Results of running SPEEDCHK on a RAMDrive with varying numbers of buffers 
and several SMARTDRiVe configurations. The RAMDrive itself is uncached.
N.B.: These times do not include deletion. Deletion times were within
      the error range of the test program. SPEEDCHK is not consistant 
      across sessions or individual runs within a session.
      System: MS-DOS 6.22; Caviar 1210 EIDE HD with 2KB internal cache, 
              989 cyl, 12 hds, 36 spt, 80% full and reasonably well defragged 
              and compacted; DFI 3486-UCF mobo, 1/8MB L2 cache; i486DX-50 CPU;
              RAMDRIVE.SYS loaded high, 10MB size, nearly empty

                               Smartdrive cache size
    BUFFERS        2MB.......1MB.......3/4MB.....1/2MB.....REMed 
       |
     REMed ........8.........8.........8.........8.........8   
       
       1   ........13........13........13........13........12
       5   ........12........11........11........11........10
       7   ........11........10........11........11........9
       10  ........10......................................9
       
       14  ........9.......................................8
       15  ........8.......................................8
       
       30  ........7.......................................7
       60  ........8.......................................7

  It seems that the BUFFERS command affects RAMDrives in the same manner as 
hard drives. REMing the SMARTDRV command improves times slightly; is 
MS-DOS 6.22 resetting it's virtual heads, crossing cache-occupied RAM in the 
process? ;-)