| 1 |
On Sunday 06 September 2009 00:48:40 Volker Armin Hemmann wrote: |
| 2 |
> > Also, have you considered that you got it all backwards? The kernel |
| 3 |
> > configuration tells you that for lower latencies, you should use 1000Hz |
| 4 |
> > and PREEMPT. It even says "Desktop" right there. Why should I take |
| 5 |
> > your word over that of the kernel devs who actually wrote that code? |
| 6 |
> |
| 7 |
> low latency means bad throughput and that hurts IO. |
| 8 |
|
| 9 |
The average desktop user on Linux needs low latency. To get that, one must |
| 10 |
sacrifice some throughput efficiency. |
| 11 |
|
| 12 |
The average server machine needs high throughput. To get that, one must |
| 13 |
sacrifice some latency efficiency. |
| 14 |
|
| 15 |
Latency and throughput cannot both be optimal as they conflict. One must pick |
| 16 |
the point of the line on the graph that best suits one's needs and move |
| 17 |
forward from there. I have no idea what hardware you run but your replies |
| 18 |
indicate a high possibility of one of these: |
| 19 |
|
| 20 |
1. server usage |
| 21 |
2. Massive bleeding edge desktop hardware that can give acceptable performance |
| 22 |
regardless of what you throw at it. |
| 23 |
|
| 24 |
have you considered that there is a large population of users whose needs and |
| 25 |
workload are totally different from yours and therefore require something |
| 26 |
completely different to you? |
| 27 |
|
| 28 |
-- |
| 29 |
alan dot mckinnon at gmail dot com |
Archives