[hnayoung]

I was wondering if there is a thing as IDE bottleneck. What I mean is if you have a device that can do UATA 133 as master and another hooked up to the same cable that runs at ATA 100 as slave, will there be a bottleneck or anything?

 

[D.D.]

yes i think it defaults to using the slowest in the chain.

 

[Player-X]

the real bottlekneck for a harddrive is the drive mechanism rather than the cables

 

[Kane]

There is a lot of incorrect or out of date information floating the net about IDE drives.

>yes i think it defaults to using the slowest in the chain.

This was true 10 years ago.

There is also the one about the IDE chain runs as fast as the master device on it. IIRC this also changed a few years back so the chain runs at the top speed of the device it is accessing.

However, there is a bottleneck from 2 devices on hte same chain. Only one device can be accessed at a time. You can't read from one while writing to the other. It's taking turns. A slow device will slow this down as a result as the faster device is waiting for it's turn.

 

[refraction]

Hard drive technology in itself is a bottleneck. There is 2 major drawbacks holding Hard drive speed back

1. The technology used for hard drives, the heads just cant move fast enough to get the data across any faster. This is true across all formats, even SATA suffers from the same problem, so dont expect any huge boosts from changing to SATA drives.

2. The IDE controllers run off the PCI bus, so the PCI bus speed is also a limiting factor on hard drives, altho with the introduction of PCI Express this bottleneck is removed, due to there being higher bandwidth availability.

As soon as the technology changes we will see an improvement, There was talk a while back about crystal hard drives which use lasers to store data on a crystal, altho i imagine itll be a good few years before that comes about.

 

[D.D.]

There is also the one about the IDE chain runs as fast as the master device on it. IIRC this also changed a few years back so the chain runs at the top speed of the device it is accessing.

However, there is a bottleneck from 2 devices on hte same chain. Only one device can be accessed at a time. You can't read from one while writing to the other. It's taking turns. A slow device will slow this down as a result as the faster device is waiting for it's turn.

ah, ic. in that case, the MS database , iamnotageek, and lot of ppl on the internet must be confused. can i please have a link that explains what u have posted please Kane? i wud like to stop this confusion from spreading further :cuss2:

 

[Kane]

>Many of the current Enhanced IDE controllers do not support separate transfer speeds of the slave and master, so if you have a fast master device and a slow slave device on the same channel, your controller may lower the maximum transfer speed to the slowest device.

I think you'll find, like I said that that is outdated information given for compatibility purposes. Most EIDE controllers these days expect people to use powerful devices on every port of every channel

 

[james.miller]

thats correct, Kane. 99% of ide controllered these days are cabable of Dual FIFO, where each device can run at it rated speed regarless of whatever else is plugged into that cable. I've seen it on every motherboard ive owned since my old pII-350 on a intel 440bx chipset.

 

[Esturk]

And keep mind unless you're using a RAID array, most hard disks don't even reach 100MB/s anyways. Some hard drive companies don't even make ATA133 hard drives because they simply don't go that fast due to technolgical limitations. Theoretical Bandwidth.

 

[D.D.]

GRR. thanks kane and james.miller. i wud have thot that MS and iamnotageek wudnt be spreading fud that was outdated 10 years ago (seeing how their articles were written within 2 years of now O_O) i dont like it when ppl post wrong info on the internet