Can you get both high compression ratio and low latency?

By ZeroPoint Media, Friday 5 April 2024

Compression isn't anything new; there are plenty of highly effective algorithms out there that can achieve impressive compression ratios. LZ4 and zSTD are two commonly used ones for storage compression, operating on a granularity ranging from 4kB to several MB pages, which is great for storage purposes.

But when it comes to compressing data in working memory or cache, latency is crucial. Algorithms that require data in the order of kilobytes or megabytes just won't cut it. What's needed is an algorithm that can quickly and efficiently compress small pieces of information, like cache lines of 512b/64B.

Comparing ZeroPoint algorithms to high-performance hardware acceleration implementations of LZ4 and zSTD (which, it's worth noting, aren't widely available on the market), the difference in latency is stark, especially when compared to open available standards. But it's important to acknowledge that this isn't a completely fair comparison. ZeroPoint technology has been developed and optimized specifically to deliver high compression ratios and low latency for small pieces of information, whereas LZ4 and zSTD are optimized for file compression. Nonetheless, the data speaks for itself.

So, is it possible to deliver both high compression ratio and low latency. In fact it is! For those interested in ultra-low latency in the nanosecond domain paired with high compression ratios, ZeroPoint technology is the only solution available on the market today.

Subscribe to our newsletter