Theorising: Will we ever see 10k rpm IDE drives?

As far as I see, the only difference between IDE and SCSI is the interface. There is nothing technically wrong to equip a 10K drive with an IDE interface. So I guess we will see them. Actually Im surprised that we havent allready. Perhaps some marketing guys still wants a clear destinction between mainstream IDE and high-end SCSI.

except for profit margins.
IDE apparently has extreemly slim profit margins. why would they want to jepodise their 10k rpm scsi line by producing 10k IDE?

Thats why im thinkin that 10k scsi will have to become "old hat" and much less profitable before we see 10k IDE.

on a technological front though you are correct. would be quite easy to equip a 10k drive with IDE electronics.

<b>"True communication is possible only between equals, because inferiors are consistently rewarded for telling their superiors pleasant lies than for telling the truth."</b>

IDE drives rely far more on the CPU than SCSI drives. In CPU and hard drive intensive tasks, such as video editing, some people will always prefer the SCSI interface over the IDE interface. Sure, IDE drives can match the performance of SCSI drives, but they'll always be more taxing on the CPU. In a gaming or home PC, CPUs have reached a point where that doesn't matter, but in some CPU intensive tasks, it does matter.

Intelligence is not merely the wealth of knowledge but the sum of perception, wisdom, and knowledge.

once SCSI has reached 15000 or so standards, then 10000IDE drives may show up (I am thinking 2nd half 2003 at the earliest... fast new-type drive for the Serial ATA bus)

no-one shouts louder than someone who is being ignored, or in the case of techies, to be heard over the noise of their PC's ;-)

Aww, predictions are difficult, especially if you're predicting the future. Being the eternal pessimist, I would say that the warranty reduction from the major manufacturers (Maxtor, Seagate, Western Digital) <i>might</i> be the advent of shoddily manufactured 10krpm drives in the IDE market.

Still, I'm rather sceptical about it. Profit margins in the IDE market are far too slim to introduce high-performance <i>mechanics</i>. IMO the average seek times of current IDE drives are textbook example. To my understanding, 4ms - 5ms average seek times have been a reality in the SCSI world for some time now. In so called high-performance IDE drives the average seek time still is >8ms. Ever wondered why?

My own guess is the density requirement. In the IDE market, it is mostly about "bang for the buck". IDE densities are <b>huge</b>, so I'd think that they require über-accurate seek mechanics. IMO there is a tradeoff between accuracy and speed. In the SCSI - or rather high-end server - configuration, speed (seek time) is the crucial factor. Increased rotation speed mostly serves to reduce the real-life seek times. Any density limitations are overcome by (noisy) RAID configurations that would be unaffordable/insane for "a couple of users at a time" configurations.

Another big hurdle is the plain vanilla PCI bus. It's 128MB/s total maximum, and 128MB/s for HD alone is very much wishful thinking when there are network cards, sound cards and the like connected to the very same bus. SATA on its own can do nothing about it. Increasing the rotation speed in order to get higher sustained transfer rate (and lower real-life seek times) would be rather useless when the ongoing IDE density race can accomplish the same thing. IMO, again IMO, we won't be seeing higher rotation speeds until mainstream IDE controllers (SATA?) are be connected to something <i>much</i> faster and more scalable than PCI bus.

In the meantime, the IDE trend seems to be about density, seek time and rotation speed. Pick one, and cover any shortcomings in the remaining two with firmware optimizations and bigger cache. IMO, WD has already paved the way for choosing density over the other two with their JB drives (8MB cache) and by offering three-platter 200GB IDE drives.

Who'll dare to argue? :smile:


<font color=red><b><i>You want WHAT on the #$#%## CEILING?!</i></b></font color=red> -Michelangelo

Ive heard this numerous times before, but I never really understood why. Is the ATA interface so much more different from SCSI? I have never studied the SCSI interface so I know little about it. But I know the ATA interface well, and it has commands for reading/writing sectors, DMA transfer etc.,etc.,etc. But many of these low level commands are handled by the IDE controller. The CPU is only involved at a higher layer in interpreting the filesystem (e.g. FAT table etc.). And the CPU has to do this regardless of the underlying physical interface. So why does ATA load the CPU more than SCSI?
Please enlighten me!

Well, back in the good old PIO days (no DMA), that statement about CPU dependency was certainly true. Anyhow, I'm under the impression - <i>please</i> correct me if I'm wrong - that SCSI controllers are able to handle e.g SCSI->SCSI transfers entirely on their own. As opposed to DMA read -> DMA write which requires CPU to set up the transfers. And other stuff like that. Handy but not really meaningful unless the drivers/OS support it.

On a final note, IMO IDE has become largely CPU independent with multiword DMA transfers and UDMA. E.g X-bit Labs has done their review with a 600MHz Pentium something processor and their performance results are reasonably comparable to recent THG results with some 2+ GHz Pentium 4 processor.


<font color=red><b><i>You want WHAT on the #$#%## CEILING?!</i></b></font color=red> -Michelangelo

I don't tkink we will ever see any 10,000 rpm IDE drives, but I think we will see plenty of Serial ATA 10,000 rpm drives in the next couple of years.
I think that Serial ATA will replace IDE quickly. I just hate these cumbersome IDE cables.

__________________________________________________
It's not important to know all the answers, as long as you know how to contact someone who does.

hmmm. we cannot stick with 5400rpm and 7200rpm indefinately, capacities are getting to the point where scandisc's and formats take hours!

an example of this is Maxtors up and comming next gen drives. Apparently they will have a monster 5400rpm drive with 4 80gb platters. yep. 320Gb on one drive, and 250Gb for their 7200rpm version!

<b>"True communication is possible only between equals, because inferiors are consistently rewarded for telling their superiors pleasant lies than for telling the truth."</b>

uhh, serial ATA is a form of IDE. :smile:
different transfer protocol, same stuff onboard the drive & controler (i think).

but your right in one sense, those nice serial ATA cables cannot come soon enough.

<b>"True communication is possible only between equals, because inferiors are consistently rewarded for telling their superiors pleasant lies than for telling the truth."</b>

Yes, SATA is designed to be as much backwards compatible as possible. However, one big difference is that there will no longer be any master/slave hassle (I think). Each HD will have it's own dedicated SATA channel and 150MB/s burst transfer rate at it's disposal.


<font color=red><b><i>You want WHAT on the #$#%## CEILING?!</i></b></font color=red> -Michelangelo