

				 CD-BENCH v2.0


	 Many  people think that the data transfer rate of CD-ROM  drives
      is constant from the beginning to the end of the disc, but this  is
      true  only  for old-and-slow models, the  so-called  CLV  (Constant
      Linear  Velocity) drives.  High speed drives use the CAV  (Constant
      Angular  Velocity) technology or the mixed CLV/CAV one.  The  speed
      of  CAV drives grows continuosly from the beginning to the  end  of
      the disc while in CLV/CAV drives it rises quickly up to the maximum
      and then remains constant.

	 Many common benchmarks report a completely wrong result  because
      they suppose that the reading speed is constant, so they read  only
      at  the  beginning  of the disc.  CD-BENCH, instead,  is  the  only
      benchmark that reads the disc in six different points and  displays
      six results, so you can see if and how the speed varies.
	 To  perform  a  complete  test (six readings),  you  need  a  CD
      containing at least 301500 sectors (589 MB).  Note that CAV  drives
      reach  the  maximum speed only towards the end of the  disc  (of  a
      *full* disc) so you should use the fullest disc you have.
	 If  you  use  the -S switch, CD-BENCH reports the  size  of  the
      inserted  disc  without performing any test.  Note  that  many  DOS
      programs (including Microsoft DosShell) report a wrong size.
	 By  the  -D  switch you can change the delay that precedes  each
      reading.   During  this  time the  drive's  read-ahead  buffer  (an
      internal  cache  memory) is filled with the data that  CD-BENCH  is
      going to read.  The default is -D=500 (500 milliseconds), it should
      be  enough to let your drive perform well.  With -D=0 the drive  is
      tested in the worst case.
	 By  the  -B  switch you can change the  quantity  of  data  that
      CD-BENCH  reads.   High  values reduce  the  influence  of  drive's
      internal cache so giving worse results; in other words you can  use
      this  switch to modulate the influence of caching on results.   The
      default is -B=15 (1500 sectors).

	 CD-BENCH also measures the average access time by performing 100
      random accesses to sectors located in the first 300 MB of the disc.
      Using  a fixed portion of the disc ensures that the result  depends
      only  on  the drive speed and not on the disc size.   If  the  disc
      contains less than 300 MB of data, CD-BENCH skips this test.
	 The  sequence of 100 accesses is only apparently random:  it  is
      always  the  same, every time you perform the test;  in  this  way,
      again, the result depends only on the drive speed.

	 You  can  effect  only  the  first  test  (data  transfer   rate
      evaluation) or only the second one (access time evaluation) by  the
      -1 and -2 switches.

	 CD-BENCH can also be used to verify whether a disc is  defective
      (that  is whether it causes read errors or not): by the -A  switch,
      in  fact,  you  get CD-BENCH to read the  whole  disc.   Of  course
      this task takes several minutes.
	 IMPORTANT NOTE.   Many drivers have  a  little bug  that  causes
      CD-BENCH  to read farther the end of the disc so generating a  read
      error.   This  problem takes its origin from a wrong formula  in  a
      Microsoft's  official  document;  a more  recent  version  of  that
      document is correct but many drivers are still buggy (this includes
      Windows 95's internal driver!).

	 CD-BENCH  interacts  with  the  8253  timer  chip  to  get  high
      resolution timing (0.1 ms) and, therefore, highly accurate results.

	 The  presence  of  a  disk cache like  SMARTDrive  can  lead  to
      unrealistic  results.   CD-BENCH  temporarily  disables  SMARTDrive
      v4.0+  or  compatible caches (PC-Cache v8.0+ and  the  most  recent
      versions  of Norton Cache are known to be compatible) and  restores
      their  original status before returning to DOS, even if  you  abort
      the  program by pressing Ctrl-Break or ejecting the disc  (I  don't
      lock them so you are free to do as you like).

	 If no parameter is supplied on the command line, CD-BENCH  tests
      all  the available CD-ROM drives; if you want to test only  one  of
      them, just specify its letter.
	 IMPORTANT  NOTE.  If you didn't install both MSCDEX.EXE  and   a
      proper device driver for your CD-ROM drive, CD-BENCH can't see  the
      drive and displays the message "No CD-ROM drive found".

	 CD-BENCH  cannot  run as a DOS task of Windows 95 or  OS/2:  you
      have to use "plain" DOS.


				   --=<>=--


      How to get good results.

      1)  Use a clean, unscratched, "silver" disc.  CD-Recordable  weakly
      reflect  the  laser beam so they are harder to read.  When  a  read
      error  occurs  the drive repeats the read operation twice  or  more
      times  and can also reduce the spin (some models only).  Of  course
      this behaviour influences negatively the result.

      2) Some discs are unbalanced and cause strong vibrations when  they
      rotate  at  high speed.  Some drives detect  these  vibrations  and
      reduce  the spin, so, if the first run doesn't give  good  results,
      try to use another disc.  BTW: never attach labels on CD-ROMs!

      3)  Visit  the  web site of your  drive's  manufacturer  and  check
      whether a new version of the driver is available.

      4)  Some old BIOS is programmed in the way that it  recognizes  any
      CD-ROM drive as a "PIO Mode 0" device.  This is not a good thing if
      you  have  a  very  fast drive (20x or more)  because  PIO  mode  0
      actually  holds your drive back.  If this is your case, you  should
      contact your mainboard vendor for a BIOS upgrade so that your  high
      speed drive will be recognized as a PIO mode 4 device.


				   --=<>=--


	 I  thank  the  persons that have tested a  pre-release  of  this
      program  on  drives different from my one so giving  me  a  general
      survey  of the market.  They are Andrea Campagnolo,  Adriano  Cata-
      pano,  Giacomo Cocchella, Vittorio Colosio, Alessandro  Del  Rosso,
      Andrea Galli, Allan Jong, Patrick  Lee, Gaetano Petronzio, Alessan-
      dro Queri, Cicci Serra.

	 CD-BENCH  is a copyrighted free program.  It can be freely  used
      and  distributed  but  cannot be  bundled  with  hardware  products
      (drives, CD recorders, etc.) without my written permission.

	 Your feedback is welcome.


      My addresses are:

      E-mail:  pnavato@poboxes.com
	       pnavato@geocities.com
	       Pino Navato, 2:335/225.18  (The Bits BBS, Fidonet)

      WWW:     www.poboxes.com/pnavato
	       (currently forwards to
		 www.geocities.com/SiliconValley/4421)

      Mail:    Pino Navato
	       Via Pittore, 164
	       80046 S.Giorgio a Cremano (NA)
	       ITALY






			   *	 *     *     *	   *
