Yup...have a Plextor 12x burner slaved to a hard drive on the Secondary IDE with no problems at all!!!
Opus

Hey Guys,

I'm glad this thread was started, I was gonna save up all my questions and hit Opus with'em when all my parts get here (heads up Opus). After everything gets installed and scavenged parts get migrated into my ANUS (Lmao!!), I'll end up with 8 physical places to connect up IDE Drives. On-Board Channel 1&2 (+master/slave)=4 as well as a Promise PCI DMA/66 Controller Card. It has 2 channels as well (+master/slave) = 4 more. As for IDE Drives that I HAVE are:

1 - 56x CD-ROM
1 - CDR-Burner
1 - IBM 13G/HD @ 7200rpm
1 - Quantum Fireball 20G/HD @ 7200rpm

My thoughts are to put the HD's on the Controller card each as a master on it's own channel and connect and set the optical drives as masters on the 2 channels on the Mobo connectors. Basiclly set it up so each device is a master with a dedicated channel. Is this the right thing to do? Is there a more optimal set up using my existing equipment?

Thanks in advance.

CustomProd

BTW - Keep an eye out ya'll for a future post. I'll put up my laundry list of parts that I'll be using to replicate the Opus2000. ANUS=POWER! :)

I would never put anything on the same IDE channel with a hard drive (besides another hard drive). Especially a cd-rom/cd=r drive. It slows down the preformance of the said hard drive on that IDE channel. Just what "I" would do though, doesn't make it right.

??? My understanding is that the rule of thumb is to never put a harddrive on the same IDE channel as a CDROM, DVDROM, CDRW, or DVDRW, because both components on an IDE channel will only run as fast as the slower device. Therefore, in a "typical" small DAW machine, without a RAID motherboard, HDD1 (system) is the master on IDE1, at the end of the IDE cable, HDD2 (audio) is the slave on IDE1, on the middle connector of the IDE cable, CDROM or DVDROM is the master on IDE2, on the end of the IDE cable, and your CDRW is the slave on IDE2, on the middle connector of the IDE cable. If it's a RAID motherboard, disable the RAID and make each of the four devices masters on their own IDE channels (1, 2, 3, & 4).
Am I wrong??? If so, please explain, Opus.

I didn't read through all of your reply sonofsmawg, but what I did read was

"My understanding is that the rule of thumb is to never put a harddrive on the same IDE channel as a CDROM, DVDROM, CDRW, or DVDRW, because both components on an IDE channel will only run as fast as the slower device."

And that's totally correct. The devices on an IDE channel can only operate as fast as the weakest link in the chain. Therefore I would never put a CD(-rom or whatever) on the same IDE channel as a hard drive.

Costomprod, I think your setup is optimal.

As far as the other setups go, I think having a burner and a cdrom/dvd on the same IDE cable has the only disadvantage of not allowing to copy from the CD reader to the CD burner. The reader and burner need to be on different IDE cables. But most people don't have a reader as good as their burner, and are better off reading/ripping and burning from the same CDburner. In this case then, it doesn't really matter if both CD drives are on the same cable. I think. :p

Tommy P.

Ya know...of all the systems I have designed, tested, performed on etc etc...all of em have had IDE drives with CDROM's or CDR/W's on the same cable....never once did I notice a performance difference. I dont think the difference is going to be that much "greater"..I dont know the exact measurements but here's what it all boils down to....it's when you put two hard drives on the same cable that will affect performance even more! When you have two hard drives on one cable and they are both being used it sucks your bus speed to half it's worth....
I'll have to find that article I had somewhere that explained that in greater detail....also was always reccomended to me back in my early days of PC's from all the guru's I used to chat with on ICQ! Man, those were the days!
Opus

Just what I need ... more confusion ... :roll:

I know what you mean :D

It gets worse...
On my Roll Yer Own Thunderbird I have the 20 gig OS/Apps drive as Master and a 60 gig Audio drive as slave -then- I have the CDR/W as master and a 60 gig 2nd audio/backups as slave. Works fine.
lou

Decisions, decisions ... do I take the Zanex, the valium, or a few aspirin? Hmmm ... maybe I'd better just take a "smart" pill!

