Where is the discussion happening?

I ran the same benchmarks for my i7-6770HQ 4-core system. This really
needs revising.

While disk space using in /boot is a concern, in this example at least
-10 would use only 8MB (10%) more space and cut the time taken from 2m1s
to 13s.

zstd.0 84M 0m2.150s
zstd.1 96M 0m1.236s
zstd.2 90M 0m1.350s
zstd.3 84M 0m2.235s
zstd.4 84M 0m3.355s
zstd.5 81M 0m5.679s
zstd.6 81M 0m7.416s
zstd.7 78M 0m8.857s
zstd.8 77M 0m10.134s
zstd.9 77M 0m11.238s
zstd.10 72M 0m13.232s
zstd.11 72M 0m14.897s
zstd.12 72M 0m19.343s
zstd.13 72M 0m26.327s
zstd.14 72M 0m30.948s
zstd.15 72M 0m40.913s
zstd.16 70M 0m59.517s
zstd.17 66M 1m15.854s
zstd.18 64M 1m36.227s
zstd.19 64M 2m1.417s

-- 
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to initramfs-tools in Ubuntu.
https://bugs.launchpad.net/bugs/1958148

Title:
  mkinitramfs is too slow

Status in initramfs-tools package in Ubuntu:
  In Progress

Bug description:
  On the Nezha board creating an initrd takes more than 1 hour. The
  compression level -19 for zstd is unwisely chosen.

  Here are durations and compression results for an SiFive Unmatched
  board with four cores using zstd:

  -T0 -1 => 13.92s, 136298155 bytes
  -T0 -2 => 15.73s, 131717830 bytes
  -T0 -3 => 26.11s, 127434653 bytes
  -T0 -4 => 29.31s, 126924540 bytes
  -T0 -5 => 36.44s, 125296557 bytes
  -T0 -6 => 39.36s, 124781669 bytes
  -T0 -7 => 46.56s, 116200665 bytes
  -T0 -8 => 51.95s, 113172941 bytes
  -T0 -9 => 55.89s, 112835937 bytes
  -T0 -10 => 61.32s, 108326876 bytes
  -T0 -11 => 64.32s, 108115060 bytes
  -T0 -12 => 76.37s, 108016478 bytes
  -T0 -13 => 148.99s, 109121308 bytes
  -T0 -14 => 156.58s, 108908574 bytes
  -T0 -15 => 228.64s, 109213554 bytes
  -T0 -16 => 380.26s, 107260643 bytes
  -T0 -17 => 453.36s, 103679714 bytes
  -T0 -18 => 714.79s, 100402249 bytes
  -T0 -19 => 1046.58s, 100188713 bytes

  Compression level between -2 to -10 offer a good compromise between
  CPU time and compression results.

  Ideally there would be a parameter that we could pass to mkinitfs. But
  for as fast solution we should simply replace -19 by -9 in
  mkinitramfs.

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/initramfs-tools/+bug/1958148/+subscriptions


-- 
Mailing list: https://launchpad.net/~touch-packages
Post to     : touch-packages@lists.launchpad.net
Unsubscribe : https://launchpad.net/~touch-packages
More help   : https://help.launchpad.net/ListHelp

Reply via email to