Main Menu
Links

Parallel Numerical Linear Algebra Libraries
  • Speed-up version developed for HITACHI SR2201 has been used by 15 customs, including Cambridge University, UK. Average speedup is 1.42 for BLAS and 1.73 for ScaLAPACK
  • Speed-up version developed for Teraflops parallel computer HITACHI SR8000 has been used by 9 customs, including Leibniz Computing Center of Germany. Average speedup is 6.39 for BLAS, 2.58 for LAPACK and 2.33 for ScaLAPACK
  • Installed on Domestic parallel computing system SHENWEI-I, DAWNING 1000A and DAWNING 2000

PQR: Generalized Eigenvalue Problems Solver Package
  • One of the kernel mathematical function of material and quantum scientific computing, problem size can reach 104 order
  • PQR has been used or bought by Institute of Physics(CAS), FuDan University of P.R. China; Ames Lab., North West University, Texas University of America, Chalmers University of Sweden, etc.
  • Supported platform including Cray T3D¡¢Cray T3E¡¢SGI Origin 2000¡¢ONYX¡¢DAWNING 1000A.
  • Speedup is larger than 2 compared with original installed software with the same function on above systems.
PRIS: Parallel Solver for Numerical Simulation Large Scale Oil Reservoir
  • Parallel solver is the kernel of parallel numerical oil reservoir simulation
  • Available for Shared Memory and Distributed Memory parallel computing systems
  • First time 3D long time simulation of million grid points large scale oil reservoir simulation using production data on domestic parallel computers with PRIS of P.R. China
  • Released the Parallel version WorkBench with Bake Hughes SSI on Nov. 2000


ParaMidSQL
  • Designed for PC Cluster
  • Shared Nothing Architecture
  • Combined MPI with Open Source DBMS, e.g. MySQL or PostgreSQL
  • Focus on Scalability and High Performance
  • Provides API of Parallel SQL execution for common users
  • Try our best not to modify kernel DBMS function, keep simple
ParaMidZQuery
  • OBWT
        • A losless data transformation method
        • Increasing compressibility after transformation
        • support fast indexing
  • Key Tech.
        • BWT
        • Suffix Array
        • Huffman &PC
  • Key Tech. Proposed
        • Overlapping Blocked Compressing
        • Parallel Compressing Algorithm
        • DBF Query Tech.
        • Research paper published in proceedings of ECIR2007
Research&Development Center for Parallel Software,Institute of Software, Chinese Academy of Sciences
Contact Webmaster