Re: Anybody's machine working most of time?

Tiny Tim wrote:

"Michael J. Mahon" <mjmahon@xxxxxxx> wrote in message news:LqednTRsPYEMiPPfRVn-hA@xxxxxxxxxxxxxx

JW wrote:

Here are two ideas that may or may not help
1. Be sure you do not have a series scheduled to be recorded.
2 Be sure you do not have your system set to automatically do a defrag, virus scan, spyware check or any other operation at the same time each day. unless the time is such that you know you won't be using the system.

"Joe Horton" <joe@xxxxxxx> wrote in message news:eJm$E0aSFHA.2132@xxxxxxxxxxxxxxxxxxxxxxx


Does anyone's machine just work all the time? My Tivo never crashed, hung
up, or failed to record anything - and I'm always fighting with my new MCE.
It's super cool when it works - but it seems every day it's slowly getting a
bit worse. Last night while watching a DVD, the hard drive light went nuts
and the DVD started playing choppy and freezing. It checked everything I
could think of.

I have a new Gateway - should I have bought something else - or do most of
you have issues too? 

I think it's clear what the OP was saying--TIVO is designed and tested
to operate reliably and repeatably as long as the hardware works.

Unfortunately, this seems not to be the case with MCE.


That may be so. But to compare a bespoke system in which the designer has complete control over all aspects of hardware and software and in only one configuration amd which only performs one function with a system with myriad combinations of hardware and almost unlimited abilities to play games, process spreadsheets, databases, documents, graphic design etc etc. is not really comparing apples with apples. a Tivo is to a bicycle as an MCE PC is to a turbocharged car. One is inherently more complex and hence needs more effort in setup and maintenance. 

I acknowledge that point later in my post, and am fully aware
of the combinatorial difficulty of getting such "mix 'n' match"
systems to work reliably--or at all.

But consider the very restricted list of "approved" hardware and
software as well.  Also consider that the designers willingly chose
to approach the problem in this way.  Consider further that they
chose _not_ to offer a "turnkey" solution to OEMs.

They created this problem themselves, and will have to live with
the support and ill-will consequences in the marketplace.  The
marketplace (that's us ;-) must live with the waste of time and
money.

Nobody should pretend that an MCE PC (or any PC) is suited to people with no knowledge of how to use one properly. 

Sure, but the "average" PC user is clearly not a rocket scientist.
In fact, given the large number of them, it is safe to assume that
at least half of them are "below average".  ;-)

Now consider that MCE PCs are often sold as consumer electronics
items--and compared to them in showrooms.  See the problem?

Allowing potential buyer's expectations to be set in this way
is a recipe for a support nightmare.

If MCE were treated as a system in beta testing, and made
available at little or no cost to "testers" recruited for
the purpose, expectations would be met or exceeded.  (And
this would be closer to the truth.)

A system designed to record video should never need to be defragged,
since it can use allocation units of constant size for the video, and
allocate all video disk at one end of a partition, while everything
else is allocated at the other end (the system should be designed so
that frequent long seeks are also unnecessary because of buffering
of shorter files).


Proper knowledge would lead one to record TV to a separate partition or even a separate disk. That would remove the problem of excessive fragmentation. Even if recorded files are fragged there should not be a need to defrag them in order to use them. I certainly don't need to. I have a dedicated drive.

Unless the space allocation scheme used by the file system is
appropriate, no amount of partitioning will prevent increasing fragmentation with use. There is no evidence that NTFS or FAT32
are so designed.

TIVOs, ReplayTVs, and UltimateTVs (not to mention several Windows and
non-Windows-based sofware solutions) routinely record multiple programs
while playing back another, so certainly three simultaneous video
streams should be no problem.


True. I didn't know it was a problem for MCE. Is there evidence to suggest it is? 

The reply to the OP to which I replied had suggested that it was
(quoted above, point 1).

The problem is that the designers apparently did not think much
about how to level the load and how to minimize wear and tear on
the hardware--primary considerations for any long-lived design.


Which designers? MCE designers or system builders? I've built my own system. It works for me :-) Empirical evidence suggests many system builders are big screw-ups. Have you ever seen the crap that Packard Bell install on their machines? It's a real pig's ear.

I have friends who bought a brand new Packard Bell MCE PC recently with AMD 64 3400+, 1024MB RAM, 2*160GB HDD. It runs much much much slower than my homebuilt Shuttle with an Athlon 64 3000+. From a cold boot it is running something like 56 processes. My Shuttle has just 30, while watching and recording live TV. 

While the OEM does have great freedom (which the MCE designers
allowed them) to build an ill-configured system, no amount of
skillful component selection can compensate for a general-purpose
OS-based approach that is constantly spending time and resources
on things which have no relation to the application problem.

A proper OS-based process priority scheme will effectively prevent
_any_ lower-priority process from interfering with the progress of
a higher-priority process.  Or perhaps you don't think of media
processing as a real-time activity...

The problem that MCE addresses is not an unsolved problem--though
MCE attempts to support more combinations of hardware and software
(like Windows in general)--but apparently little research was done
into how others have solved the problem (unfortunately, also a common
Microsoft practice).

I'm afraid that the ultimate source of the problem of MCE reliability
is a level of complexity beyond the ability of its designers and
implementors to master.  This complexity is not a characteristic
of the problem, but of the means MCE's designers chose to solve it.


MCE is just an application laid on top of Windows XP. In principle it should really be no different to Snapstream, MythTV, Beyond TV, GBPVR etc. etc.. I must admit there do seem to be hurdles to overcome that are unecessary. Of course, as I built my own system I would expect to face more of those than someone buying a pre-built solution. Frankly it is a disgrace that customers of HP, PB, Dell, Gateway etc. face as many problems as they do. But does the fault lie with MCE, the system builders or the end users just being dumb f##s and stuffing all sorts of virus and spyware laden crap on their PCs and failing to run regular Windows updates and so on? I suspect most of the fault lies with the latter two. 

Of course, people who look at a "MCE PC" and think of it as both a
Media Center and a PC are mistaken--but you can hardly fault them
for getting this impression.  Neither Microsoft nor the OEMs say
much about restricting the application set until you run into
problems!

I would say that, in general, when you layer a demanding real-time
application on top of a general purpose OS, you deserve all the
trouble that you engender.

Notwithstanding that, Beyond TV seems to have made it work and has
met customer expectations better than MCE, albeit with somewhat
narrower (less grandiose?) aspirations.

MCE will eventually get where it would like customers to believe
that it already is, if it doesn't disappoint too many prospective
customers too often in the long road to getting there.

I very much regret the use of the mass market as a kind of
"extended beta testing" environment, where people who get
it to work correctly are "delighted" and people who can't
are "disgusted".  Marketing a "solution" to the masses calls
for a robust solution that cannot easily be broken from the
"front panel".  Microsoft and their OEMs have much to learn
about this.

To paraphrase Dijkstra, if you run out of neurons before you run out
of complexity, you're toast.


-michael

Home page:  http://members.aol.com/MJMahon/
.

Loading