1 |
On Thu, 22 Dec 2011 12:11:32 +0100 |
2 |
Francesco Riosa <vivo75@×××××.com> wrote: |
3 |
|
4 |
> > I tried messing around with this with Amazon EC2. The problem was |
5 |
> > that due to latency I only really saw the benefit for VERY high |
6 |
> > levels of parallelization (think -j25+).. However, make isn't |
7 |
> > actually "distcc-aware" so it just runs 25 jobs of anything in |
8 |
> > parallel. So, anytime a makefile launched a ton of java or python |
9 |
> > jobs the host ground to a halt as it wasn't distributed and it was |
10 |
> > way more than the host could handle (especially java - which |
11 |
> > swapped like there was no tomorrow). |
12 |
> |
13 |
> Just wanted to point out that (if there is enough memory) recent |
14 |
> kernels manage much better parallelism, even excess of it, once |
15 |
> reached the maximum load augmenting threads only bring minimal loss of |
16 |
> "real" time. |
17 |
|
18 |
Does that include handling complete lack of memory and heavy swapping? |
19 |
|
20 |
-- |
21 |
Best regards, |
22 |
Michał Górny |