[mythtv-users] Good Books/Websites on building computers

Michael T. Dean mtdean at thirdcontact.com
Mon Mar 26 16:32:34 UTC 2007

On 03/26/2007 11:41 AM, Jon Sustar wrote:
> Does anybody have any recommendations on good books or websites that 
> have indepth explanations of the steps to build a computer?

This is exactly what you're asking for:


>   I have a computer background, but I'd like to get a better 
> understanding of what hooks up to what, what needs to be compatible 
> with what, etc. 

But probably not at all what you want.  ;)  It's an excellent book with 
which you build a modern computer from scratch--building the hardware, 
the compilers, the kernel, and applications.  (Who needs Intel, AMD, 
NVIDIA, Via, SiS, WD, Seagate, Centon, Kingston, Linus, Richard 
Stallman, etc., anyway? ;)

Note: I haven't actually made it through the book, but it's on my list 
of things to do--somewhere around "run a marathon."

For what you want, though, my recommendation is to find a friend (or 
several) who's a true computer geek, loves talking about computers, and 
has tons of parts around the house, so you can get one-on-one 
instruction with hands-on experience (and current information straight 
off the 'net).  I really think you'll get much more.

Also, if you do go with books or websites, before reading through the 
info, think first about the reason for the existence of the source, the 
background/ties of the author, and the date of publication.  Doing so 
will help you see the bias so you can temper it with sources that 
present information with other biases.

    - Website with articles designed to inform, while presenting 
information that is beneficial to advertisers/hardware companies (that 
allocate pre-production samples for review sites to perform testing)
    - Book written by a technology professional who has worked for a 
specific company for many years
    - Information presented by someone who has only used Windows-based 
    - Information presented by an open-source zealot (I'm a user and a 
believer, but I wouldn't consider myself a zealot.)

You get the idea.


My bias:  "Hardware is significantly overpowered compared to software 
today, such that the only real consumer-type applications that stress 
modern hardware are modern gaming and /real-time/ high-resolution video 
encoding/decoding.  Therefore, rather than buying parts that could give 
a 2% performance improvement for 50+% price increase, hardware purchases 
should be made without taking into account performance (instead focusing 
on price, value, reliability/longevity, support, etc.--tempered with a 
risk/reward ratio that accounts for consistently-falling technology 
prices and consistently-improving technology performance), and the only 
time performance should be a factor is for those specific applications 
that stress the hardware.  Even then, performance gains that do not 
specifically support the tasks for which the system is being designed 
should not be a factor."

And, back to MythTV:
Note that the *only* part of a Myth system that falls into the category 
of "hardware-stressing application" under my definition, above, is the 
frontend.  Although the backend may do a significant amount of 
high-resolution encoding (when transcoding HDTV)/decoding (when 
commflagging and when transcoding HDTV), it does not need to do so in 
real-time.  (Even though software encoders need to do real-time 
encoding, they are only SDTV-compatible, so they do not do 
high-resolution encoding--where HDTV fits today's definition of 
high-resolution based on today's hardware.)  Neither of my 2 backends 
could do real-time HDTV decoding, but they do just fine with commercial 
flagging and transcoding.  I have an Athlon XP 2400+ and an Athlon XP 
2000+ for backends (with 4 total HDTV capture cards and no SDTV capture) 
and they generally do commflagging (often across the network) at about a 
2:1 time ratio (i.e. 2 hours to flag a 1-hour show).

More information about the mythtv-users mailing list