[KLUG Members] IDE cd burner

[Adam Tauno Williams]

>>EXACTLY!  And my point is that bang-for-buck one needs to consider
>>if spending the extra danero for a fast burner gets you anything.
>Yep.
>>It depends entirely upon the application,  if your going to be copying
>>CD's and can't put the drive on it's own channel then you might as
>>well buy a slower drive.
>Or add another channel.  Better yet, device-to-device is best with SCSI.
>TekRam's DC-315U UltraSCSI card is ~$20, so you only need to eat the
>drive premium of ~$25-50/drive.

And then you have a SCSI card for other purposes as well.  Very nice SCSI
scanners can be found at curb-side prices.  SCSI tape drives (DAT) are very cheap.

>>If your creating CD's from ISO images on a HD on a seperate channel,
>>and your machine is relatively robust,  then you get what your
>>paying for.
>Yep.  Although ever system I build from now on will use a $125
>2-channel/2-disc 3Ware ATA RAID controller in RAID-0 (for workstations) or
>RAID-1 (for servers).  That also frees up the on-chipset ATA/IDE controllers.
>>Can't you add a third IDE channel to a machine?
>Yep, with a fourth.  You can even add more, although some vendor's
>drivers are not good at multiple cards (especially in Windows), so you should
>alternate cards from different vendors.  Or you can just pick up a 3Ware card 
>and use it to drive your ATA/IDE disks.  Mega performance with an on-board, 
>32-bit microcontroller with integrated RAM (in addition to the actual IDE
>controller).
 
Has anyone used a USB burner with Linux?  What's the top end speed for those?

>>Are there any hdparm settings that can be used on CD-ROM drives?
>Yes, although most ATA/IDE CD-ROMs don't seem to support the full ATAPI
>spec.  
>I.e., when I run "hdparm -v /dev/hdX" it often fails to read a few
>values.
>><rant>System performance is more complicated than the speed ratings
>>of each individual component.
>No crap.  The Pentium IV is a perfect example, which is much slower MHz
>for MHz  than a Pentium III (let alone an Athlon!) in most functions.  And
>forget the "Pentium IV optimized" crap because it usually means using _lossy_
>float- point via SSE/SSE2 (which is really interpolated integer math) which
>affects precision (quite noticable with image compression -- like with MPEG-4
>encoding).

Any processor optimization (IMHO) is just noise.  Unless the application you USE
is compiled that way, it simply doesn't come into play.  And most people use
binary RPMS or other packages compiled for the i386,  or the i486 if they get
lucky.  Not that compiling a typical OSS package is hard,  but most people
aren't going to do it.

>>I've seen both individuals and companies waste significant
>>amounts of money on "upgrades" that accomplished nothing
>>because they didn't bother to examine the interrelations
>>of the various subsystems,  but viewed them as discrete
>>entities.
>Especially servers!  You need I/O in servers, not just dual processor. 

"I/O, I/O, its off to work we go..." :)

For even your average business app run on the desktop I think I/O is overlooked.
 Good I/O helps alot when you start to swap.  It is true that the MCSE's of the
world have brought the slavering-CPU-mania of PC land into the glass house.

>That's  why I buy servers with ServerWorks chipsets.  It is also why Intel
>cross- licensed their products.  I understand the i845 SDRAM chipset for
>Pentium IV is of their design.  Too bad Intel forcing mainboard manufacturers
>to "cripple" it as SDR-only (when it does do DDR).
>>I call this the "Best Buy" mentality,
>No joke.  BTW, did you hear that Best Buy bought out another retailer
>that was implementing Linux as their POS device?  Now they are going Embedded
>NT like  Best Buy.  Ever ask cashiers at Best Buy about the "reliability" of
>their system?  Every 3rd time I go, the backend is down and rebates won't
>print.  And when people comment about how "cool" the POS screen is, the 
>cashier always says, "yeah, when the system is working."

Best buy bought someone!  I thought you had to make money to do something like that.

>>Bob probably refers to it as "The Gateway Syndrome". :) </rant>
>Gateway 2000 royally screws up their Athlons.  They use the utmost

Doesn't suprise me.

>quality in Intel components, but the cheapest in Athlons.  They don't even use
>Athlon-rated power supplies!  That's the ultimate insult.
>The problem isn't that AMD doesn't "certify" hardware, its just that
>people don't bother to check.  Same thing with Linux, people assume that
>RedHat, Caldera and others don't "certify" their products with various
>databases, storage, backup, etc... but they do.  In fact, we recently had a
>discussion about one tech who had to argue with his boss that RedHat had a
>higher/better vendor certification than HP/9000-UX on his exact, proposed 
>database (Oracle), storage (Want to say "something-Stor?) and backup (cannot
>remember the library) components for their configuration, while 1/3rd the 
>price!  His boss still refused to go Linux.

No suprise there either,  HP-UX is a "safe" bet.  There are a suprising number
of UNIX stalwarts who still dismiss Linux as a toy.  I meet them all the time.

Systems and Network Administrator
Morrison Industries
1825 Monroe Ave NW.
Grand Rapids, MI. 49505