Feature Overview

Here is a quick overview of features that make PCVS a robust solution different than other validation engines.

Run the validation from anywhere

After installation, PCVS can be configured in a minute for the current directory:

$ pcvs scan .
$ pcvs run

Scale a test-suite depending on the target machine

From PCVS approach, a benchmarks is a collection of programs. coupled with a compiler and a launcher, sets of tests can be build to dynamically adapt the machine to test. With a single edition, a test-suite can be ported from a validating an MPI implementation on a simple workstation (no tests requiring more than one node & 4-8 MPI processes) to largest supercomputers (thousands of nodes). PCVS allow this thanks to criterions, a variadic component to apply to a program to build tests. It may be populated as follows in a profile:

criterion:
    iterators:
        number_of_mpi_processes:
            values: [1, 2, 4, 8, 16, 32]

Automatic Test-suite builder

PCVS relies on a specific test description syntax in order to build an efficient test-suite. To help with that process, PCVS can pre-generate templates:

$ pcvs scan /dir

Definition - Execution - Reporting in one place

One main advantage of PCVS is the capability to gather all validation modules in one single place, easy to install as a single user. Among others:

  • Highly customizable test generation framework

  • Orchestrator designed to run tests at scale

  • Autonomous reporting web platform.

  • Store results persistently as a Git repository for easy imports/exports & validatino progression.