Okay, here's the deal flat-out, straight-up, etc:

As mentioned earlier, the devices on an IDE chain will function only as fast as the slowest device. In the case of a CD-rom/Burner, the device type is PIO4 or PIO5... This kind of device communicates at roughly 24MB/s. If you're burning from CD-to-CD at 24X (pretty damn fast) you're still only reading from the source CD at 3,600 K/s (maybe a bit more to create a buffer), and you're writing at 3,600K/s. A combined total of 7,200K/s or about 7MB/s. Clearly there is plenty of bandwidth to spare with this arrangement... The only problem you might encounter is a "buffer underrun" which will cause a burn to fail. This happens when the system cannot cache the files to be burned fast enough for the CD-rom to write it. The Bandwidth of the IDE channel is not the only issue here. It also has a lot to do with the Speed of your CPU and the amount of memory you have. When a lot of information is being transferred through an IDE channel a certain amount of Processor power is needed. This amount is actually kind of a lot with non-DMA devices. and will consequently slow down the buffer and transfer rates when burning.

An ATA100 IDE controller has a bandwidth of about 100MB/s that's a considerable difference from the PIO counterpart. Maybe not when booting up or letting some other function of a poorly coded Microsoft OS take over. But certainly when loading and saving audio/video files. Loading and Saving operations will require a sustained read/write period for the target device, if you're Hard Drive was running at almost 1/4 of it's full potential, I'm sure you'd notice a difference. The main advantage of an ATA100 drive setup is that it utilizes UDMA (Ultra Direct Memory Access). This function allows the target device to read/write at near maximum speed without using a proportional amount of CPU time. Not only will this help burning (when the source is a HD), but also it will prevent your system from slowing down when doing a routine save while working on a large project.

I hope this all make sense, and if it doesn't just remember that it all boils down to this:
Don't put your CD-Roms on the same channel at your Hard Drives.

Good Luck,
-Miles

I completely agree with milesg! Well said.

Good goal, Miles, but there is one problem with that if you're building an audio machine without either a raid controller or an auxiliary IDE controller - If your dedicated audio recording HDD needs to be on an IDE channel, and your boot drive needs to be on an IDE channel, and you only have two IDE channels (no add-on controllers) then where do you put IDE optical drives? This is why I've been advocating raid Mobo's, not necessarily for the raid function as much as for the two extra IDE connectors. Without those, the most practical way to hook up drives is to give the audio drive its own channel, and take the chance that apps won't load as fast nor the machine boot as fast with an optical drive on the same channel with the boot/apps drive. In a machine with only 2 IDE channels AND 2 optical drives, I guess you'd have to put the audio HDD on the same channel with the boot/apps drive, and let the two opticals take turns slowing each other down. (Especially with the new "burn proof" drives, where coasters are an endangered species)

This is why my new system will have only SCSI and Firewire opticals, leaving room for 8 HDD's on 4 IDE channels. Plus, if I find that a second HDD on a channel screws it up for speed, I will still have the option of 4 drives at a time, each on their own channel. (320 GB online at a time should be OK for a while)

For all you "lurkers" and posters, I'm sorry this subject is so confusing. All I can say is, I'm not the one who designed a PC using off-the-shelf and out-of-the-junk-drawer parts, then perpetuated the limited architecture years beyond its useful life - but a least we now have an option for more IRQ's, if all goes well. One outa eight ain't bad... Steve

Yes, I agree that having only 4 slots to place any IDE device is a bit of a limiting factor. SCSI or IEE1394 is of course a good solution.

