On 09/04/2015 05:58 PM, Ranjan Maitra wrote:
If I take the traditional computing scheme, such as compiling the Linux
kernel, much of the need is to also have fast storage in the code source
tree.
Depending on what compiler you use and what option you give it, the I/O
generated by object files (just an example) would kill your performances.
Good point. Perhaps, one way to get around this is to also move executables and libraries
generated by me to the "System" side of things, using perhaps a make install.
I used gcc with -O3 or -Ofast. Any suggestions on this?
Not being a specialist, I would say that the safer is the best option.
Optimisation flags comes with drawbacks on some codebase.
There was a time, a long time ago, when I compiled the kernel, -O2 and
-O3 lead me to different misbehaviour.
Being "old" now, I dont compile my kernel anymore ;-)
My suggestion would be to leave the system on the "slow disk", as you
boot one time, and put the source tree either in "tmpfs" or SSD.
Any suggestions of a specific PCI-e SSD. I would prefer 128 GB or something not
larger than that.
I know OVH really liked Intel brand, and they provide dedicated servers
where you can choose many OSes.
That's just what I know.
But I'm certain your feedback would be interesting.
At least, buy your in a store where you have choice *and* where you can
replace it if ever fails to run with your OS.
Could be more expensive, but...
--
users mailing list
users@lists.fedoraproject.org
To unsubscribe or change subscription options:
https://admin.fedoraproject.org/mailman/listinfo/users
Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines
Have a question? Ask away: http://ask.fedoraproject.org
