1 |
> Something _really_ weird happened to your quoting; you quoted my |
2 |
> email, but your email client said you wrote it. |
3 |
> |
4 |
> On Mon, Dec 17, 2012 at 2:00 AM, J. Roeleveld <joost@××××××××.org> wrote: |
5 |
>>> On Sun, Dec 16, 2012 at 10:39 AM, J. Roeleveld <joost@××××××××.org> |
6 |
>>> wrote: |
7 |
> |
8 |
> ^-- weird --^ |
9 |
|
10 |
Very weird, especially as I am using the same client now (squirrelmail) |
11 |
and I don't see that line at all now. |
12 |
|
13 |
>>>> On Sunday, December 16, 2012 01:52:46 PM Volker Armin Hemmann wrote: |
14 |
>>>>> Am Samstag, 15. Dezember 2012, 20:57:24 schrieb J. Roeleveld: |
15 |
>>>>> > Even on a system with only 2 sockets, it can be useful to have NUMA |
16 |
>>>>> > available. |
17 |
>>>>> |
18 |
>>>>> or not, because it costs you performance. |
19 |
>>>> |
20 |
>>>> When does it cost performance? |
21 |
>>>> In all situations? |
22 |
>>> |
23 |
>>> It adds some additional logic to memory allocation (put an allocation |
24 |
>>> near the process that uses it) and to process scheduling (keep the |
25 |
>>> process near its memory, but bump it to a more distant idle core if |
26 |
>>> necessary). |
27 |
>> |
28 |
>> That's the way it's supposed to work, yes :) |
29 |
>> |
30 |
>>> In all honestly, it's not a performance loss you're likely to notice, |
31 |
>>> unless you're so in need of squeezing out every spare cycle that you |
32 |
>>> most definitely _have_ hardware where there are disconnected memory |
33 |
>>> banks. I'm not convinced it's even measurable for us mundanes and our |
34 |
>>> hardware. |
35 |
>> |
36 |
>> I don't think I would notice it either, but as the system I have |
37 |
>> supports |
38 |
>> it, I want to use it. |
39 |
>> And then I want to be certain it actually supports it correctly. |
40 |
>> |
41 |
>> The system I'm talking about is used for testing purposes. Running |
42 |
>> multiple VMs. As far as I know, Xen has support for it, just need to |
43 |
>> configure it properly. |
44 |
>> And for this usecase, I think NUMA with only 2 physical CPUs should make |
45 |
>> a |
46 |
>> positive difference. |
47 |
> |
48 |
> Don't get me wrong; I was arguing that it shouldn't hurt to have it |
49 |
> enabled. :) |
50 |
|
51 |
I know, just wanted to add the use-case for considering NUMA a usefull |
52 |
option even with only 2 physical CPUs :) |
53 |
|
54 |
-- |
55 |
Joost |