Best performance setup?

[Intimidator]

What is the most efficient and best performance setup for the following IDE channels running W2Kpro:

Componets: WD40, WD20GB HD's, Toshiba 32x CD, Cendyne (LTR 52246S) burner.

My thoughts would be:

IDE 1 Master 40GB
Slave 20GB

IDE 2 Master Toshiba 32x CD
Slave LTR 52246S

Any thoughts? Is it better to run cable select vs the master/slave settings?

Is it better to have the 20GB as the master and 40GB as the slave? The second drive is just for backup purposes using Norton Ghost 2003.

Thanks for your time!

[digitalpimp]

Umm I would keep the harddrives on IDE1

But on IDE2, I would have the CD Burner as Master and the CDROM as Slave

also, try to always use the jumpers to set master or slave designation over cable select

also, make sure DMA is selected for the HD and CD drives in the Device Manager

Thanks,

digi

[CowboySlim]

WD40 as PM
Toshiba as PS
WD20 as SM
LiteOn as SS

Background: IDE channels can only transfer one way at a time. Therefore, if you are backing up from one HDD to another on the same channel, while the 40 is transmitting, the 20 is idle; while the 20 is receiving the 40 is idle. As a result, they both go off-on continously during the backup.

With the alignment as above:
HDD to HDD transfers are continous
Burns from the 40 to the LiteOn are continous
Reads to the 40 from the LiteOn are continous
Direct, on-the-fly, copies (if you want to, but not the best practice) from the Toshiba to the LiteOn are continous

Outside of that, I agree with everything in the above post.

[tazdevl]

If you're going to do a lot of disc to disc copying, I'd recommend this, otherwise what slim said is fine.


WD40 as PM
Toshiba as PS
LiteOn as SM
WD20 as SS

[Intimidator]

tazdevl & cowboyslim,

Thanks for the prompt response. I would have never thought of setting it up this way. I was under the impression that HD's should stay on the same IDE channel because a CD-Rom is not as fast as the HD and this will decrease the HD performance. Will this cause a decrease in HD performance this way? I was not aware of the continuous vs noncontinuous transfer. Both of these ideas are very unique.

An occassion I will do disc to disc copy but mostly I will be backing one HD to the other HD using NG 2003. I will also be burning files directly from the 40GB HD to the burner.

Thanks again!

[digitalpimp]

Intimidator

No, having a harddrive and a cdrom drive on the same IDE channel no longer slows both down to the slowest one of the two. In the past (a long, long time agao) this was the case, however, it is no longer the case.

digi

[CignaXI]

Intimidator

No, having a harddrive and a cdrom drive on the same IDE channel no longer slows both down to the slowest one of the two. In the past (a long, long time agao) this was the case, however, it is no longer the case.

digi

So it doesn't matter if one drive is in PIO mode while the other is in ultra dma mode? Why?

Btw I have a WD120GB SE, WD20GB, LTR-163D DVD and a LTR-16101B Burner.

[CowboySlim]

I'll skip the PIO issue out of lack of knowledge. Current drives such as the WDC HDDs should be run in the DMA 5 mode and the opticals (LiteOn & Toshiba) in the DMA 2 mode.

[Intimidator]

What is the difference b/t:

LiteOn as SM
WD20 as SS

and

WD20 as SM
LiteOn as SS

How will this affect performance? It sounds like the same to me?

Thanks!

[CDRecorder]

I would suggest this setup:

Primary Master: 40GB Hard Drive
Primary Slave: 20GB Hard Drive
Secondary Master: Lite-On CD-RW drive
Secondary Slave: Toshiba CD-ROM drive

[CowboySlim]

How will this affect performance? It sounds like the same to me?

Yes, nowadays, it is the same. One or two hardware generations ago, however, it used to be best to make the faster drive (the HDD) the master and the optical or tape drive (which are slower) the slave. The theory being that the master controlled the channel and controlled it at its highest native speed.

I think that CDRecorder (fka LiteOnGuy) is missing the requirement to backup from one HDD to the other HDD. When they are both on the same IDE channel, you are choking yourself because IDE does not allow for two way, simultaneous data flow. Think of it this way: The back up starts by the drive being backed up sending data to RAM (while the backup drive is in idle) until the RAM is full; then the data goes from the RAM to the backup drive (with the drive being backedup idling) until the RAM is flushed; then the process repeats until the backup is done. If the drives are on opposing IDE channels as primary and secondary, the process is essentially continous.

[aviationwiz]

Here's my setup:

PM: CD-RW
PS: DVD+R/RW
SM: DVD-ROM
SS: Unused

RAID SATA: HDD0
RAID: HDD1

[CowboySlim]

Your killin' me, AW, that RAID setup, as wonderful as it is, has nothing to do with the subject matter.
OTH, I don't have RAID, but I ought to get it just to learn about it.

Well, anyway, I learned about the other in the meantime.

So it doesn't matter if one drive is in PIO mode while the other is in ultra dma mode? Why?

It does matter. An IDE channel can't be run in mixed mode. That is, if an older (PIO mode) drive that doesn't support DMA, shares a channel with a newer drive that does support DMA, the newer drive must be set to run in PIO mode.

