Independent Device Timing

You definately get the award for asking questions - and. thats not all bad either. What you have referenced is your Northbridge controller and most people put hard drives on a RAID controller - that's what it was designed for. And, while I don't think you can run optical drives on those particular IDE headers, quite frankly - I'm not 100% on that either because I've never tried it. Guess I look at it this way - with 4 IDE headers - when would one need to run a CDRW on the same cable with a HDD and why would one want to? That takes away all your flexibility with RAID configurations if it did work. I've got one MoBo with 4 IDE headers and 2 SATA headers (not set up yet in any machine) but thats 6 HDD's and 4 opticals or whatever - have yet to see or hear of anybody that needs that kind of capacity and that includes file servers. Good question though.

 

Hi.

In the context that you ask the question I think you'll find that the Independent Device Timing simply means that all devices will actually run. You can run IDE-0 and IDE-1 at different speeds but not a master and slave at different speeds on a particular IDE port. It means that devices with two different speeds WILL RUN, it doesn't mean that they will run at their own different speeds. I think the definition of IDT has changed over time and doesn't mean what it used to mean. I think it currently means you can run each device at it's own speed. I think it used to mean what I described above.

Also,I think you mentioned in another post that you were thinking of getting round cables for air flow. The flat ribbon cables were designed to reduce cross-talk between data lines. The new 80-pin cables have 40 extra strands that act as insulators between signal channel strands. With the higher speeds it's important to maintain isolation. The round cables defeat this purpose as they fold data channels together. I would recommend that everyone switch to the 80-pin cables when convenient, even for the cd/dvd applications. The higher isolation of the lines can only benefit the data flow.

Another comment. Some card controllers (such as the Promise Ultra100 TX2 card) advertise separate speeds for each device. These cards actually have four IDE controllers built in that allow this capability.

-------
Western Digital says:

If your EIDE controller supports asynchronous data transfer rates (different rates for each device on the controller), you can add a second device that uses a different ATA rate without affecting the rate of an existing device. If the controller does not support asynchronous data transfers, the system will run at the fastest rate supported by both devices. For example, if you added an ATA/66 drive to a system with an existing ATA/100 drive, both drives would operate at ATA/66. Check with your controller card manufacturer to determine if your card supports asynchronous transfers. This may also be referred to as "independent device timing ".
----------------

Giles

 

Hi.

Us old programmers that lived daily with the fact that 0(zero) is a valid number get used to IDE-0,1 vs IDE-1,2 and seldom think about it. We just automatically work it out. Doesn't help matters a bit tho.

You are correct re the old and new definitions of IDT. I think that is what's happened. However it's important to remember that IDT (old definition) means that devices of differing speed ratings can run together on one IDE port, it doesn't mean they will run at their individual speeds. The new difinition just floated into the SATA drive capabilities since they now assume IDT means serial transmission.

I think the round cables were made to sell as a fad to help with air flow only and not as a result of generating a better cable. If data is mis-sent from the hdd to the computer it is read again (up to a pre-specified amount) until the data is read correctly. This being the case you can actually lose speed and never know it. If the round cables have individual line shielding they would be fine. Otherwise, for precision the 80-pin cables are called for.

The Western Digital term "asynchronous data transfer rates" doesn't really apply to IDE ports. The term "asynchronous" is serial transmission like USB or SATA drives. Since IDE ports are parallel IDT could not apply to them unless you actually have four separate ports. As long as there are two plugs on one cable plugged into one IDE port it can't be IDT thus the statement from Western Digital holds true for most all motherboards.. With SATA you have a plug per drive so you can run drives of separate speeds without inteference.

Also, what your bios and system is reporting re the speed of each individual device is not the final speed at which the data will be transmitted to the computer. It will always be at the speed of the slower device when intermixed on a particular IDE port. There are cases where the speed has actually been clocked at somewhere between the higher and lower speed as if it were averaging somehow. That could be because of buffering by the slower device and thus a higher burst transfer rate. I've never pursued this so I don't know for sure.

Thanks for the pointer. I'll try and read through that. Might clear up some things.

Giles

 

Guess I did the ol' ass_u_me when you referenced 686B because thats usually the Southbridge controlller used in conjunction with a Highpoint controller for RAID. Now that I've eaten a little crow twice in one day (Daddad understands), I guess I'm encouraged by your intellectual curiosity because you really do ask some pretty good questions - thought provokers - but again, ain't tried it so can't say.

I read with interest what Giles has to say and as a practice, have never mixed opticals with HDD's on the same cable. In the old dog new tricks category, some of your questions are most refreshing.

 

Hi.

Oh, criticism is just fine. That's how we boil it down to some useful information for the user. On the same token, I tend to overexplain things. That isn't ment for anyone in particular. It's so anyone reading it might get to the point we are at.

You're absolutely correct about SATA and IDT. I can't imagine why a company as large as Western Digital with the power behind their comments would assign the IDT terminalogy to apply to "asynchronous data transfer rates" when IDT was a term applied to being able to run devices at different data rates on the same IDE ports, which are parallel in design. With the weight of their words they have essentially reasigned the capability. Maybe someone just wrote that blurb without their morning coffee and thusly changed IDT forever.

