Skip to content

Latest commit

 

History

History
23 lines (12 loc) · 1.47 KB

README.md

File metadata and controls

23 lines (12 loc) · 1.47 KB

PARSEC Benchmark with tweaks (hardcoded ROI computation, parallel make)

This is a fork of the Parsec benchmark fork modified for cross-compilation, with a few tweaks.

Hardcoded ROI computation

I need to natively compile the binaries for RISC-V, and display the ROI on execution, but the gcc-hooks-configured builds failed with obscure ASM-related errors.

Since I couldn't find out a clean way to achieve this compilation, I just hardcoded the ROI routines.

Note that the parsec-benchmark fork not only improves cross-compilation compatibility, but also fixes some issues on native (RISC-V) compilation. Some benchmarks can't be cross-compiled because of x86 ASM code (raytrace and x264).

Parallel make

The parsecmgmt script runs invokes make in sequential mode (without the -j option); I've added the jobs parameter with one recipe per processor (-j $(nproc)), in order to use speed up the compilation as much as possible.

GCC 10 compatibility

At least two packages (dedup, splash2x.raytrace, and likely the splash2 counterpart) require a change in order to compile on GCC 10, which I've applied to its makefile.

dedup: Remove provided, non-portable, SSL library dependency

The dedup package relies on the provided SSL library, which is non-portable (as it has x86 ASM). At least in the gcc-pthread configuration mode, it can be removed without side effects though, making dedup compilable on riscv64.