Examples¶

After running Pkg.test("ParallelAccelerator"), a good next step is to try out some example programs from the ParallelAccelerator examples/ subdirectory. These can be run either from the command line or from the Julia prompt. For instance, to run the black-scholes example from the Julia prompt:

julia> include("$(homedir())/.julia/v0.5/ParallelAccelerator/examples/black-scholes/black-scholes.jl")

On your first run, you should see output similar to the following (this was on an 8-core Linux machine with 8 GB of RAM, using ICC and MKL):

iterations = 10000000
SELFPRIMED 31.675743032
checksum: 2.0954821257116845e8
rate = 2.268548657657791e7 opts/sec
SELFTIMED 0.44081047

The SELFTIMED line in the printed output shows the running time, while the SELFPRIMED line shows the time it takes to compile the accelerated code and run it with a small “warm-up” input.

For the first function you accelerate in a given Julia session, you might notice a long time being reported for SELFPRIMED. This delay (about 25 to 30 seconds on an 8-core desktop machine) is the time it takes for Julia to load the ParallelAccelerator package itself. If you run the function again, you’ll notice that SELFPRIMED gets smaller. Here’s an example interaction:

julia> include(“$(homedir())/.julia/v0.5/ParallelAccelerator/examples/black-scholes/black-scholes.jl”) iterations = 10000000 SELFPRIMED 31.675743032 checksum: 2.0954821257116845e8 rate = 2.268548657657791e7 opts/sec SELFTIMED 0.44081047

julia> include(“$(homedir())/.julia/v0.5/ParallelAccelerator/examples/black-scholes/black-scholes.jl”) iterations = 10000000 SELFPRIMED 1.62395378 checksum: 2.0954821257116845e8 rate = 2.3933823208592944e7 opts/sec SELFTIMED 0.417818746

julia>

Notice that SELFPRIMED dropped from almost 32 seconds to about 1.6 seconds. This is because, for the second run, the ParallelAccelerator package was already loaded. The remaining 1.6 seconds is mostly the time it took for the ParallelAccelerator compiler to compile the accelerated code.

It’s instructive to compare the running time of the ParallelAccelerator code with the plain Julia version:

julia> include(“$(homedir())/.julia/v0.5/ParallelAccelerator/examples/plain-julia/black-scholes/black-scholes.jl”) iterations = 10000000 SELFPRIMED 0.000573466 checksum: 2.0954821257116845e8 rate = 3.640941015492507e6 opts/sec SELFTIMED 2.746542709

This time, SELFPRIMED is almost 0 (because there is no package to be loaded and no work for the ParallelAccelerator compiler to do), but SELFTIMED is rather larger than it was under ParallelAccelerator:: 2.7 seconds.

You can also run example programs from the command line:

$ julia ~/.julia/v0.5/ParallelAccelerator/examples/black-scholes/black-scholes.jl

Pass the --help option to see usage information for each example:

$ julia ~/.julia/v0.5/ParallelAccelerator/examples/black-scholes/black-scholes.jl -- --help
black-scholes.jl

Black-Scholes option pricing model.

Usage:
  black-scholes.jl -h | --help
  black-scholes.jl [--iterations=<iterations>]

Options:
  -h --help                  Show this screen.
  --iterations=<iterations>  Specify a number of iterations [default: 10000000].

The examples/ subdirectory contains many more examples. Most examples also have a corresponding plain Julia version available in examples/plain-julia/. Some examples require additional Julia packages to be installed.