1 |
Hello, |
2 |
|
3 |
On Sun, 09 Nov 2014 00:15:56 -0800 Zac Medico wrote: |
4 |
> On 11/08/2014 10:53 PM, Andrew Savchenko wrote: |
5 |
[...] |
6 |
> > On old hardware it may take dozens of minutes of CPU time. I have |
7 |
> > that *.pickle files, I have sqlite metadata cache, I have 100% CPU |
8 |
> > usage. It's not an I/O problem. Just take into account that due to |
9 |
> > instruction set, Lx cache, frequency and memory speed difference |
10 |
> > old CPU-based system may be 20x times slower than recent one. |
11 |
> |
12 |
> It could be useful for us to collect profiling data generated on old |
13 |
> hardware. If you'd like to help, you can use python's cProfile module to |
14 |
> generate statistics for us to analyze. The statistics can be nicely |
15 |
> visualized as a shaded call graph by using gprof2dot and graphviz [1]. |
16 |
|
17 |
Sorry for delay, I generated samples on two old hosts. |
18 |
|
19 |
Tarball contains per host directories. Each one contains: |
20 |
- pstats file; |
21 |
- generated pdf with call graphs and timing; |
22 |
- host-related information: |
23 |
* emerge --info |
24 |
* /proc/cpuinfo |
25 |
* /proc/memnifo |
26 |
|
27 |
*.pstats and *.pdf filename should describe command unambiguously, |
28 |
e.g. emerge-pv_python:2.7_python:3.3-python-3.3.5-r1 means: |
29 |
emerge -pv python:2.7 python:3.3 while using python-3.3.5-r1 as |
30 |
interpreter. |
31 |
|
32 |
hitomi system was not fully updated for a bit more than a year, |
33 |
while another one for about half a year. So dependency calculations |
34 |
may be of different intencities. Sets of packages installed are |
35 |
similar but not the same: |
36 |
2502 on hitomi |
37 |
2953 on desktop |
38 |
|
39 |
I understand that most interesting data will be from emerge -DNupv |
40 |
output, but to obtain it I must first fix all blockers and |
41 |
conflicts, this really takes a lot of time, so please be patient. |
42 |
|
43 |
And last but not the list: on both systems portage is configured to |
44 |
use sqlite3 cache for metadata, so there was no I/O delays (CPU |
45 |
usage about 100% according to top). |
46 |
|
47 |
In order not to abuse mail list with large attachments, data is |
48 |
available here: |
49 |
ftp://brunestud.info/gentoo/portage.tar.xz |
50 |
|
51 |
Best regards, |
52 |
Andrew Savchenko |