split-inventory current (single-root, node-per-leaf CHK dict) stats

Robert Collins robertc at robertcollins.net
Sat Oct 18 10:19:13 BST 2008


CHK:
Commits: 54474
                      Raw    %    Compressed    %  Objects
Revisions:      82786 KiB   0%     43308 KiB   0%    54474
Inventories: 28021377 KiB  92%  14487903 KiB  97%   332965
Texts:        2328229 KiB   7%    274863 KiB   1%   210683
Signatures:         0 KiB   0%         0 KiB   0%        0
Total:       30432393 KiB 100%  14806075 KiB 100%   598122

pack-0.92:
Commits: 54474
                      Raw    %    Compressed    %  Objects
Revisions:      81242 KiB   0%     41880 KiB   8%    54474
Inventories: 74481588 KiB  96%    155399 KiB  32%    54474
Texts:        2328229 KiB   3%    274863 KiB  58%   210683
Signatures:         0 KiB   0%         0 KiB   0%        0
Total:       76891060 KiB 100%    472143 KiB 100%   319631


RAW performance: CHK has 46GB less raw data though inventories are still
92% down from 96%.

Compressed performance: CHK is 14.4GB larger due primarily to inventory
data not being delta compressed.There are also 270K more objects to
index which increases index size substantially (not show in the above
details).

I believe we can get the raw performance of CHK even better, I think
that that is the next place to improve things; group-compress or xdelta
work could start concurrently.

-Rob


-- 
GPG key available at: <http://www.robertcollins.net/keys.txt>.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20081018/67e5c176/attachment.pgp 


More information about the bazaar mailing list