The performance increase is minimal as it's not parallel ATA holding our drives back, it's our drives themselves that are the bottleneck. What it will bring is a lot of dongles being sold, new power supplies (serial ATA drives require a new 15 pin connector, unless they changed that part of the design) and more cost per motherboard to supply the same number of ATA channels as serial ATA provides 1 channel per controller vs 2 channels in master/slave per controller with parallel ATA. This is unless everyone agrees to sell their new serial ATA controller chips for half of what they charge for parallel ATA controller chips. True, the connectors take less board space and the traces are fewer and more easily routed, but there will have to be more of them.
Nice pic of that new power connector
here.
The article writer seems to envision serial ATA taking over the world and moving SCSI aside.
As for the voltage concerns mentioned, these are in reference to the future limitations of parallel ATA - which we aren't even close to. While we will get there eventually as drives finally catch up to the ATA bus, be it parallel or serial, there's no rush.
Smart command queuing is part of the serial ATA II spec. Is II the one that's going to be rolled out to start with?
I guess I just expected more somehow. In an era of 15 devices per SCSI channel, 63 per 1394, 127 per USB....1 per serial ATA controller seems rather the wrong direction. And if we're going to require new power connectors and part of the goal is to clean up cabling, why not put the power wires and data wires all in the same cable as the spec? Personally I find routing power wires to be just as much of a pain as parallel ATA cables at this point and the power cables will be even more of a nuicense when serial ATA makes the data cable less obtrusive.
--Mc