[linux-audio-dev] Re: [linux-audio-user] disaster day #1

Bob Sanders rsanders at conejo.engr.sgi.com
Tue Jun 7 15:32:56 UTC 2005


> 
> The deeper I get into the hardware world (day job) the scarier it is.  You
> know how some PSUs cost more than others?  Sometimes a lot more?  There
> are reasons for that.   Ditto RAM.
>

Power supplies that go into an OEM product tend to get a lot of engineering
resources thrown at them.  But the end user has neither the time, tools,
nor the experitse to deal with all that, so lots of cheap, junk power supplies
get used.  Plus, people generally only read the total watts, not the inidvidual
currents and additive wattage of +3.3V, +5V, and +12V.

Also, there are power supply vendors that use sub-spec wire and connectors.
The power supply can deliver to voltage and current, but the wiring harness is not
capable of delivering that to the motherboard and peripherials.  They are betting
that no one or a rare few will actually get near the max. requirements, so shave
pennies off each supply in a mis-guided attempt to reduce costs.
 
> I'm frankly amazed that we all get by without ECC on *every* machine.
> It's frightening.
>

Not really.  Most home users/workstations use 2 to 4 DIMMs at most.  And
even then, most of that is not active.  So a transient memory error rarely
has a chance to be noticed.

When you expect 60% to 80% memory utilization 24x7, and the number of DIMMs
increases, then the chances of hitting a memory error are more likely.  Some
systems in the field are running 4 TB of ram.  That's a lot of DIMMs.  The
first few months deployed are real nightmares.   It's not uncommon to see
200 DIMM failures.  But all those are not always real, they may be transistive.

And most error logging for memory errors is pretty poor, so finding a root
cause is diffcult.

Bob
-  



More information about the Linux-audio-dev mailing list