How to use Multi-core Compression
What is Multi-core?
A core is a processing unit in a CPU, and a CPU with two or more cores is called a multi-core CPU. Multi-core CPUs have become very common these days. Four or eight-core CPUs are widely used, and even CPUs for mobile devices are made as multi-core CPUs. Therefore, whether a software supports multi-core CPUs is an essential factor in evaluating its performance.
Bandizip supports multi-core compression. Since data compression requires a lot of CPU work, using multi-core helps to improve the compression speed.
For example, Bandizip compresses your file(s) 6 times faster with 4 cores 8 threads CPU, and 13 times faster with 8 cores 16 threads CPU.
The pictures below show the comparison between Bandizip that supports multi-core and Mac Finder that only uses single-core.
Why is multi-core not used during decompression?
Compression with Bandizip increases CPU usage to the maximum, but decompression doesn’t increase the usage to that extent.
This is caused by a characteristic of data compression algorithms. Most data compression algorithms use a method of referring to front data. Due to this, not like its reversal, decompression is almost impossible to be processed in parallel (parallelization) and even with a large number of cores, the speed improvement is minimal.
Why is CPU usage not high during compression?
There may be a problem that compression doesn’t increase the usage either. Compression is processed by the following steps:
Reading the file to compress (Storage) ▶ Compression (CPU) ▶ Writing to the archive (Storage)
In general, a bottleneck occurs during the compression, resulting in high CPU usage. However, in some cases, such as using slow-speed media (external USB drive, network drive, etc.) a bottleneck occurs on storage I/O and therefore CPU usage is not high.
- When you compress multiple files of small size (less than 100KB) the speed improvement is minimal even with multi-core.
- 7zip supports multi-core compression when compressing your file(s) in ZIP format, but it works only for multiple files and not for a single file.