summaryrefslogtreecommitdiff
path: root/lapack
AgeCommit message (Collapse)AuthorFilesLines
2014-05-23enabled abd tested optimized trtri lapack functionswernsaar3-332/+167
2014-05-21removed lapack/getri because it was never usedwernsaar4-774/+0
2014-05-21enabled optimized trti2 lapack functions againwernsaar1-1/+1
2014-05-21enabled lauu2 and lauum lapack functions againwernsaar1-1/+1
2014-05-18enabled and tested optimized potrf lapack functionswernsaar1-1/+1
2014-05-18enabled and tested optimized getrs lapack functionswernsaar1-1/+1
2014-05-18enabled and tested optimized dgetrf functionwernsaar1-1/+1
2014-05-07changed makefiles for lapack developmentwernsaar1-1/+2
2013-12-01Merge remote branch 'origin/develop' into armv7wernsaar7-260/+25
2013-11-24added experimental support for ARMV8wernsaar1-0/+33
2013-11-14Refs #283. Fixed the incorrect usage of long data type for Windows 64.Zhang Xianyi4-10/+10
2013-11-07Refs #309. Fixed trtri_U single thread computational bug.Zhang Xianyi3-250/+15
2013-11-03added missing file arm/Makefile in lapack/laswpwernsaar1-0/+33
2013-07-31Refs #266. Fixed the compiling bug with Open64 5.0.Zhang Xianyi1-1/+1
2013-07-14Refs #191. A walk around for dtrtri_U single thread bug.Zhang Xianyi2-0/+247
This function caused the failure of ERKALE serial test. I replaced it with LAPACK source code.
2013-07-11Refs #214, #221, #246. Fixed the getrf overflow bug on Windows.Zhang Xianyi2-2/+2
I used a smaller threshold since the stack size is 1MB on windows.
2013-07-08Refs #221 #246. Fixed the overflowing stack bug in mutlithreading BLAS3.Zhang Xianyi2-1/+49
When NUM_THREADS(MAX_CPU_NUNBERS) is very large ,e.g. 256. typedef struct { volatile BLASLONG working[MAX_CPU_NUMBER][CACHE_LINE_SIZE * DIVIDE_RATE]; } job_t; job_t job[MAX_CPU_NUMBER]; The job array is equal 8MB. Thus, We use malloc instead of stack allocation.
2013-02-23Fixed the overflowing bug in single thread cholesky factorization.Zhang Xianyi1-2/+4
2012-08-09Refs #130 Fixed laswp building bug with DYNAMIC_ARCH=1.Zhang Xianyi2-0/+10
2012-08-09Refs #130 Prevent reading ipiv array beyond the bound in ?laswp. Use laswp ↵Zhang Xianyi12-668/+2687
instead of laswp_oncopy in getrf.
2011-01-24Import GotoBLAS2 1.13 BSD version codes.Xianyi Zhang66-0/+11503