Skip to content

Latest commit

 

History

History
70 lines (52 loc) · 2.61 KB

README.md

File metadata and controls

70 lines (52 loc) · 2.61 KB
Raw

SYCL Academy

Exercise 4: Handling Errors

In this exercise you will learn how to handle synchronous and asynchronous exceptions.

1.) Catch synchronous exception handling

Synchronous errors, such as failure to construct an object, are reported immediately by the runtime throwing an exception.

To catch synchronous exceptions simply wrap your application in a try catch block and catch a SYCL exception and print out the error message provided by the what member function.

2. ) Catch and asynchronous exceptions

Asynchronous errors, such as an error occurring during execution of a kernel on a device, are reported via user-supplied asynchronous error-handlers.

To catch asynchronous exceptions create an async handler using a lambda expression which takes an exception_list, iterates over it and rethrows the exceptions to be caught by the surrounding try-catch block.

Build And Execution Hints

For For DPC++ (using the Intel DevCloud):

clang++ -fsycl -o sycl-ex-4 -I../External/Catch2/single_include ../Code_Exercises/Exercise_04_Handling_Errors/source.cpp

In Intel DevCloud, to run computational applications, you will submit jobs to a queue for execution on compute nodes, especially some features like longer walltime and multi-node computation is only abvailable through the job queue. Please refer to the guide.

So wrap the binary into a script job_submission and run:

qsub job_submission

For ComputeCpp:

cmake -DSYCL_ACADEMY_USE_COMPUTECPP=ON -DSYCL_ACADEMY_INSTALL_ROOT=/insert/path/to/computecpp ..
make exercise_04_handling_errors_source
./Code_Exercises/Exercise_04_Handling_Errors/exercise_04_handling_errors source

For hipSYCL:

# <target specification> is a list of backends and devices to target, for example
# "omp;hip:gfx900,gfx906" compiles for CPUs with the OpenMP backend and for AMD Vega 10 (gfx900) and Vega 20 (gfx906) GPUs using the HIP backend.
# The simplest target specification is "omp" which compiles for CPUs using the OpenMP backend.
cmake -DSYCL_ACADEMY_USE_HIPSYCL=ON -DSYCL_ACADEMY_INSTALL_ROOT=/insert/path/to/hipsycl -DHIPSYCL_TARGETS="<target specification>" ..
make exercise_04_handling_errors_source
./Code_Exercises/Exercise_04_Handling_Errors/exercise_04_handling_errors_source

alternatively, without cmake:

cd Code_Exercises/Exercise_04_Handling_Errors
/path/to/hipsycl/bin/syclcc -o sycl-ex-4 -I../../External/Catch2/single_include --hipsycl-targets="<target specification>" source.cpp
./sycl-ex-4