Contents

• Introduction
• Contributing your improvement to ATLAS
  • Signing up for the ATLAS mailing list
  • License issues
  • Due Credit
  • Inclusion in the ATLAS tarfile
  
  
• Coding Conventions for contributed code
• A note about ATLAS kernels
• Speeding up the Level 3 BLAS
  • Building the General Matrix Multiply From the L1 Cache-contained Multiply
  • The L1 matmul
  • Putting it together with some examples
  • More timing info
  • Complex L1 matmul
  • Providing ATLAS with kernel cleanup code
  • L1 matmul usage notes
  • Getting ATLAS to use your kernel
  • Contributing a complete GEMM implementation
  
  
• Speeding up the Level 2 BLAS
  • Speeding Up GEMV, HEMV, SYMV, TRMV and TRSV
  • Speeding Up GER, GERU, GERC, HER, HER2, SYR and SYR2
  
  
• Speeding up the Level 1 BLAS
  • General comments for Level 1 optimization
  • Testing a kernel
  • Timing a kernel
  • Ramblings on special cases
  • Notes for ROTG, ROTMG, ROT, ROTM
  • Notes for ASUM
  • Notes for AXPBY
  • Notes for AXPY
  • Notes for COPY
  • Notes for CPSC
  • Notes for DOT
  • Notes for IAMAX
  • Notes for NRM2
  • Notes for SCAL
  • Notes for SET
  • Notes for SWAP
  • Getting your new kernel used
  
  
• A quick reference to ATLAS programming resources
  • ATLAS's prefetch header file
  
  
• Conclusion
• About this document ...