Closed GoogleCodeExporter closed 9 years ago
With the release of LZO 2.05 (currently 2.06), the difference in decompression
speeds is ~50% and compression speeds are almost identical**. Also, it's a lot
of effort to get a compression algorithm to kernel mainline. So, I think better
would be to pick optimizations done in newer LZO versions and port them to the
kernel, instead of introducing a new and only marginally better compression
scheme.
** http://permalink.gmane.org/gmane.linux.drivers.driver-project.devel/15608
Original comment by nitingupta910@gmail.com
on 28 Dec 2011 at 4:43
Thanks for your response! I did not know about the new versions of LZO, which
do seem like good improvements. Perhaps this is indeed the better way to go.
How does LZO scale down on low-end systems compared to Snappy? I am thinking of
mobile phones/tables and low-end router/gateways etc. Many benchmarks are on
reasonably fast PCs, while most low-end systems already live with constraints
of memory and cpu.
It may not even be possible to update RAM or CPU on older PCs too.
Original comment by ande...@gmail.com
on 28 Dec 2011 at 5:59
Unfortunately, I do not have access to any such embedded-style (or plain old)
systems but yes, I would surely be interesting to know how Snappy compares to
LZO on such devices.
Original comment by nitingupta910@gmail.com
on 28 Dec 2011 at 6:20
LZO as in the kernel was recently refreshed to the latest version. Not sure if
snappy still holds while still supporting all those archs.
Original comment by nitingupta910@gmail.com
on 2 Oct 2012 at 10:11
Original issue reported on code.google.com by
ande...@gmail.com
on 28 Dec 2011 at 3:59