summaryrefslogtreecommitdiff
path: root/mpn/x86_64
Commit message (Collapse)AuthorAgeFilesLines
* Provide Nehalem redc_1.Torbjorn Granlund2013-09-211-0/+534
|
* Fix typo.Torbjorn Granlund2013-09-211-1/+1
|
* Short-circuit a branch.Torbjorn Granlund2013-09-211-3/+2
|
* Add "Contributed ..." lines.Torbjorn Granlund2013-09-2011-3/+25
|
* Provide Sandy/Ivy bridge redc_1.Torbjorn Granlund2013-09-201-0/+531
|
* Complete rewrite.Torbjorn Granlund2013-09-201-191/+441
|
* Provide mullo_basecase for Conroe, Wolfdale, Nehalem, Westmere.Torbjorn Granlund2013-09-201-0/+416
|
* Postpone pushes, short-circuit a branch.Torbjorn Granlund2013-09-201-12/+25
|
* Whitespace cleanup.Torbjorn Granlund2013-09-192-2/+2
|
* Allocate more stack under DOS.Torbjorn Granlund2013-09-191-2/+2
|
* Cosmetic fix to mulx byte output.Torbjorn Granlund2013-09-181-1/+1
|
* Provide mul_basecase and sqr_basecase for Conroe, Wolfdale, Nehalem, Westmere.Torbjorn Granlund2013-09-182-0/+1933
|
* Cleanup, streamline.Torbjorn Granlund2013-09-182-142/+146
|
* Provide SBR and HWL mullo_basecase.Torbjorn Granlund2013-09-182-0/+776
|
* Preserve xmm6-xmm8 under DOS.Torbjorn Granlund2013-09-161-2/+12
|
* Use R8 for bit testing.Torbjorn Granlund2013-09-151-2/+2
|
* Correct c/l table.Torbjorn Granlund2013-09-151-2/+2
|
* Complement c/l table.Torbjorn Granlund2013-09-151-1/+1
|
* Complement c/l table.Torbjorn Granlund2013-09-151-3/+11
|
* Replace mul_1 code.Torbjorn Granlund2013-09-151-128/+128
|
* Rewrite for a slight speed-up for small and large operands.Torbjorn Granlund2013-09-151-82/+96
|
* Implement temp fix to properly handle overlap.Torbjorn Granlund2013-09-121-3/+3
|
* Really handle overlap correctly in rp != up (mod 16) code.Torbjorn Granlund2013-09-101-33/+31
|
* Rewrite rp != up (mod 16) code to make it handle any allowed overlap.Torbjorn Granlund2013-09-101-46/+70
|
* Refresh cycle table.Torbjorn Granlund2013-09-101-5/+10
|
* Remove accidentally left-in debugging code.Torbjorn Granlund2013-09-091-2/+2
|
* Use fastsse com for atom.Torbjorn Granlund2013-09-091-0/+26
|
* Use some fastsse code for bd1.Torbjorn Granlund2013-09-094-0/+89
|
* Provide basic AVX copyi/copyd.Torbjorn Granlund2013-09-092-0/+315
|
* Streamline.Torbjorn Granlund2013-09-051-9/+10
|
* Retune.Torbjorn Granlund2013-09-053-259/+225
|
* Extend FFT tables.Torbjorn Granlund2013-09-051-33/+75
|
* Retune.Torbjorn Granlund2013-09-041-32/+32
|
* Use macro mulx consistently.Torbjorn Granlund2013-09-041-6/+6
|
* Implement larger "corner".Torbjorn Granlund2013-09-041-55/+73
|
* Provide Haswell redc_1.Torbjorn Granlund2013-09-031-0/+422
|
* (mulx): Handle negative offsets.Torbjorn Granlund2013-09-031-7/+7
|
* Retune.Torbjorn Granlund2013-09-011-114/+67
|
* Tweak for slightly better speed for relevant CPUs.Torbjorn Granlund2013-08-311-8/+8
|
* Provide SBR/IBR sqr_basecase.Torbjorn Granlund2013-08-311-0/+473
|
* Fix comment typo.Torbjorn Granlund2013-08-316-6/+6
|
* Provide basic x86_64 sqr_diag_addlsh1 for sqr_basecase-less CPUs.Torbjorn Granlund2013-08-311-0/+103
|
* More of: Provide fat build fallback files.Torbjorn Granlund2013-08-301-0/+21
|
* Provide fat build fallback files.Torbjorn Granlund2013-08-303-0/+63
|
* Move x86_64 top-level basecase files to k8 subdir.Torbjorn Granlund2013-08-295-0/+0
|
* Support Haswell.Torbjorn Granlund2013-08-291-0/+6
|
* Cosmetic change.Torbjorn Granlund2013-08-291-2/+1
|
* More of last change.Torbjorn Granlund2013-08-296-1622/+0
|
* Handle mulx insn using m4, allowing use without assembler support.Torbjorn Granlund2013-08-297-0/+1681
|
* Fix typo.Torbjorn Granlund2013-08-291-1/+1
|