makeinfo.
Updated version.texi and ran makeinfo.
Changed -INTMAX_MIN to INTMAX_MIN in t-set_sx.c.
Added missing ChangeLog items.
Updated date on docs.
Switched back to doing autoreconf -i on eno, as MinGW32 is incapable of handling simple things like making an MPIR lzma tarball.
Updated AUTHORS, NEWS, ChangeLog and corrected version numbers.
Touch configure.in, autoreconf -i.
Revert back to older version of config.guess as it got clobbered by autoreconf -i -f.
Added additional contributors to mpir.texi.
Added reference to Bodrato's paper.
Revert texinfo.tex, as revision 3985 clobbered it.
Fixed a pile of other clobbered files.
Added reference to Bodrato's paper to mpir.texi.
Added small piece on toom 8.5.
Fixed issue with missing symbols in Windows command line and updated release
Corrected NEWS file regarding 32/64 bit integers on Windows.
Attempt to fix intmax_t issue in mpirxx.h.
Try to correct intmax_t fix in mpirxx.h and propagate to = operator.
Second attempt at fixing the intmax_t issue in mpirxx.h.
Change all #if HAVE_LONG_LONG to #ifdef LLONG_MAX
Further attempt to fix the intmax_t issue in mpirxx.h.
Debugging code only.
Try #include stdint.h and limits.h for fun.
Remove support for intmax_t in mpirxx.h on linux.
Added missing underscore on _MSC_VER.
Attempt to add back long long support in mpirxx.h for systems that need it.
Changed a GMP_EXTERN_INLINE to __inline__/__inline.
Replaced inline with macro.
Hopefully fixed a bug in my mpn_addmod_2expp1_1 macro.
Hacked fft tuning code in to tuneup. Also added tuning values for k102.
Made fft multiplication tests do something non-trivial when past fft threshold
Removed symbol conflicts of the form n_blah, with flint.
(i)fft_blah => mpir_(i)fft_blah, n_revbin => mpir_revbin, fermat_to_mpz =>
Added #include "config.h" in all tests which needed it.
Added some additional clarifications about --enable-gmpcompat to the docs.
Slight speedup of fft for integer multiplication.
Added a test for intel compiler when excluding bad gcc-4.3.2.
Moved some more fft prototypes from gmp.h to gmp-impl.h.
Export mpn_sumdiff_n unconditionally.
Ran autoconf -i.
Lots of tuning values.
Sandy Bridge tuning values courtesy of JP Flori.
ARM and AMD Bobcat tuning values courtesty of Julien Puydt and Leif Lionhardy
Updated authors and contributors.
Copied final tuning params across from *nix.
Fixed some bugs in gmp-h.in which were causing test failures in flint.
unsigned long => mpir_ui, long => mpir_si in docs.
Changed default gcc flags to -O2 to avoid multitudinous sins of the GCC
Corrected a typo in manual.
Ran autoconf -i.
Change CFLAGS in another location.
an autoreconf -i again.
-----------------------------------------------------------------------
--
multiprecision integer library forked from gmp
More information about the debian-science-commits
mailing list