Settle a debate or fuel the fire about IDE configurations.

[Buz2b]

A couple of friends and I have been debating the setup on a system as far as the "ideal" solution for a non-RAID IDE setup. The theoretical system would be a system (WinXP) with 1 HDD, 1 CDROM and 1CDRW. Some say that the system is better off with the CDROM as the secondary master and the CDRW as the slave; or visa versa. Some point out that DAE would not be possible with this setup and that the CDROM should be the Primary Slave with the CDRW as the Secondary Master. Then the debate starts about putting a slower device (the CDRW or CDROM) on the same IDE as an UDMA 100 HDD and how that "could" slow down the system. Remember, this is non-RAID and/or using NO IDE controller card.
I'd be interested in hearing what others think of this. Thanks for your input.

 

[Peter]

Modern IDE bridge chips allow mixing and matching different speed devices on one channel, without
degrading performance of the faster drive.

However, on IDE no matter who made it, the two devices cannot share bandwidth,
so as long as one is being accessed, the other one is blocked out.

Then, remember the actual IDE _controller_ is in the Master drive, so you want the newer (or faster)
drive to be Master and the older (slower) one to be Slave.

Conclusion: You want the CDRW to be on a different channel than its source drive(s), and the fast
HDD at Master with the CDROM as slave.

regards, Peter

 

[Buz2b]

<< CDRW to be on a different channel than its source drive(s) >>

OK, while that is fine if using DAE or copy on the fly from CDROM, what about copying from the HDD or making a HDD image file?

 

[Gog]

<< The theoretical system would be a system (WinXP) with 1 HDD, 1 CDROM and 1CDRW. Some say that the system is better off with the CDROM as the secondary master and the CDRW as the slave; or visa versa. Some point out that DAE would not be possible with this setup >>



Why wouldn't DAE be possible with this setup? The harddrive and the CDROM that is doing the extraction are on different channels and therefore there shouldn't be any intereference.

 

[dunkster]

Peter's post was one of the best summaries of IDE device configuration that I've seen posted.

Buz2b's response just points out that no matter how you configure 3 devices on 2 IDE ports, some compromise in overall connection flexibility/efficiency will suffer from master/slave configuration.

A dedicated IDE port for each IDE device is obviously ideal for read/write to/from any device to any other device.

It's kind of amazing that folks are still buying mobos with only 2 IDE ports, considering that most have a minimum of 3 IDE devices either installed or planned for expansion.

Hope this helps!

 

[Peter]

When you're making a HDD image file from CDROM on the same channel, interruptions don't matter.
HDDs are obviously capable of accessing their sectors in any order and at any time as the software pleases.

From HDD to CDRW you're on separate channels again.

Real men use SCSI. That gives you just one headache (when you see the bill) and then no more

regards, Peter

 

[Buz2b]

<< Real men use SCSI. >>

HA! That is one area I have never "dived" into Peter. BTW, will "Serial ATA" that I keep hearing of make that argument no longer valid because of it's speed? I know it won't be "chainable" like SCSI. Just curious.

<< It's kind of amazing that folks are still buying mobos with only 2 IDE ports >>

Yeah and it's amazing that everyone doesn't have DSL either (at least I think so)! Unfortunately not everyone I deal with are even aware of RAID or IDE controller cards in general when they buy their MB's.

So is it the "concensus" that the best compromise would be Primary Master=HDD, Primary Slave=CDROM; Secondary Master=CDRW? And if a second HDD is added then it would be put as Primary Slave with the CDROM moving to Secondary Slave (or Master?)? The argument started over what would be the better setup (overall speed) for someone that burns Audio and Backup CD's (semi) regularly.

 

[Peter]

DSL ... that'll change. SiS are working on a low cost single chip DSL solution to go onto ACR riser cards.

Serial ATA will be the same old crap on nicer cables. Still no bandwidth sharing, and still only two drives per channel. Remember, what we know as IDE today still is the same arrangement that we had in the very beginning of 286 AT computing: Rather un-intelligent, sector handling controller on ISA bus, steering one or two drives. It's just the controller has been moved into the Master drive, with a supercharged version of ISA running on the "IDE" cable, and an added DMA data pump in the PCI-to-IDE bridge to at least get off the CPU's butt while transferring data.

Back to the topic ...

Two HDD, CDRW, CDROM? Both HDDs as masters, CDRW and CDROM as slaves. Hold your CD images on the HDD that is not on the same channel as the CDRW.

regards, Peter