1 |
El mar, 30-06-2020 a las 08:20 +0200, Fabian Groffen escribió: |
2 |
> Hi, |
3 |
> |
4 |
> On 29-06-2020 21:13:43 -0500, Sid Spry wrote: |
5 |
> > Hello, |
6 |
> > |
7 |
> > I have some runnable pseudocode outlining a faster tree verification |
8 |
> > algorithm. |
9 |
> > Before I create patches I'd like to see if there is any guidance on making |
10 |
> > the |
11 |
> > changes as unobtrusive as possible. If the radical change in algorithm is |
12 |
> > acceptable I can work on adding the changes. |
13 |
> > |
14 |
> > Instead of composing any kind of structured data out of the portage tree my |
15 |
> > algorithm just lists all files and then optionally batches them out to |
16 |
> > threads. |
17 |
> > There is a noticeable speedup by eliding the tree traversal operations which |
18 |
> > can be seen when running the algorithm with a single thread and comparing it |
19 |
> > to |
20 |
> > the current algorithm in gemato (which should still be discussed here?). |
21 |
> |
22 |
> I remember something that gemato used to use multiple threads, but |
23 |
> because it totally saturated disk-IO, it was brought back to a single |
24 |
> thread. People were complaining about unusable systems. |
25 |
> |
26 |
> In any case, can you share your performance results? What speedup did |
27 |
> you see, on warm and hot FS caches? Which type of disk do you use? |
28 |
> |
29 |
> You could compare against qmanifest, which uses OpenMP-based |
30 |
> paralllelism while verifying the tree. On SSDs this does help. |
31 |
> |
32 |
> Thanks, |
33 |
> Fabian |
34 |
|
35 |
I am talking only based on my own experience but, at least on my case, there are |
36 |
noticeable differences between I/O schedulers. That is something that maybe |
37 |
could affect that tests... with latest kernels we only have noop, deadline |
38 |
(default) and bfq... in my case bfq works far better for keeping the system |
39 |
responsible (on both, rotational and SSDs). But you can test different |
40 |
combinations to see the one that fits better for you. |