DGEMM for Alpha Chip This is an optimized BLAS(Level 1, 3 and some 2) and some lapack(GESV, LASWP, GETF2, GETRF) library for Alpha. And some of Level 3 routines(GEMM, TRSM, TRMM, SYMM) and lapack(GETRF, GESV) are parallelized by posix thread. Now, Linpack TPP performs 1069 MFlops(ES40, 21264 677MHz DDR 2nd cache, Tru64 Unix V 5.1, this is a peak performance. The avarage is about 1060MFlops) faster than Compaq's announce performance(Compaq's performance is 1031MFlops). On Linux, TPP performs a little bit slower. If you want to use SMP version, you should link this library with libpthread.a("-lpthread" link option) and you can specify the number of CPUs by environment value "STATABO_NUM_THREADS"(currently, the maximum number of CPUs is 64). If this environment value does not exist, the library assume that you have 2 CPUs.

Current Version:   20000719 License Type:   GPL

Home Site: http://members.jcom.home.ne.jp/kgoto/ Source Code Availability: Yes Available Binary Packages: Debian Package:   No RedHat RPM Package:   No Other Packages:   No Targeted Platforms: Linux/Alpha and Tru64 UNIX Software/Hardware Requirements: The EV4/EV5/EV6 architectures are automatically and dynamically detected by this library. So you don't have to take care.

Other Links: http://cnls.lanl.gov/avalon/ (Avalon, an Alpha Linux cluster which uses DGEMM for Alpha software) Mailing Lists/USENET News Groups: None User Comments: None See A Screen Shot? (Not Yet)

  SAL Home   |   Numerical Analysis   |   Misc