Having both optical drives on the same channel and both HDs on a different channel (Boot and Audio) is still much better than sharing an IDE channel with a slower device... Sure, splitting the bandwidth of both hard drives isn't the best it can be. However, it's still better than lowering the HD speed to 24MB/s and THEN splitting it with the optical drive. The speed of most ATA100 HDs is about 40BM/s these days. If you double that, you still haven't reached the maximum bandwidth of the ATA100 IDE channel. So sharing a channel wouldn't prove to be much of a bottleneck (in theory at least).

The goal is to keep your DMA devices out of PIO mode. Those of us who can afford a second controller or SCSI card, of course will enjoy the benefit of greater speed... In the mean time, I don't think that's it's much of a limiting factor with the speed of today's drives.

-Miles

My new components are arrived, I`m ready to put them together, but there`s still the IDE Channel Setup. I got a DVD, a CDRW(plextor 24x) and two HD`s.
I went to my local computer guru and explained my problems. he told me, as long as I`m not using the optical drives it should not be a problem, especially with never drives.
Now what`s worse? 2 Harddiscs slowing each other down.(like milesg mentioned)
and shall I put the audio or the OS as the master device?

Or one harddisc(master) and one optical drive(slave), as Opus purposed. Sorry, but the question is still not solved.
Did ever anybody do some testing with both setups?

You should set the boot (OS) drive to Master of your primary channel and slave the Audio Drive to that.

You're more than welcome to also slave one of your CD devices on there... but I'm pretty sure that you'll notice a slow down.

Someone please correct me if I'm wrong, but I think that a HD connected on the same channel with a PIO device will not allow DMA mode to be enabled on that channel. Not only decreasing throughput, but also raising your CPU needs.

Changing the IDE chains after your system is configured, should not pose a problem, so experimenting with different configurations is something that I would highly recommend.

-Miles

PS I'd test this stuff out myself, but I'm only running one 500MB drive on a 386 w/4MB of EDO ram... hehe, j/k

This thread is really comical. It seems everyone has an opinion, but in the end, there really hasn't been a definitive answer. I'm not a puter genius like some on here, but applying simple logic, it's clear that we all need cold, hard facts.
Is there anyone here who has the knowledge to actually conduct a test? What we need is for someone to take 2 HDDs, a CDROM, and a CDRW, on two IDE channels, and test them in all possible configurations under the same conditions using an accurate, scientific means of comparison. I'm talking about actual numerical and/or graphing results.
There are a lot of people on RO building computers right now, and it's obvious that this IDE configuration dilemma needs to be resolved. Since it's obvious that lots of people here are configuring their IDE devices in different ways, then obviously some are doing it wrong and are losing out on performance.
So, is there anyone with knowledge, equipment, and software who is willing to take the time and effort to solve this perplexing dilemma?

The debate continues!

I've decided to put both hard drives on the primary controller, and the CDRW on the secondary. (basically because my motherboard only came with one ATA100 cable, haha)

Being an amatuer weekend warrior my new system is probably overkill for what I will use it for. I doubt if I will ever push it hard enough to notice a difference. By the time I do I'll be looking to build a 10 GHz system with six 1000 GB drives.

My question is, could this whole issue be avoided with a $15 PCI controller card for the CDRW? Or would that be bad?

Thanks to all who have replied, I love this forum!

"So, is there anyone with knowledge, equipment, and software who is willing to take the time and effort to solve this perplexing dilemma?"

SOS, you forgot one other ingredient - Time... I am more than a little short on that commodity for the next month or so, but DO have the other ingredients. By the time I get free, I imagine there will be 40 more posts and at least 2 or 3 more "final" answers - No matter, I will still eventually thrash this within an inch of its life once I get to it. In the meantime, anyone who wants to do a simple test could just go to echo's website and download the Echo Reporter (SCRATCH THAT - I just went there and apparently they have chosen not to share that useful utility any more)Here is a link to some other benchmarking utilities -

http://www.acnc.com/benchmarks.html

I haven't tried any of these except coretest, of which I have an old version. Some of these are DOS only - do NOT try to run them in a DOS window, you must reboot into REAL DOS and run them from a dos prompt. If this scares you, wait for someone else to do it and post.

