Serial ATA CD-Rom Drives?

OutRigged asks: "With Serial ATA hard drives starting to go mainstream, and being almost equal in price to their parallel equivalents, one would think we'd have Serial ATA CD-ROM drives by now. Yet wherever I look, all I see are PATA based CD-ROM drives. It's obvious that an optical drive will benefit little, if at all from using SATA, but why not switch for the sake of the cable size? CD-ROM drives are usually at the top of the case, and with the 1m limit in length, along with the small size of the cables, I see no reason not to use a Serial ATA interface in a CD-ROM drive."

Comments

bridge chips

[uberhund2]

Most of the SATA hard drives are still just parallel ATA with a bridge chip to convert them to serial. I imagine that once manufacturers switch to native SATA hardware, the reduced costs will send SATA to CDs, DVDs, etc.

Re:Internal Firewire?

[drinkypoo]

Like SATA, PATA, and (IIRC) unlike SCSI, 1394 can be installed as an asynchronous or synchronous bus. As such, it need have no more latency than anyone else. Programmatically it looks much like SCSI, being a register-based protocol. SCSI also has inherent support for things like multiple devices per channel and hotplug (All the mentioned standards support multiple devices per channel.) You can also do tagged queueing of firewire disks; they are typically attached via SCSI emulation. After all it's not the drive that supports ordered queueing, but the controller/host adapter.

As for saying what with drives today SATA's 150MB/s will never be saturated, what about drives tomorrow? Why pick a standard without room for growth? Firewire is at 1.6Gbps today (though only in sampling quantities) and the 1394 WG has a plan to move to 3.2Gbps over fiber, providing 400MB/sec. As for "drives today", 800Mbps 1394 is adequate, with 100MB/sec transfer rates, since individual drives rarely provide more than 20MB/sec transfer under any conditions. However, 1.6Gbps 1394 provides more bandwidth than SATA, and allows you to connect enough drives to utilize it. What's more, it's designed for external use, so like SCSI it is irrelevant where you put your devices. It allows greater cable lengths than modern-day high-speed SCSI, however.

It remains to be seen how the upcoming serial version of SCSI will perform, but it is safe to say that it will continue to be costly. 1394 is easy to implement, flexible, full-featured, and here today. The only thing preventing hard drive manufacturers from making 1394-native (or apparently native - some cheaper SCSI drives actually have a SCSI to IDE bridge built onto the controller board) hard drives today is the lack of demand. What I don't get is why there is a lack of demand; More and more PCs can now firewire boot, macs can firewire boot, and it would be lovely to diminish the number of goofy interfaces on the system board. Realistically, you need only USB2, IEEE1394, and an AGP slot to cover 99% of users. (Not counting the DIMM slots, CPU socket, etc, of course.) Input devices and network interfaces can go on USB2 (I would also like to see systems have onboard GigE, though), your video can go in an AGP8x slot, and storage devices can live on 1394. This would produce a truly legacy-free PC, without making it more expensive; in fact simplifying it to this degree would reduce the cost. Meanwhile, those who require legacy IO can plug something into USB.

ATA is terrible. SATA is much better, but still has silly limitations, namely cable length (if it doesn't support external drives, it sucks) and the number of devices. SCSI is pretty decent but the cabling is complex and every piece of the system is overly expensive. Firewire can replace all three of them even if we don't have native firewire drives, but it would make much more sense if we DID have them. Cheaper, faster, cleaner, better. Legacy-free. In other words, all the things we've been asking for. Why is there such opposition to such an idea?