The scsi() syntax is used for both RISC-based and x86-based computers, and is used in all versions of Windows NT. Using the scsi() syntax indicates that Windows NT needs to load a SCSI device driver and use that driver to access the boot partition.
Note
The term scsi() syntax refers to the scsi(W)disk(X)rdisk(Y)partition(Z) form of the ARC pathname.
On an x86-based computer, the device driver is Ntbootdd.sys, which can be found in the root of the system disk (usually C:). It is a copy of the device driver for the disk controller in use. On a RISC-based computer, the driver is built into the firmware, as required by the RISC standards, so no file is required.
The W, X, Y and Z parameters have the following meaning when using the scsi() syntax:
W
is the ordinal number of the controller as identified by the Ntbootdd.sys driver.
X
is the SCSI ID of the target disk.
Y
is the SCSI logical unit number (LUN) of the disk that contains the boot partition. Y will almost always be 0.
Z
is the partition number. All partitions receive a number except for type 5 (Extended) and type 0 (unused) partitions. These numbers start at 1, as opposed to all other entries, which start with 0.
When using scsi() syntax, the value of W depends upon Ntbootdd.SYS. Each SCSI device driver used in Windows NT has its own method of ordering controllers, although the controller order corresponds to the order that the BIOS on the controllers loads (if they are being loaded).
Additionally, if you have more than one SCSI controller, and they use different device drivers, you should only count the controllers that are controlled by Ntbootdd.sys when determining the value of the W parameter. For instance, if you have an Adaptec 2940 (which uses Aic78xx.sys) and an Adaptec 1542 (which uses Aha154x.sys), W will always be 0. What changes is the Ntbootdd.sys file: