Performance Test v5.7 (c) 1994,2007 Sergio G. Luvoni
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

(Terrific english-doc version!!)

 This test program the basic parts of the PC in how much the yield
of the hardware. It can carry a record of all the PC'S that is going proving.


Requirements:
               PC AT 286 or superior
               MSDOS 3.31 or superior
               512 Kb. memory
               * May be run on PC XT.

Use:
    TEST {options}

Where {options} can be:

    /D:  Present messages of the next accion to be executed (For debug).
    /S:  Show comparatives without performs any test.
    /L {> file}:  Shows the records of the data base
    /C:  It does not attempt to detect the type of processor and/or coprocessor
    /F:  It does not attempt to detect the processor clock frequency
    /V:  It does not attempt to detect the video adapter (BIOS PCI/AGP)
    /H, /?  :  Shows these options


Files:
              TEST.EXE  (The program)
              TEST.DAT  (PCI Database)
              TEST.DOC  (This document in spanish)
              TEST.ENG  (This document in english)
              TEST.LOG  (File where are kept the accomplished tests)


 The tests are split into three:

             1.- Test of the microprocessor and floating point unit.
             2.- Test of video system.
             3.- Test and information of the fixed disk drives.
             4.- After the foregoing, it gives a general value of merit.


1.- Test of the microprocessor and floating point unit:


  Consists of showing how rapid operates the CPU, showing the result
  in real time. To comparison manner, is presented a table with the values
  of various CPUS that they were proven with this program with precedence.
  The values were drawn taking as optimum the configuration of the
  checked equipment.

  The program is capable of detecting the type of CPU and/or FPU that are found
  installed in the system.
  It can detect Intel CPUs, AMD CPUs, Cyrix/IBM CPUs, NexGen CPUs, UMC CPUs,
  IDT WinChip, etc.

  The reference of the same is based in which, if could run in an PC of 1982
  the performance would be of 1.

  Examples : (CPU Index only)

            1  = IBM PC                     370 = Pentium 66
            3  = XT 12                      430 = Cy5x86 100
           12  = 286 20                     580 = Pentium 100
           26  = 386SX 33                   610 = Am5k86 100
           65  = 386DX 40                   690 = Cy6x86 120
           85  = 486SX 25                   790 = Pentium 150
           120 = 486DX 33                   915 = Pentium 166MMX
           180 = 486DX 50                  1090 = Pentium 200MMX
           200 = 486DX2 66                 1230 = Pentium 233MMX
           350 = 486DX4 100                2200 = PII 400


  If the proven CPU does not arrive to the logical scale values that would have to
  to have, the motives can be:

     For hardware:

          A.- Smaller clock frequency to the real required.
          B.- Too many wait states in the system memory.
          C.- Banks of decrease report quality.
          D.- CPU Cache (internal and/or external) disconnect or absent.
          E.- Bus Frequency low.
          F.- CPU "stressed".
          G.- Motherboard quality.

    For software:

          A.- Resident programs.
          B.- Shells (Norton Commander, XTree ++Gold, etc.)
          C.- Caches (Smartdrv, Ncache2, etc.).
          D.- To run the program in an environment that it will not be DOS-Flat.
              (For example Windows, OS/2, Linux, etc).
          E.- Virus.

    It is convenient for better results, to use the program from the
    commands line directly. Not to execute it in Windows(tm) or OS/2.



2.- Test of the video system.

  Here it is test the speed of the report of the video system. The test
  it is simple and according to my experience, it is quite linear with the
  performance of the video cards in various video modes. The test is in text mode (80x25)
  so much in monochromatic as the color. Keeper relationship to LANDMARK 2.0 (tm) concerning
  the results values.
  It does not use the processors of video cards (3D, MPEG, etc).
  In this way the results can be compared with any application.
  Just as previous test, there is a comparative table of different adapters.
  The @@performance of a same video card can vary according to the equipment
  where it is proven. As a rule the variation is produced in PCI/VLB/AGP cards.
  In ISA video cards generally the performance is maintained no matter where
  plugged (386 or Pentium).
  This program is capable to detect PCI/AGP video adapters.


  If the performance is less than the waited, the motives can be:

       A.- Low speed of the Bus of the system.
           ( in the BIOS, to verify the value of the BUS Clock Clock/3,
           Clock/4, etc. in ISA cards)

       B.- 16 bits-Cards operating how of 8 bits. It is amended with a
           "jumper" located in the card. To see manual of the same.
           (generally in Trident VLB and ISA video adapters)

       C.- Clock of the Local Bus Slot wrong.
           In some video cards there is a "jumper" that establishes the speed
           of the local bus according to the plate VLB that have to be. Again
           it is needed of the manual of the video and motherboard
           manufacturers.