BTW, AW, I still owe you that anecdote that I picked up from that British manufacturer of pilot seats. Follow me to Beer Garden.

[CDRecorder]

How will this affect performance? It sounds like the same to me?

I think that CDRecorder (fka LiteOnGuy) is missing the requirement to backup from one HDD to the other HDD. When they are both on the same IDE channel, you are choking yourself because IDE does not allow for two way, simultaneous data flow. Think of it this way: The back up starts by the drive being backed up sending data to RAM (while the backup drive is in idle) until the RAM is full; then the data goes from the RAM to the backup drive (with the drive being backedup idling) until the RAM is flushed; then the process repeats until the backup is done. If the drives are on opposing IDE channels as primary and secondary, the process is essentially continous.

You're exactly right. I did miss that point. With that in mind, I would suggest this setup:

Primary Master: boot hard drive
Primary Slave: CD-ROM drive
Secondary Master: backup hard drive
Secondary Slave: CD-RW drive

[Intimidator]

WD40 as PM
Toshiba as PS
WD20 as SM
LiteOn as SS

Question?? How can you setup this pattern using the supplied IDE cables that came with the HD and MB?

The CD and burner are mounted in the top of the tower just like most towers and the HD's mount in the cage UNDER the 3.5" floppy bays.

So it is like this:

4 5.25" bays then under that 2 3.5" floopy bays and under that 5 3.5" HD bays. The cables cannot be setup in this fashion to connect this sequence above? How would I do this? There is not enough distance b/t the 1 and 2 connectors on the IDE1 or IDE2 ribbons to make this possible.

Any thoughts?

[rdgrimes]

IDE does not allow for two way

Guys, copying data from one drive to the other is not 2 operations, it's one. Having HD's on one channel, and opticals on one channel is a perfectly good set-up most of the time.
Given the complexities of varying motherboards, controllers, drivers and drives, the only way to know what works best on a given system is to test it. Unless you need to access the 2 drives in question, simultaneously from 2 different processes, they will work fine on the same channel. An example would be reading from one drive in one program while writing to the 2nd drive from another program. That specific instance would require them to be on separate channels for the process to complete quickly.
Copying a CD, or data from a HD, is one process.

[minix]

hi again rdgrimes

As I've just said in another forum, IDE bus doesn't have anything similar to "disconnect" function in SCSI bus.

When one drive makes an operation, the IDE bus is locked until the operation is finished.
This means that if a "read operation" is performed, a huge amount of "cycles" will be wasted doing nothing between the "read command" is put in the bus and the desired data is actually delivered.
This can be even several seconds if the drive wasn't spinning, and anyway it's a lot of time with mechanical drives.

SCSI bus can be used by the other drives during that useless time between the command is sent and data is available.
I don't why I wrote this. I forgot that it's here:
http://www.feurio.com/English/faq/faq_v ... nect.shtml

Copying a disc is 2 operations, of course. A drive reads, the other writes.

I've heard that something similar is/was developed for IDE bus, but I don't know if there are avilable controllers or old drives work perfectly with it... more info appreciated.

the only way to know what works best on a given system is to test it

This is the important point.
We can't tell you what is best in your system. You have to test it yourself. Sometimes performance is better with some drives and worse with others in the same IDE controller.

If everything performs well (reading and writing buffers show the performance level), don't change it.

[rdgrimes]

When one drive makes an operation, the IDE bus is locked until the operation is finished

This is the best argument against having your system HD share a channel with an optical drive. Every time you access the optical drive, your HD will be frozen for that time. Simply inserting a CD can bring your system to a halt. Most times it's not a big deal, but if you're a heavy multitasker like me, it can be a real pain. It's also the reason why I no longer put any HD's on the main IDE controllers. (That and the fact that I have too many burners)

[CowboySlim]

Question?? How can you setup this pattern using the supplied IDE cables that came with the HD and MB?

This is a separate question from the previous which had to do with functionality. OK, the short answer is that you can't do that with "Spec" UATA/100 cables. Spec compliant cables are no longer than 18" end to end, have 5 - 7" spacing between the drive connectors, are flat and have 80 wires. Anybody that tries to tell you anything else is misinformed or purposely misleading you. That includes purveyors of the round, 36" cables. If you doubt this, look up the IEEE standard yourself. That these non-spec, 36", round cables work acceptably for many people is a separate issue.

Now for the answer to the question of physical connection. I have used 2 HDDs and 2 optical drives in the functional relationship noted above; however, not with spec cables. I've used flat, 80 wire, 24" long, with 8" between drive connectors for this functional arrangement in an Antec SX830 case, which is physically similar to that described above. These cables, at 24" long, are not compliant with the UATA/100 standard. However, they work fine with the HDDs noted above as those drives transfer data no faster than the ATA/66 standard.

[CowboySlim]

IDE does not allow for two way

Guys, copying data from one drive to the other is not 2 operations, it's one. Having HD's on one channel, and opticals on one channel is a perfectly good set-up most of the time.
Given the complexities of varying motherboards, controllers, drivers and drives, the only way to know what works best on a given system is to test it. Unless you need to access the 2 drives in question, simultaneously from 2 different processes, they will work fine on the same channel. An example would be reading from one drive in one program while writing to the 2nd drive from another program. That specific instance would require them to be on separate channels for the process to complete quickly.
Copying a CD, or data from a HD, is one process.

