[DosFreak]

I've seen the 150m spec plastered all over creation but never any REAL benchmarks. I need comparisons between ATA & SATA dangit!

[JP-]

well maybe when there are some proper production drives you might see some

[Brian Frank]

I'd be intrested in the SATA benches too, as I don't see how the interface is going to speed things up in real life--at least without a faster RPM on the ATA drives.

[Alien]

Well, I think I read that Seagate has already made a drive for it, & Iwill has apparently made mobos & PCI cards, so might not be too long.

Rumour has it that someone was going to try & use the technology for possible networking applications, but then religious luddites heard about it..... :p

[BladeRunner]

If Serial ATA offered no real-world performance boost I'd still go for it.
I love the idea of one tiny little thin cable running between all my IDE devices rather than the bulky ones we are using now.

No HD manufacturer has any plans to go above 7200rpm with IDE drives at the present time.
In fact the only company who publically showed research into this was IBM who had 10000rpm IDE units designed, their failing point was heat - they were simply generating too much.
However now IBM have sold off their HD division to Hitachi it would appear that none of them are working on faster units.

[JP-]

Maybe we should start looking at alternatives to the spinny platter HDD idea. Suppose youd have to consider how it would store the data when the PC was off, and how you could get GBs of storage for a decent amount of moola

Im sure someone will come up with something

[CrazyKillerMan]

I think it is Sharp Electronics that is developing a film substrate that goes on the silicon that in non-volitile and about 1000 faster than flash memory to replace the regular old volitile memory.

This could theoretically replace the hard drive, seeing as though it is non volitile, and there is no moving parts either so very little heat.


The only problem is that they know it works, but they dont know why.

[Uykucu]

The next thing in storage is molecular level optical storage systems which will store the data in the lattice's own fabric (crystal structure obviously)
This idea popped up after they have learned how to read quantum state of a particle without bouncing a photon off it (therefore changing the state and kicking it to god knows where).

I dunno exactly which companies are working on it and although heard rumors of prototypes have not seen one therefore i discredit them myself.

What this means for you is basicaly instantenous access to your couple of thousand terrabyte of data at will at something as small as 1 cubic centimeter. The power consumption will be next to nothing for reading the data but unless they know something i don't like changing the planck's constant, writing a MB file should take something between a Hydrothermal Power plant and your personal Nuclear reactor in your case.

I have no idea how they are planning to fit or cool those down.

What crazy killer is describing is something similar maybe a lower technology application.

BTW those are called solid state or crystaline drives...

[CrazyKillerMan]

Yea - I was reading the article earlier - i know it involved crystals but I couldnt remember the name of the process.

Black box's do have solid state storage now.

I also checked out some thesis's on the matter and it was very very intense.

I dont know how similar it was though - I'll get back to you when I find the article.

[pmistry]

Fujitsu was also experimenting with 10 000 RPM IDE drives, but they too suffered with heat issues.

[JP-]

I dont get it, how can an IDE 10,000rpm drive suffer from 'heat issues' but a 15,000! rpm SCSI drive can be made with no prob, they both spin dont they !??!?!

[BladeRunner]

*Shrugs*
I don't work for IBM

Now by trade I am a software person although I do dable with hardware.
However I'm not a hardware engineer to the degree of knowing exactly what goes on inside various computer components.

I can only assume that the technology behind SCSI & IDE HD's is different throughout the unit, not just at the interface level.
I kind of assumed that the drives were more or less the same with the exception of the interface.
However if this was the case then "porting" 15k SCSI units to IDE wouldn't be a problem.

The 15k SCSI units do get hot though, we've got five of them here in a RAID array and they certainly generate some heat.
That is both Fujitsu & IBM publically saying there is a heat issue with 10k IDE units, so there must be something in it.

Maybe somebody with a little more mechanical knowledge can explain what inside the HD makes the SCSI unti so much different from the IDE where SCSI can reach 15k and generate heat, but not enough to effect stability or performance where as with IDE 7,200 appears to be our current limit due to heat.

[Uykucu]

Yo need a bloody large heatsink and extra fans in the case with 15K SCSI's they do heat seriously.
Try to run a DB with couple of thousand users...

Although not sure.
SCSI drives are bigger then IDE's so might have more plates etc.