       D.- Not optimum configuration of the PCI bus or the Cache Pipeline.
           Only it is applied if the plate is PCI. Here one must to prove with
           the options that possess the BIOS, not always it is the same to achieve
           the better performance.

       E.- Clock card low. This is very specific.
           The old Trident video cards possess a program called "SETBOARD"
           where it is configured this. To see the help of the same
           utilitarian (ISA/VLB).

       F.- Poor quality of video adapter. Not to confuse. The names of the
           video cards they are that nothing the names of the chip of the same.
           Well, one it can have for example the same chipset Trident 8900 but
           the board are of different manufacturer (Phoenix, Quadtel, etc.)

       G.- In the ancient machines, influences the CPU clock.

       H.- Defective component. Incompatibilitys. Etc.



3.- Test of the fixed disk drives.

  This test accomplishes a reading at under level of the disks installed
  in the system (until four disks, primary and secondary controllers).
  In him are read sequentially all the cylinders and randomly.
  Turned out to be it the average. The test is at physical level,
  no matter if the unit this not partitioned, or what type of files system has.
  Also it is presented a comparative table. These values can vary
  according to the equipment. A same disk can operate more rapid in a modern machine
  (for example: Ultra DMA, PIO Mode etc.).

  The accomplished test is the net data transfer, these influence
  the transfer of the bus, the access time, the rotation speed.

  Also it shows additional information on the disk drive:
  Serial number, Cache size, access time, rotation per minute,
  etc.


  It is preferable NOT TO execute it under Windows or OS/2.

  The variations of performance can be motive of:

       A.- Speed of the Bus.

       B.- Transfer manners.
           (PIO MODES 0,1,2,3 and 4. To verify which is the sustained maximum)

       C.- 32-bit mode deactivated. As long as it will be sustained.

       D.- "Block Mode" deactivated. Just as the previous, it must be
            sustained by the disk in question.

       E.- Ontrack's Disk Manager. This program is used so that the
           ancient BIOS sustain disk of more of 528 Mb. The same
           it can influence of negative or positive way.
           This variation is seen reflected when the disk is installed in
           a PCs that it support "naturally" and from must eliminate
           said program.

       F.- Wrong format of the disk, generally it is possible format
           a disk drive to a value of smaller capacity to which possesses.
           This can influence the performance. To check the data of the
           manufacturer of the disk with the values that possesses the BIOS
           of the same.

       G.- Cylinders in wrong state. Regrettably the test does not verify if
           read a bad sector. Here the BIOS is delayed a little concerning
           the reading or directly aborts the test.

       H.- Sound cards with IDE interface incorporated.
           This sometimes it is disregarded. Some cards possess a
           IDE port to permit to plug a CDROM unit in.
           Only modern BIOS accept this.
           Sometimes enters conflict with the controller of the motherboard
           (especially if possesses a double, that it sustains four units )
           and to make to operate the disks one must to deactivate all
           (Block Mode, 32 Bits, PIO, etc.). This is solved disabling
           the controller of the sound card, if it is possible, with a
           "jumper". or with a configuration program if the by software (PnP)
           If it does not possess a disconnection of the same, one must to
           make it in the multi-funtion board or in the motherboard.
           This motive is one of those that causes that DOS operate of way
           correct in an equipment, but not make it well in Windows,
           Windows 9x and is not wanted to install OS/2.

       I. - Ultra DMA Modes not installed or installed incorrectly or
            if is possessed old CD-ROM in the same controlling, provoke this
            conflicts with some of the advanced characteristics of
            the modern disks.



I wait that this program serves to draw the better performance of
their/its/your/his equipment without having to change it by one new.


For example you/them I give the obtained numbers from my old PC:

CPU          :  i486 DX4-120 Mhz. (Overcloked)
Memory       :  16 Mb. of EDO-RAM (SIMM 8Mb x 2).
MotherBoard  :  UMC 849XX Chipset, Fake Cache, 3 VL-BUS Slots, AMI-WINBIOS
Video        :  Trident 9440AGi Local Bus. 1Mb VRAM.
Multifuntion1:  Winbond Local VL-459GW Bus c/double IDE port + 1 LPT + 2 COM
Multifuntion2:  Generic 8Bit-ISA 2 COM, 1 LPT, 1Gb
Hard drive   :  NEC D3727 1.058 Gb.  4 partitions.
Other        :  CD-ROM Philips PCA 62 (6X), Sound Blaster 16 Multi-CD.