Reiterating that I have NOT tried these, Bench32 looks like a good try for Win2k-XP, and Michaels is the only choice here for Win98/ME. Also, be aware that HD Tach claims NOT to preserve data on the test drive, and some will NOT work on partitioned drives. Testa Labs has some interesting info on drive characteristics, here's a link -

http://www.tcdlabs.com/hdtach.htm

Note the comments on speed of inner vs. outer disk tracks - makes me think of inner/outer partitions on each drive, outer for record/play, inner for archives...
From what I read, I wouldn't bother with the demo version of HD Tach. It seems pretty limited by the description. If you feel like spending $49.95 after trying the freebies, cool.

To run tests yourself: first, share an IDE channel with a HDD and an optical drive, go into system properties and enable DMA if it lets you, disable autoinsert if you haven't already, run the benchmark, note the # of tracks in read/write. Delete any files written by the benchmark SW between system changes. Then, switch to two HDD's on one IDE channel, repeat the tests, note the read/write performance (make sure to enable UDMA again if it lets you, this would also answer the question about optical drives allowing this when on a channel with a HDD)

This would allow anyone with enough awareness to even be posting here to do a basic test for themselves.

"My question is, could this whole issue be avoided with a $15 PCI controller card for the CDRW? Or would that be bad?"

Starman, if you have the resources (slot, IRQs, $15 :=)) a controller card would be a good solution. Every one I've seen has two IDE connectors/channels, so you would have 4 devices each with their own IDE channel. Not having personally used an outboard controller, I am only guessing here (an educated guess, hopefully) but REMEMBER: your primary and secondary IDE controllers EACH take their own separate IRQ. (For those who haven/t tried this, right click on your "My Computer" icon on the desktop, click "properties", click on the tab "Device Manager", then double-click on the "computer" line at the very top of the list. This takes you to the IRQ/memory assignment list. scroll down to IRQ 14 and 15 - they should each have two listings, one will say "Intel Bus master controller", or something similar, and "primary IDE controller if it's IRQ 14, "secondary IDE controller" if it's IRQ 15.)

My point is, without yet trying it, I believe that a PCI controller, while giving you 2 more IDE channels, would also require 2 more IRQ's. Finding these in a machine that prefers not to share IRQ's in the interest of speed, could get interesting. In the case of some Mobo's, it might not even be doable. Still, sharing an IRQ, if it works at all, is preferable to sharing bandwidth all the time, so if it were me I would go for it. $15 is a cheap education.

If you don't have the patience to wait for me (and few do, I'm slow and cantankerous) you should be able to run a basic test for yourself with the above info.

one more test to try:
If a controller card is using one IRQ, would it be better to put your CDRW, and CDR on let's say a ATA66/100 controller card(promise) and your HD's on the main board controller(seeing as how they have they own IRQ-14 & 15), OR vice-versa, would there be better throughput for the hd's on one IRQ and let the controller do the work. Right
now I am using my new machine(the OPUS- ANUS) in scenario B and it is working great. I am running my OS/APP & Audio drive drive off the promise TX2 controller(no raid)card, and my CDRW & CDR & removable drive bay, off the main board controllers. Now I am wondering if I should do it the other way so the HD's would have their own IRQ's. Food for thought. what I do know is that right now, my machine is getting work done, and I am making music , not F**king with it. I could try this test, but for many reasons I just don't give a rats ass right now(maybe in a few weeks). Making music is much more fun !!
Take Care All(TCA),
JT

Resurrected from the dead! A post from April 2002!

I just got some questions answered, and in the process found out that my data HD- which hasn't been performing up to par as an ATA100 disk is in PIO mode, thanks to the crappy DVDRom I have on that IDE channel. How nice.

And, of note, there is STILL not an answer to where to put 2 HDDs and 2 optical drives....as far as I can read! :D