[ale] Is this a good deal

[Jim Kinney]

Think of it this way: each process you want to run will consume at least one cpu core. The video conversion may consume more than one if it can parallelize frames.

The engineering aspects of this depends on known performance of the software doing the video process. If it doesn't have to be real-time (i.e. watching the video as it's being transcoded), less speed and more cores and RAM are a better buy. If it is a live process you'll need to sacrifice virtual machine performance with fewer cores at higher cpu speed and moderate ram with really good storage IO. CPU controlled ram is faster than chipset controlled but it takes cpu cycles to manage ram. I'm not sure if ram refresh is a cpu process or embedded in the ram now. The later would be great.

The AMD ryzan performance error is being fixed with replacement. Apparently the fixed cpu does a kernel compile in under 30 seconds now. :-)

On September 4, 2017 10:16:04 AM EDT, Narahari 'n' Savitha <savithari at gmail.com> wrote:
>Good email.  Thank You
>I am under constant confusion.  I was at the ALE  Sunday meets last
>weekend
>and before that.  I am trying to build a PC where I can record the
>SiliconDust output to the disk and stream that to my roku or appletv or
>chromecast. (whichever is easy)
>
>I have gotten great advise about this but what I am have understood so
>far
>is that a fast CPU is a must for transcoding from the recorded mpeg2 to
>.264 format.
>
>As Steve has pointed out if I get an i7 (dont know about Ryzen as
>reliable)
>and put like 32GB of RAM on it would I be able to do both transcoding
>and
>run say 2 or 3 VM's on them ? or am I stretching it too much.
>
>-N

-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.ale.org/pipermail/ale/attachments/20170904/389a4862/attachment.html>