Values :

   CPU    : 425       (Very good, 355 without overclock)
   FPU    : 992       (Very Good)
   Video  : 16208     (good)
   Disk   : 1751      (Good -)
   TEST   : 416

And the numbers obtained from my following model:

CPU          :  Pentium 166MMX
Memory       :  32 Mb. EDO-RAM (SIMM 8Mb x 4)
Video        :  S3 Virge 80C325 PCI 4 Mb.VRAM.
Motherboard  :  VXPRO+,USB,UltDMA. VIA Apollo VPX Chipset, 512 Kb.Cache, AWARD
Hard drive 1 :  NEC D3727 1.058 Gb. 4 partitions.
Hard drive 2 :  QUANTUM BIGFOOT 1.2 Gb. 1 partition.
Hard drive 3 :  Seagate Medallist 4.3 Gb. 2 partitions
Other        :  CD-ROM Philips PCA 62 (6X), Sound Blaster 16 Multi-CD.


Values :

   CPU    : 915       (Normal)
   FPU    : 2090      (Normal)
   Video  : 10626     (Regular)
   Disk 1 : 1751      (Regular) **
   Disk 2 : 2099      (Regular)
   Disk 3 : 2375      (Good)
   TEST   : 690


And the numbers obtained from my other machine:

CPU          :  Pentium II 400 Mhz. 512 Kb L2-Cache
Memory       :  160 Mb. (32 + 64 + 64Mb PC100 DIMM)
Video        :  3Dfx Voodoo 3000 TV Out 16 Mb.
Motherboard  :  Soyo 6VBA 133, AWARD BIOS
Hard drive 1 :  NEC D3727 1.058 Gb. 4 partitions.
Hard drive 2 :  Seagate Medallist 4.3 Gb. 2 partitions.
Other        :  Generic 36X ATAPI CD, Sound Blaster 16 Multi-CD.

Values :

   CPU    : 2204      (Normal)
   FPU    : 5750      (Normal)
   Video  : 40037     (Very Good)
   Disk 1 : 1751      (Regular) **
   Disk 2 : 2627      (Good)
   TEST   : 1280

** Note: The values (very good, regular, etc.) are conditioned to the type of
         equipment and of the performance that would have to offer. Therefore my disk 1
         in the old equipment outweighs in the new machine.

IMPORTANT:

The values can be recorded in a file for then to make a better
comparative. For this is necessary go through all the tests.
The value Index TEST is calculated thus: (20 * CPU + 7 * Disk + (Video/100) ) /50
The numbers to the beginning of the line indicate the record of the data file.
For example: the disk "1" belongs to the same machine that video card "1"
that it is the same equipment that "1". In this case the "1" is my old equipment.


Problems:

Under Windows or OS/2 can not detect IDE units.
Under Windows or OS/2 may fail detecting CPU and clock frequency.
Older Pentium frecuency clock measurement may can fail.

Documentation and base source code:

COMPTEST 2.60. Copyright (c) 1988-1994 Norbert Juffa
PCI. Copyright (c) 1996-2001 Craig Hart


Trademarks:

i486, Pentium 166MMX(P55C),PII,PIII,IV,Xeon (c) Intel
Am486, 5x86, K5, K6x, K7 (Athlon), Duron (c) Advanced Micro Devices
Cy6x86, 6x86MX (c) Cyrix-IBM Corp.
Trident (c) Trident Microsystems
Voodoo 3 (c) 3Dfx Interactive.
NEC (c) NEC Corporation
BIGFOOT (c) Quantum
Medallist (c) Seagate
Windows, Windows 9x (c) Microsoft Corp.
OS/2 (c) IBM Corp.
Landmark System Speed Test (c) Landmark Research International
All the mentioned brands are property of their/its respective manufacturers.


Grettings:

To all that they made possible to obtain values for this program and
ordered me their/its data.

Next steps:

 * Graphics benchmarks.
 * MMX/SSE benchmarks.
 * 32/64 bits benchmarks.
 * CDROM/Floppy/Removal disk benchmarks.
 * Windoze port?


Author:

Sergio G. Luvoni <sluvoni@fi.mdp.edu.ar>

Home Page: http://www.geocities.com/sluvoni

Geronimo Costa 1442
T.E. 054-223-478-4961
Mar del Plata - (CP 7600)
Prov. of Buenos Aires - ARGENTINA.

