sata/ide bandwith limitations
Out of curiosity, What's the limiting factor in speeds right now, the interface or the drives? For instance, could an ide hard drive perform faster if it wasnt limited by the ide interface? If i got an ide to sata adapter, whereas the sata has 150mb/s IIRC, would the ide drive go faster because it's not limited by the ide bandwith limitations? I'm just curious as if this would work. Thanks for any replies.
IF you have a super fast drive you can expect around 55MB/s transfers from it. The only time you exceed that speed is in "cache burst" which is used up rather quickly.
<font color=blue>There are no stupid questions, only stupid people doling out faulty information based upon rumors, myths, and poor logic!</font color=blue>
The limiting factor of transfer rates right now is basically the drives IMO. We need to get past 7,200 RPM, which thankfully WD did for us a couple of days ago (they are going to begin selling 10,000 RPM IDE drives, 10K was previously only reserved for SCSI).
The interface for drives is a limiting factor, but no harddrive company is "unlazy" enough to basically almost eliminate the interface overhead. This would require making a new interface, and one which motherboards would have to be equipped with (otherwise you would have to buy a controller card). MB manufacturers would be too lazy to do this as well unless this new upcoming interface was widely accepted as something that would sell.
Like you could make a new interface called "Super ATA 75" And if your harddrive's read speeds were almost constantly around 75MB per second, then you'd have an accurate statement.
Maxtor tried to do this a bit with the introduction of ATA 133 (on parallel), to try and get rid of the interface overhead as much as possible.
But the whole ATA 100 idea is just a big lie, the only way to get data transfer to nearly that level if through simultaneous transfer in RAID mode.
For a truthful and honest statement, they should call the current gerenation of IDE drives "ATA 50"
Benchmarks don't lie :-)
Typical transfer rate for "super-fast" hdd are:
SCSI 15k rpm 60 MB/s
IDE UATA 100/133 7.2k/10k rpm 55÷58 MB/s
burst transfer rate 90÷95 MB/s
this value are still lower than the standard UATA 100 interface and it is not expected to go very soon, beyond 70÷80 MB/s. Even considering to build a 15k ide hdd the limit still apply because increasing the speed must be reduced the data density and the resulting transfer rate does not proceed parallel to the speed increase.
Very probably the limit is the PCI bandwidth when RAID 0 system are used. In fact a RAID 0 array can easily reach 100 MB/s against a theoretical value of 133 MB/s of the PCI bus. Only a 64 bit/66MHz PCI bus with its 266 MB/s can sustain the potential of a high end RAID 0 system. But the interfaces ATA or SCSI are not yet involved as bottleneck.
Principle of the highest harassment for engineers says: having two events, one bad and one favorable, with the same probability to occur, it is most probable that the bad event will occur
Actually Unoc, you are very wrong about the transfer rates here.
If 15K SCSI is transferring from drive to drive, it can easily achieve over 100MB per second.
Also, if latest 320 SCSI drives are targeting your RAM or processor, the rate there is about 75MB per second, on a well defragmented drive.
Have a look here about Fujitsu's latest models
<A HREF="http://www.fcpa.fujitsu.com/products/hard-drives/mas-15k-rpm/specifications.html" target="_new">http://www.fcpa.fujitsu.com/products/hard-drives/mas-15k-rpm/specifications.html</A>
IDE drives will only achieve 55 to 58MB per second under the <b>best</b> of circumstances.
Also, 64-bit PCI runs at 533MB per second, not 266.
Benchmarks don't lie :-)