Level Extreme platform
Subscription
Corporate profile
Products & Services
Support
Legal
Français
Windows swap file
Message
General information
Forum:
Visual FoxPro
Category:
Windows API functions
Miscellaneous
Thread ID:
00172049
Message ID:
00172295
Views:
26
>>There are several drawbacks to using a separate IDE drive, especially on the same IDE channel. IDE isn't a terribly smart interface; the communications mode for a channel is determined by the -least- capable device on the channel. IOW, if you have a new UltraDMA33 drive and an old original IDE drive, everything will talk at the capability of the old IDE drive. If you're going to use a separate drive for a swap file and it isn't as capable as the main drive, put it on the secondary IDE channel.
>
>Thanks for the insight, Ed. My swap drive happens to be on the secondary channel on one PC and the other is using SCSI anyway so, by chance, my setup is ok. I will keep your reply in mind for the future though - I wasn't aware of the comms mode determination. Do you know what happens with an IDE CDROM drive as second device on IDE channel?

The channel uses the best common mode available to all devices on the channel. IOW, if you have the latest UDMA 33 drive with busmastering drivers and a cherry on top, and an IDE CD-ROM drive that runs in PIO Mode 2 on the same channel, the channel runs at PIO Mode 2.

It gets worse; unlike SCSI, which has a disconnect/reconnect protocol available to it, only one device can have an activity 'in process' at a given time. If your CD-ROM and primary drive share a channel, and you have to do a long seek, the channel can't be used by anything for the duration of the long seek, which could be hundreds of milliseconds. Disconnect/reconnect on the SCSI bus allows an action to be started and the SCSI bus then released by the tartget (the device doing the work); it reacquires the bus when the task is complete. Most decent busmastering SCSI adapters (like the Adaptec 294x family) will support multiple targets (devices) active simultaneously and provides good mechanisms for bus arbitration when multiple devices are active, With TCQ, there's even an option for multiple simultaneous operations in progress on a single device, with some flexibility in ordering the service of requests.

In reality, it doesn't make much difference in the typical desktop. The hard drive itself, not the disk-to-host transfer speed that limits the performance of the drive system given the typical desktop, with a single or dual disk environment. PIO Mode 3 supports data rate a bit over 8MB/sec between disk and host; there are few IDE drives that can do better than that (even the top-rated 10KRPM UW2 SCSI drives would be hard-pressed to saturate PIO Mode 4, at 16.6MB/sec.) If the drives are on separate channels, the have no real effect on each other; there's no problem with transfer mode or channel tie-up if the devices are on spearate channels. Once it does become an issue, SCSI is a logical next step.

FWIW, all the systems I use here at home have Adaptec SCSI of some kind on them, and all but one has the CD-ROM or DVD drive on the IDE channel.
EMail: EdR@edrauh.com
"See, the sun is going down..."
"No, the horizon is moving up!"
- Firesign Theater


NT and Win2K FAQ .. cWashington WSH/ADSI/WMI site
MS WSH site ........... WSH FAQ Site
Wrox Press .............. Win32 Scripting Journal
eSolutions Services, LLC

The Surgeon General has determined that prolonged exposure to the Windows Script Host may be addictive to laboratory mice and codemonkeys
Previous
Reply
Map
View

Click here to load this message in the networking platform