RD,

Please excuse me, perhaps my writing style is somewhat lacking in clarity. Allow me to refer you to the last bullet on page 389 of this source.
Actually this is the second edition of a series of which the third has been printed, but not quite yet available. As such, I can't suggest that you purchase the 2nd edition. Therefore, I will quote from it: "ATA forbids simultaneous I/O on an interface, which means that only one device can be active at a time. If one device is reading or writing, the other device cannot read or write until the other device yields the channel. The implication of this is that if you have two devices that need to perform simultaneous I/O, e.g., a CD writer that you use to duplicate CDs from a CD-ROM drive, you should place those two devices on separate interfaces."

Now, these are my words: With respect to Intimidator's desire to use one hard disk as a back up to another, it would take approximately twice as much time if they were both on the primary channel as compared to being on separate channels.

BTW, I suggest that you, or anybody else who has an interest learning more, go here to preorder the 3rd edition (I certainly will). Hopefully, it will be available soon.

[rdgrimes]

Cowboy, your quotes are very interesting, but do not relate to actual experience. The fact is that you can copy a CD on the fly with both drives on the same channel, I do it all the time. It's one operation, not 2. I do this all the time, at speeds up to 48x.
The same is true for hard drives, copying is one operation, not 2. I've actually tested this, it's posted over at CDF somewhere. (Thread title is "IDE configuration myths"). The transfer speeds on the same channel were identical, even a little faster at times, than when the drives were on separate channels.
I can assure you that it does not take twice as long with the drives on the same channel. I suggest you test this for yourself if you don't believe it.
No doubt there are older controllers that may limit this, but I've been doing it for several years on a number of different boards.

[Inertia]

I agree with rdgrimes on this one. Theoretical considerations aside, the fact remains that with actual testing on modern computers it usually works perfectly well to copy from and to two devices on the same IDE channel.

See his thread at IDE configuration myths.

As rdgrimes has mentioned, the only way to know for sure is to test it on your own system, but personally this works fine on both of my computers.

The Web (and books as well) are full of bad, old, or misleading information that may not updated to current conditions. This is particularly so when advice is being given based on second hand theory rather than empirical conditions.

[rdgrimes]

I agree with rdgrimes on this one

BTW, the CDF links are currently broken, you have to substitute "forum" for "club".

[CowboySlim]

@Inertia and rdgrimes

Well, Buckaroos, as my ol' Pappy used to say: "You can lead a horse to water, but you..............................."

So gather 'round the campfire and maybe we can separate some mythology and anecdotal observations from good ol' fashioned book larnin'.

It has to do with the open systems architecture as defined in the IEEE Specs/Standards. Once these have been voted on and become official, everybody has to design hardware and software that is in compliance with them or their designs just won't work. It's just that simple - that's all there is to it. It doesn't matter if the motherboard with the IDE channels/connectors and the ATA/ATAPI devices are 5 years old or new this year. The standards specify one way data transfer only. (Now, this is different from SCSI, which does have two way data transfer.) What else can I say?

Well, Buckaroos, still not good enough? So I just ran an actual comparison test. Transferring 936MB of photos from a freshly defragged partition on my primary IDE channel HDD to an empty partition on a secondary IDE channel hard drive. Then transferring from the latter partition to another empty partition on the secondary channel.

............and, the envelope please! 1min 45sec interchannel; 3min 00sec intrachannel. I accept the fact that my shoot from the hip 2 to 1 estimate is somewhat overstated.

Ready for that drink yet, Buckaroos?

I take no offense that disagreement was expressed with my postings in this issue. However, I am gravely dissapointed in the attitude expressed concerning the veracity of Robert Bruce Thompson's book (my referenced source). That disdainful attitude does not convey an overwhelming sense of intellectualism.

About the book? I bought the original edition and was impressed enought that I bought the second edition last summer. The third edition will be available soon and I'll order that one too, shortly. Why, one might ask? As you may have noticed, I don't know everything and I've got to deal with that issue next.

[cfitz]

Slim, I'm sorry but your test was not valid if I understand it correctly. It sounds like your interchannel test went from drive A to drive B, while your intrachannel test went from drive B to drive B. The first issue is whether or not drives A and B are identical. If not, then the test is invalid because performance differences between drives A and B could account for some of the disparity.

More importantly, when copying from drive B to drive B you introduced seek times into the equation. The transfer from A to B, if A is defragmented and B is empty as you described, was a nice smooth transfer. Drive A's head seeked to start of the files, drive B's head seeked to the start of the empy space, and then both drives smoothly scanned the disc surfaces without having to do any more seeks. This is a sequential transfer, and it is the fastest way drives can transfer data.

The copy from B to B, on the other hand, is a random access transfer. Drive B's head has to seek to the start of the files, read some data, then seek to the empy space, write some, then seek back to the files, etc. All this seeking really slows down the transfer.

cfitz