| 1 |
On 18/08/14 15:31, J. Roeleveld wrote: |
| 2 |
> <snip> |
| 3 |
valid points, and interesting to see the corrections of my |
| 4 |
understanding, always welcome :) |
| 5 |
> Looks nice, but is not going to help with performance if the application is |
| 6 |
> not designed for distributed processing. |
| 7 |
> |
| 8 |
> -- |
| 9 |
> Joost |
| 10 |
> |
| 11 |
this is the key point i would raise about clusters really -- it would be |
| 12 |
nice to not need for example distcc configured and just have portage run |
| 13 |
across all connected nodes without any further work, or to use a tablet |
| 14 |
computer which is "borrowing" cycles from a GFX card across the network |
| 15 |
without having to configure nvidia grid: specifically these two use |
| 16 |
cases have wildly different characteristics and are a great example of |
| 17 |
why clustering has to be designed first to fit the application and |
| 18 |
viceversa. |
| 19 |
|
| 20 |
/me continues to wonder if 10GigE is fast enough to page fault across |
| 21 |
the network ... ;) |
Archives