XSBench v20

Adds HIP version of XSBench and other small bugfixes.

XSBench Version 19

This update adds formal support for accelerator architectures using a variety of programming models. XSBench is now implemented using five different models, each in their own source base:

• OpenMP threading
• OpenMP target offloading
• CUDA
• OpenCL
• SYCL

Full details on the changes introduced in v19 can be found in CHANGES.txt file.

XSBench Version 18

This update introduces an option to run a simplified event based method using the "-m event" option. The event based simulation is an alternative to the default history based simulation. The main difference between the two methods is how parallelism is expressed and whether macroscopic cross sections lookups are independent or not. The event based model does not capture costs associated with buffering particle states, but is still useful for determining if there are benefits of the method on various computer architectures due to its very different style of parallelism.

See the CHANGES file for more details.

XSBench Version 17

See CHANGES for details.

XSBench v14

This version adds in the ability (triggered by the -G command line argument) to run using only the nuclide grids and no unionized energy grid. When run without the -G flag, it is identical to v13 of XSBench.

XSBench v13

fixed printf specified in io