ABSTRACT
With computers processing an ever-increasing amount of data, expanding
DRAM capacity is not always feasible due to economic and
structural issues. In this context, memory compression emerges
as an alternative provided by the ZRAM module, which expands
the DRAM’s data capacity by using a reserved area to store compressed
pages directly into the main memory. Therefore, reducing
secondary storage accesses caused by page swaps and potentially
reducing the program’s execution time. This paper evaluates the
ZRAM performance, comparing the speedup obtained when running
NAS Parallel Benchmarks under different disksize values for
ZRAM, in systems with 8 GB, 4 GB, and 1 GB of DRAM. Also, we
compared our baseline system with an older one, to study the effect
of the systems improvement over the ZRAM usage impacts.
Our results indicate that the impact of ZRAM on the application’s
execution time increases with the reduction of the available main
memory space. We show no variations in speedup with 8 GB of
DRAM, and up to 10% improvement in scenarios with 4 GB of
memory. In the 1 GB scenario, compression allowed certain benchmarks
to achieve speed increases of up to 39%, even with a low
memory footprint application, although most others did not benefit.
We conclude that ZRAM has a more pronounced impact in memoryconstrained
environments, and can have a significant impact even
in applications with a low memory footprint. The code is available
at https://github.com/Guilherme-Eduardo/ZRAM.