MPIR - Parallel Algorithms and CUDA
Present : Carl Witty, Bill Hart, Michael Abshoff, Glenn Tarbox Virtually Present : Jeff Gilchrist, Gonzalo Tornaria
You can chat in a Linux text console by installing "irssi" and running: "irssi -c irc.freenode.net" and then type "/join #sage-devel"
- Multimodular algorithms
- Scalar algorithms
Peter Montgomery's remainder algorithm a mod b, precompute b1 = B mod b, b2 = B2 mod b, b3 = B3 mod b, then write a = a0 + a1*B + a2*B^2 +..., then compute a0 + a1*b1 + a2*b2 +.... and do final reduction mod b. Multiplications can be done in parallel.
- Addition and subtraction can be parallelised using nails - non-unique representation of numbers
- Classical algorithm is embarrassingly parallel - bad if you have an n log n algorithm in that range
Glenn Tarbox (Owner of cuda1, AMD K10 with NVIDA CUDA card - expert on large scale parallelisation)
- What are the top level integration issues, e.g. by libraries using MPIR
Michael Abshoff (Sage release manager)
- Link into Sage via cython and link in CUDA
- Memory bandwidth limits algorithms - matrices n**2 entries to get in and out, matrix multiplication O(n**2.7), but for integers n limbs to get in and out O(n log n log log n) operations to multiply
- AMD Math library AML provides BLAS interface uses GPU - but that's for linear algebra
- PTX NVIDIA GPU assembler code for inner loops
Gonzalo Tornaria (theta functions expert)
- Is there a way to encode integer multiplication in linear algebra? (A. Perhaps vectors - multimodular, but not matrices)
Launch threads - issues based on hierarchy of memory - CPU registers-> memory per processor block-> main graphics memory-> system memory
- Can launch all the threads on all cpus in a couple of cycles
- How GPU would compare to carefuly programmed FPGA?
- E.g a Stratix IV can have around 1000 18x18 multipliers, but maybe that's not too much, and this is probably very expensive hardware
- Carl Witty does FPGA programming - says it is probably very expensive
- Accoding to the spec the stratix can have parallel high-bandwith communication
- Up to 48 8.5Gb/s tranceivers or 24 11.3 gbps tranceivers
- Deal with 533MHz DDR3 memory directly
- What about ATI hardware - why not support OpenCL?
- Carl Witty says - Nobody ships OpenCL yet but in a year or two it should be common
- Intel has Larrabee and will ship some intel library, NVIDIA has CUDA and PTX, ATI has AML (basically BLAS) + low level interface - OpenCL is slated to be common
- OSX will ship with OpenCL
Cell port will happen as it is funded by EPSRC Grant - will be proof of principle code to apply for a port to Cell2Xi
- What is the focus for mathematicians? E.g. mathematica is reputedly going to have parallel CUDA or something similar