Re: 64Bit installation on a 500 GB Sata DRIVE
Johannes Wiedersich wrote:
> Frank Lin PIAT wrote:
>> But unless you have a specific need to use stick to 32bit kernel
>> (driver, old application [in]compatibility), it is wise to move to
>> 64bits.
>
> Why?
>
> I know that 64bit is the future. I have the feeling, however, that many
> applications are still better supported on 32bit than on 64bit. This
> applies especially for multimedia, web pages and viewing certain
> restricted pdfs, but I have often regretted the decision to move one of
> my workstations to 64bit, just because it adds more hassle than it gains
> in performance.
>
> Of course this is just my humble opinion. It may work for others.
>
> So, again: What is the advantage of running 64bit that warrants the
> extra hassle?
>
> Cheers,
> Johannes
>
>
Hi,
I have all the systems I use/admin in 64bits (amd64), it's mainly Debian
and Ubuntu, I don't feel any "hassle" anymore. It used to be a problem
for a few web pluggins and a few semi-deprecated programs, but not
anymore. Most systems are multimedia workstations, everything works very
well, including flash (native 64bits), Java (native 64bits), all
possible codecs (w64codecs and more), Mplayer, Xine, Blender...
The only downsides are Cinelerra, well it doesn't work better on 64bits
than on 32bits, just keeps crashing (dropped it in favor of Blender,
which as a simple but powerful/stable video editing mode) , and
Cinepaint which is not easy to find, but Gimp with GEGL as been catching
up on features so no longer a problem.
Regarding performances, it's sometimes just the same, often much faster,
never slower, so no reason to look back to i386 for me. There's many
comparison floating around, especially on www.phoronix.com, you can have
a look at this Ubuntu vs Fedora roundup
http://www.phoronix.com/scan.php?page=article&item=ubuntu_810_vs_fedora_10&num=4
Skip the first tests which are about games (tells smore about the video
driver performances), both distributions have been tested in x86 and
x86_64 in the exact same conditions.
I imagine the performance gain should be even more sensible with some
databases or intensive scientific number-crushing apps.
Tom
Reply to: