DISKTIMER GS v2.0 (C) 1988-89 Joe Jaworski All Rights Reserved. DisktimerGS is FREEWARE. You may distribute it to anyone without charge. There is no Shareware fee required. However, the program is copyright material and is not to be commercially distributed without prior permission. DiskTimerGS and SCSI HACKER with complete instructions is also available on disk from So What Software, 10221 Slater Avenue, Suite 103, Fountain Valley, CA 92708. (714) 964-4298 Version 2.0 Update Information 1) Version 2.0 now includes Multi-Block READs. This test can be used to optimize interleave of your SCSI drive for maximum performance under GS/OS 5.0. The test performs 16 2000-block READ commands sequentially across the disk starting with block 0000. The times reported are in seconds. 2) Minimum ProDOS volume size to run version 2.0 is 16MB. Needless to say, DiskTimerGS cannot be run from floppies or hard disks partitioned less than 16MB in size. DiskTimerGS MUST be launched from the volume or disk to test. 3) Interrupts are now enabled during all tests. For TransWarpGS owners, this allows you to maximize interleave while your machine is accelerated. This also means that moving the mouse, pressing keys, or accessing the Control Panel during the test will give you dirty results. Any Permanent Init files that hog interrupts on a frequent basis will also distort the numbers. 4) If used in conjunction with THE SCSI HACKER, use the Advanced Disk Utility from System 4.0 (not 5.0) to write partitions, then run DisktimerGS under 5.0. This is because the 5.0 native mode SCSI driver forces you to select 1:1 or 2:1 interleave ONLY, and then low-level formats your drive. As always, use HDSCPartition after low-level formatting with SCSI HACKER. 5) If you cannot see a difference in Multi-Block READ results under changing interleave factors, chances are your SCSI drive has either Read-Ahead buffering or built-in caching. As a general rule, 6:1 interleave (5:1 with TransWarpGS) yields good results with many hard drives under 5.0. With full cylinder caching, 1:1 will give best results. 6) All other tests remain the same as version 1.0.