Slow if share IDE port between HDD and CD-ROM ?

[Cl.Massé]

Are the following true for my PC (2 years old with 2 IDE ports and no
SATA) and it runs WinXP:


"...make sure that your CD-ROM drive isn't on the same IDE channel as
your hard drive. Sharing of IDE channels can dramatically slow down
CD-ROM and hard disk access."

http://www.tweak3d.net/tweak/cdrom/



"Make sure your hard drive is not connected to the same IDE port as
your CD/DVD-ROM. Each IDE port is programmed to operate at the slower
of the two devices on the port, so you could be slowing down access
to your primary hard drive by leaving a CD-ROM on the same channel.
Put your CD/DVD-ROM on the Secondary IDE port."

from: http://www.techbargains.com/hottips/hottip12/index.cfm

I confirm that all that is ridiculous. What is true though, is that
transfer between a HD and a CD drive is slowed down if they are on the same
IDE cable, and that the master should be set to the faster drive.

 

[Rod Speed]

I confirm that all that is ridiculous. What is true though,
is that transfer between a HD and a CD drive is slowed
down if they are on the same IDE cable,

Nope. Basically because very little software even
trys to use them simultaneously when copying.

and that the master should be set to the faster drive.

That is just plain wrong too.

 

[David Maynard]

So from what you say, if I have an XP system partition on one hard
drive then the swap file (assuming I want to fiddle with placing my
swap file) should not only be on another hard drive but that other
hard drive should be on another IDE channel. Is this correct?

In theory it would be better but the majority of the speed increase from
putting it on a second drive is reduced latency from head movement.

Say the swap file is at the front of the first drive and you're loading a
program located in the middle of the first drive (as in a one drive
system). The heads are slamming back and forth between the loading program
and swap file; and head movement is agonizingly slow compared to read/write.

Even if the drive is on the same channel, if the swap is on a second drive
then it's head can stay in the swap area while the other drive's head stays
in the program area and the system doesn't have to wait for them to run
their little butts back and forth across the drive.

In normal use you would probably never notice the difference between the
same or different channels, however, because the read/write overlap, or
lack thereof, is such a small portion relative to everything else going on.
(A time critical application, or large transfers, might though).

RAID 0, the question you asked, is a different matter as the whole point to
it is simultaneous access to make the array look like a faster 'single'
drive. I.E. a simultaneous read from two drives is theoretically twice as
fast as two sequential reads. You get 'twice' the data in the same time as
'one read' (to both drives).

On the other hand, if it has to read/write them sequentially, being on the
same channel, then it's no faster than one physical drive, since sequential
access is sequential access, and there's no point to it. The only thing
you'd get out of it is the worst of everything: sequential access like a
'normal' drive, no 'second drive' to put the swap file on because it's in
the RAID array, and lower reliability due to the added feature that if one
dies you loose everything on both.

----

I also use a PCI card to give me an extra two IDE channels. The card
is based on the Silicon Image 0680 Ultra-133 chip.
(I don't use its RAID capability. http://tinyurl.com/a685d)

In terms of performance does it make a difference if a 7200rpm hard
drive is on one of the two IDE channels on the motherboard (whose VIA
Via 266A/8235 chipset provides ATA133) rather on one of the IDE
channels on my adaptor card?

Just off hand I can't think of any reason why it would make a difference
one way or the other.

 

[Cl.Massé]

Nope. Basically because very little software even
trys to use them simultaneously when copying.

I disagree. It's precisely the case when one rip a CD or burn a CD. In the
latter case, the burning could fail, there isn't enough RAM to cache 650 MB.

That is just plain wrong too.

Perhaps expressed like that. The most recent and sophisticated drive. It's
because the master takes charge of the communication with the controller and
with the other drive.

 

[Rod Speed]

I disagree.

You're wrong anyway.

It's precisely the case when one rip a CD or burn a CD.

Nope, those dont use both drives simultaneously and the speed of the
operation is entirely determined by the speed of the cdrom drive anyway.

In the latter case, the burning could fail,
Nope.

there isn't enough RAM to cache 650 MB.

There doesnt need to be. There's a buffer in the cdrom drive for a reason.

Perhaps expressed like that.

No perhaps about it.

The most recent and sophisticated drive.

Still just plain wrong.

It's because the master takes charge of the communication
with the controller and with the other drive.

Completely wrong again.

 

[Kadaitcha Man]

Rod Speed, <[email protected]>, the irrelevant, bewigged hempseed, and
servant who performs all the menial tasks, jabbered:

Nope, those dont use both drives simultaneously and the speed of the
operation is entirely determined by the speed of the cdrom drive
anyway.

So, Wodleypoo, you techno-****tard ****, I can put an old 2x cdrw into a
52x32x52 burner and burn at 32x, yes? Hmmm?

--

 

[Cl.Massé]

You're wrong anyway.

Anyway you are wrong to speak me that way, I'm not your dog.

On protonic, there is a huge bunch of *real* technician, and it is entirely
free. I tried it as a client and as a tech, and it is incredible. There is
even a quality assurance, and the techs are *polite*.

 

[Rod Speed]

Anyway you are wrong to speak me that way,
Nope.

I'm not your dog.

I wouldnt have for a dog.

On protonic, there is a huge bunch of *real* technician, and it is
entirely free. I tried it as a client and as a tech, and it is incredible.
There is even a quality assurance, and the techs are *polite*.

Dont let the door hit you on the arse on the way out.

 

[Folkert Rienstra]

Surely that's slow for copying from one HD to the other?

Nope, not on copying, i.e. one reading, the other writing.

 

[Folkert Rienstra]

Just to clarify. Would 2 drive, RAID 0 performance be the same if the two
drives were master/slave vs one on each IDE channel?

Depends on the access pattern. Long sequential access makes the drives read-ahead
and the fact that drives can't be accessed simultaniously is offset by delivering
the data from the cache. So it doesn't work so well for random access but then that
is already quite slow.

 

[Folkert Rienstra]

I confirm that all that is ridiculous.

What is true though, is that transfer between a HD and a CD drive is slowed
down if they are on the same IDE cable,
Nope.

and that the master should be set to the faster drive.

Nonsense.

 

[Folkert Rienstra]

I disagree. It's precisely the case when one rip a CD or burn a CD. In the
latter case, the burning could fail, there isn't enough RAM to cache 650 MB.



Perhaps expressed like that. The most recent and sophisticated drive.

It's because the master takes charge of the communication with the controller
and with the other drive.

Nope. Only on POST with the selftest.

 

[Cl.Massé]

It's because the master takes charge of the communication with the

Nope. Only on POST with the selftest.

Sorry, but "master" need a thing, POST has nothing to do with that. You
could at least support what you say. So, what's the difference between a
drive working as a slave and one working as a master?

 

[Rod Speed]

Doesnt happen except at drive setup time.

More strictly when the drives are enumerated.

Sorry, but "master" need a thing, POST has nothing to do with that.

Want to try that again in english this time ?

You could at least support what you say.

How odd that you didnt yourself.

So, what's the difference between a drive working
as a slave and one working as a master?

Nothing as far as normal drive ops are concerned.

 

[Folkert Rienstra]

Sorry, but "master" need a thing,

That's far too technical for me.

POST has nothing to do with that.

Yup. Read the ATA spec.

You could at least support what you say.

Why? You sure don't.

So, what's the difference between a drive working
as a slave and one working as a master?

None other than that one responds to ID device0
and the other to ID device1.

Master and slave was abandoned right after the first ATA spec.
That's 6 intermediate specs ago already.
Haven't they told you that in your very polite and very competent
http://www.protonic.com/ forum?

 

[Cl.Massé]

So, what's the difference between a drive working

None other than that one responds to ID device0
and the other to ID device1.

Well, elusive answers, I suspected it.

By the way, ID0 and ID1 correspond to the position on the cable, ID0 is the
master only if the first drive is set to master or if both are set to cable
select.

Sorry if your business run slowly because of benevolent people, er...
well... yuppie!! If you had employed more people, you would have less idle
people competing with you. The Batavian redundancy rate is shameful.