In searching for "asynchronous data transfer rates ", you shouldn't get any hits for IDE because "asychronous" is serial and IDE is parallel (except, of course, according to Western Digital re IDT).

I've always had trouble with the short IDE cables. I think they are that length because of saving manufacturing costs rather than a calculated length. Wavelength (in meters) = 300/frequency (in MHz) so a ATA/100 MHz cable would be 3 meters=118.11 inches and ATA/133 MHz cable would be 2.25 meters=88.58 inches. Of course it would be a problem trying to tuck all that cable in between the floppy and the hdd. Still, they could have made life much easier with the cables a little longer than they did.

Since hard disk drives are in the very early stage of being "on their way out" it probably doesn't really matter. In either case a very interesting exchange.

The season is upon us. Eat hardy and enjoy. The best to you and yours.

Giles

 

giles

Would you like to elablorate on that comment. Are you looking at "chip" storage replacing HD? Would be a logical progression. Can envision chip/HD hybrids as a stepping stone.

 

Hi Dennis L.

Actually there are two technical interchanges going on at present. One is in the area of monitors and TVs. A year ago there was only 4 manufacturers in the world that made Cathode Ray Tubes (CRTs). All displays are moving quickly towards the LCD and Plasma screens. The prices are coming down and a lot of people are starting to buy them outside of government and business applications.

You're correct on the hard disk interchange. The USB Pen Drives are starting to gain a little and it won't be long before one of the manufacturers will get a bright idea in that area.

Say you're running XP with a lot of software loaded. Excluding MP3's or movies you're probably using a fair bit less than 10Gigs even with your data. (This doesn't apply to everyone of course, just average.)

You can get a small cable that expands a USB port to 4 USB ports for about $6.00. Two of those would give you 8 USB ports. They have USB Pen Drives of 1Gig available. That gives you 8Gigs of on-line memory disk space at the speed of the 2.0 USB port. With a modified RAID setup on the USB ports you could look at all those drives as one single drive.

The cost would be prohibitive at present but it's starting, just like the monitor/TV interchange. The current 1Gig memory drives are about $300USD.

At first they will probably come out with about a 5Gig and go from there. Cost will come down, the voltage of the units will drop so heat will drop on higher density units. From mechanical to solid state. It's a natural flow of technology. Less parts, cheaper manufacturing costs. They will probably design a better interface than the USB, probably pick up the SATA and run the ports to the back plane of the motherboard.

On one hand it doesn't make sense to go serial rather than parallel. Pushing 8 lines of data simultaneously makes more sense than pushing all those 8 lines into one data channel. (IDE to SATA.) However, if a motherboard can generate a saw-tooth pulse to run a cpu at over 3GHz then it can push data through a single line at over 3GHz serially. That would generate about 300Million bytes per second speed. Without a mechanical head to slow you down you can develop memory drives (MDs) to read and write at over 3 Gigs speed. Thats twice the speed of the new SATA150 drives and the speed of CPU's are going up all the time thus the clock speeds on motherboards are going up all the time.

There are other problems that hold things back at the moment but they will fall by the wayside soon enough. It's in the early stages. Essentially all the capability is available right now. Just needs some market profit potential to push it along. It may need a medium other than chips to contain the memory for the drives. If that gets changed it'll be as important as the chip development itself. Outta get interesting.

Giles

 

Hi Christer.

Any mistake anybody else in computers can make I've already made at least 5 or 10 times. Just ignore it and get on with getting on. We all help each other.

That was a very good page you referenced. It has a good explanation of why round cables are a potential problem and the user probably will never know. It also specifically says that round cables WILL NOT help in cooling. Everybody should read that page a couple of times. Really good pointer.

Thanks for the Merry Christmas. I got a new electric brad nailer for my woodworking shop. YEAAAAAAA!!!

Giles

 

Hi.

I haven't tried those round cables so I can't really comment. My system has an exhaust fan on the power supply that is self regulating. Normally it runs about 1288 rpms. There's a fan on the cpu (2.2Gig P4) that runs at 2700 rpms but no other cooling in my system. I normally run at 87F but I have seen it at 89F several times. I can force it up to 104F with PRIME95 but that is far below what the P4 is rated at. My system is so quiet it can sneak up on you. No mods at all.

I was thinking about setting up a system with about a 2.5 to 3.0Gig P4 and seeing how far I could overclock it. I would run all drives external, disconnect the power supply fans and no other fans. Insulate the case on all six sides, seal it in plastic and run an insulated tube from a chest freezer set at about -10F. (I've got one available.) use a small in-tube fan and blow directly onto the cpu and run a tube back to the freezer. Those fans can run forever and i would power the fan external from the computer so it never turns off, that way no condensation.

I think I could get about 5 to 8Gigs out of it if I could get a motherboard and memory that would go along.

That should be great fun.

Giles

 

Hi.

Well, according to the published old definitions of IDT your C drive should be actually transfering data at the DMA Mode 2 speed of the DVD-ROM. A speed test with and without the DVD-ROM plugged in would be interesting.

Giles