Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with
or
.
Download ZIP
Delayed processing of Vector / Matrix expression in Julia with various backends (Julia,CUDA-PTX,...)
Julia C
Branch: be-julia-tst2
Pull request Compare This branch is 30 commits behind master.

Fetching latest commit…

Cannot retrieve the latest commit at this time

Failed to load latest commit information.
.gitignore
README
demat.jl
demat_base.jl
demat_be_julia.jl
demat_crash_julia.jl
demat_demo.jl
libcuda.jl

README

## Delayed Expressions README
# Copyright 2012, Krzysztof Kamieniecki (krys@kamieniecki.com)

## USAGE
For DEMO load("demat_demo.jl") from julia command line
To use load("demat.jl")

## TODO
# * need scaler / parameter types so that a new kernel is not generated for every new 
# host side value of a parameter, the user should decide where the placeholders are used
# * OpenCL is portable and in program buildable but cannot work with Nvidia Cu* libraries (like CuBLAS)
# * PTX is low level and not portable, but is in program buildable and can work with Nvidia libraries
# * de_barrier() or de_barrier(:x,:y) to extract only x & y ?
# * x[] = v (assignment)
# * x = v[] (ref)
# * x[a:s:b] = v (sub view assignment)
# * x[a:b] = v (sub view assignment)
# * x[] = v[a:s:b] (sub view extraction)
# * x[] = v[a:b] (sub view extraction)
# * (d[c] = a[c]; [~c] = b[~c]) should be converted to "for i = 1:N d[i] = c[i]?a[i]:b[i] end"

# * A BUNCH OF OTHER STUFF

Something went wrong with that request. Please try again.