View Single Post
Old 05-18-2020, 07:07 PM   #92
sherman
Guru
sherman ought to be getting tired of karma fortunes by now.sherman ought to be getting tired of karma fortunes by now.sherman ought to be getting tired of karma fortunes by now.sherman ought to be getting tired of karma fortunes by now.sherman ought to be getting tired of karma fortunes by now.sherman ought to be getting tired of karma fortunes by now.sherman ought to be getting tired of karma fortunes by now.sherman ought to be getting tired of karma fortunes by now.sherman ought to be getting tired of karma fortunes by now.sherman ought to be getting tired of karma fortunes by now.sherman ought to be getting tired of karma fortunes by now.
 
Posts: 628
Karma: 1318712
Join Date: Aug 2008
Location: Auckland - NZ
Device: Kobo Aura H2O, Kobo Forma
Quote:
Originally Posted by NiLuJe View Post
(TL;DR: All hail our evil overlord Facebook for coming up with zstandard, because zlib is the absolute pits).
Interestingly I did some quick and dirty compression tests on my Aura H2O a few weeks ago for a project that may or may not happen. Speed wise, there didn't seem to be an awful lot of difference between zstd and zlib TBH. Which makes sense, because from what I've read, most of the performance uplift comes from multithreading, and maybe some architecture specific optimizations, with a Cortex A8 class CPU being very low on the priority list.

Now granted, the compression ratio may be better, but I wasn't to concerned about that for my specific desires.

If you ever want speed, LZ4 is waaaaaay faster. But that comes at the expense of compression ratio, so you win some, and lose some.
sherman is offline   Reply With Quote