Hi,
The xz (lzma) library is ported to nanonote. And also my zim benchmark
program.
Lzma seems to be more problematic than expected. Due to memory limitations
zimlib is not able to decompress any data on the nanonote with xz. Even the
original command line xz tool fails to run. There is a memory limit parameter
in xz and I've played with it. 64M (which is far beyond the available RAM of
32M) is too less for the decompressor and 128M it starts but fails to allocate
memory. I sent a mail to the author if there are other tweaks to reduce memory
consumption in the decompressor.
So I was not able to run any benchmarks with lzma. But I tested with bzip2.
The device needs about 2 seconds to decompress the data. With the benchmark
program it is no problem to increase the cluster cache size to 8, so that it
caches the last 8 clusters. I think a default cluster cache size of 5 would be
reasonable for that device.
Tommi
Show replies by date