3 ms·
A big part of the problem might be xz decompression, that's been my discovery anyway. https://bugzilla.redhat.com/show_bug.cgi?id=1717728 https://bugzilla.redh
by cmurf 7y ago
A big part of the problem might be xz decompression, that's been my discovery anyway.
https://bugzilla.redhat.com/show_bug.cgi?id=1717728 https://bugzilla.redhat.com/show_bug.cgi?id=1717728
- usr1106 7y agoI tried various compression algorithms. I don't think there was a CPU bottleneck, at least not with the less aggressive compressions. blktrace showed the difference compared to ext4, squashfs does all reads one by one block. It was in a previous job. I don't have access to the details anymore. And the kernel was not the newest. But squashfs looked unmaintained already then and that's what they're saying elsewhere in the discussion. So I fear nothing has changed.