ftp.pasteur.org/FAQ//ftp-pasteur-org-FAQ.zip/FAQ/pc-hardware-faq/video/part2

Internet Message Format | 1997-06-24 | 63.2 KB

Path: senator-bedfellow.mit.edu!bloom-beacon.mit.edu!news.moneng.mei.com!sol.ctr.columbia.edu!news.uoregon.edu!news.rediris.es!newsfeed.mad.ibernet.es!News.Amsterdam.UnisourceCS!newsgate.unisource.nl!xs4all!news.radio.cz!nntprelay.mathworks.com!news.mathworks.com!howland.erols.net!cs.utexas.edu!utnut!kwon!watserv3.uwaterloo.ca!newshost.uwo.ca!jones.heart.rri.uwo.ca!mjscott
From: [e-mail] (Michael Scott)
Newsgroups: comp.sys.ibm.pc.hardware.video,comp.answers,news.answers
Subject: comp.sys.ibm.pc.hardware.video FAQ, Part 2/4
Followup-To: poster
Date: 11 Jun 1997 17:15:06 GMT
Organization: The University of Western Ontario, London, Ont. Canada
Lines: 1249
Approved: [e-mail]
Expires: 21 Jul 1997 00:00:00 GMT
Message-ID: <5nmmeq$[e-mail]>
Reply-To: [e-mail] (Michael Scott)
NNTP-Posting-Host: jones.heart.rri.uwo.ca
Summary: This is a monthly posting containing a list of Frequently
         Asked Questions (and their answers) pertaining to video
         hardware for IBM PC clones.  It should be read by anyone who
         wishes to post to the comp.sys.ibm.pc.hardware.video
         newsgroup.
Originator: [e-mail]
Xref: senator-bedfellow.mit.edu comp.sys.ibm.pc.hardware.video:177478 comp.answers:26697 news.answers:105664

Archive-name: pc-hardware-faq/video/part2
Posting-Frequency: monthly (second Monday)
Last-modified: 1997/02/19
Version: 1.0
URL: http://www.heartlab.rri.uwo.ca/vidfaq/videofaq.html

**********************************************************************
COMP.SYS.IBM.PC.HARDWARE.VIDEO Frequently Asked Questions - Part 2/4
**********************************************************************

Q) What is a shadow mask?

   Monitors work by aiming a beam of electrons at a blob of phosphor,
which in turn glows.  This glow is what we perceive as a pixel on the
screen.  Your standard colour monitor has three dots (dot triad) at
each location on the screen; red, green and blue.  There is a corresponding
electron gun for each colour which emits an electron beam of varying
intensity - this corresponds to colour brightness.  To ensure that the
electrons from each gun strike the corresponding phosphor, a 'shadow mask'
is used.  Because the three electron beams arrive at slightly different
angles (from the three separate electron guns), it is possible to construct
and align the shadow mask such that the electron beam from one gun will
strike the correct phosphor dot, but the other two phosphors will be in
shadow.  This way, the intensity of red, green and blue can be separately
controlled at each dot triad location.  The shadow mask is usually an invar
mask (64% iron & 36% nickel) which is a thin plate with small holes punched
in it.  Only about 20-30% of the electron beam actually passes through the
holes in the mask and hits the screen phosphor, so the rest of the energy
is dissipated as heat from the mask.  As a result, shadow mask monitors are
prone to colour purity problems as they heat up due to slight shifts in the
position of the holes relative to the phosphor dots.  Shadow masks - or
their equivalent - have made mass production of CRT's possible.



Q) Why does my monitor have 1/2/3 faint horizontal lines on it?

Your monitor likely uses a Sony Trinitron picture tube.  Trinitron 1
tubes can be recognized because they are curved only in the horizontal
plane, but are flat vertically.  Typically, the number of lines seen
depends on the monitor size:
        < 17"   1 line
        17-21"  2 lines
        > 21"   3 lines
Because of the technical nature of how CRT's work, few people understand
the details of how they operate.  As such, many laymen have viewed Sony's
Trinitron design as being a proprietary black box because they don't
understand Sony's technical documents.  A fairly well accepted description
of the way these tubes work follows.

For a description of how a standard shadow mask CRT works, see "What is a
shadow mask?".

   Sony's Trinitron design uses a variation of the shadow mask
called the aperture grill (or guard grill).  Rows of very fine
metal strips run vertically down the screen, separating columns of
coloured phosphor which are arranged in alternating stripes of red,
green and blue.  This configuration allows the phosphor strips to
be placed closer together than conventional dot triads, and the
fine vertical wires block less of the electron beam than traditional
shadow masks, resulting in a brighter image, and less thermal
buildup and distortion.  Coupled with changes in the way that the
electron guns are arranged, this design results in a crisp, bright
image.  However, the vertical strips are so fine that they can be
set into motion when contacted by the electron beam (thermal
changes) .  This would result in a shimmer on the screen which would
be quite distracting.  To remedy this, Sony puts horizontal
stabilizing wires across the vertical ones.  This reduces shimmer,
but results in one or more fine horizontal lines being visible on
the monitor.  As mentioned above, the number of lines increases with
monitor size.  Usually, these lines are only visible to a discerning
viewer when looking at a bright, solid background.  If you're an
experimentalist, try gently smacking the side of a Trinitron monitor,
and look to see the wires shimmer.  Basically, the horizontal lines
are a minor trade-off when compared to the superior brightness
and vertical flatness of the Trinitron screen.

What follows is a statement from Sony about their aperture grille
design:

     Since its introduction in 1968, Sony has produced more than 70
million CRT's.  Award-winning Trinitron CRT's are used in a
multitude of applications, including high resolution displays for
the computer industry.  Today, Trinitron CRT's are used by Sony and
other leading manufacturers to meet the ever increasing demands and
expectations of computer users throughout the world.

     One of the unique features of the Trinitron CRT is what is
called the Aperture Grille.  An Aperture Grille consists of a
series of long vertical slits fastened with strong vertical tension
to a steel supporting frame.  Electron beams pass through the
Aperture Grill to illuminate phosphor on the faceplate.  The
vertical tension of the Aperture Grille absorbs any thermal
expansion, thus eliminating the problem of doming or color spill
and resulting in a superior picture quality.

     Since the CRT requires a vacuum to function, a damper wire
which is approximately 15 microns in diameter is strategically
placed on the Aperture Grille to reduce susceptibility to
resonance.  The "line" that some customers see on the screen is not
a fault but the damper wire which has always been an integral part
of our Trinitron technology.

     We hope that our customers will continue to consider the
overall attributes and excellence of the Trinitron system when
evaluating our products.



Q) What's the difference between fixed frequency and
multisynchronous monitors?
[From: Michael Scott ([e-mail]) and Bill Nott
([e-mail])]

There are two primary measures of the maximum effective pixel
addressability and refresh rate that a monitor is capable of.  The
maximum rate that a monitor can refresh the screen is measured in
Hertz (cycles/second) and is called the vertical refresh rate (or
vertical scan rate).  The horizontal scan rate is the number of times
that the monitor can move the electron beam horizontally across the
screen, then back to the beginning of the next scan line in one
second.  Most early analog monitors were fixed frequency, meaning
that they were intended to work only at one specific vertical refresh
rate (often 60 Hz) and one horizontal rate (often this is expressed
as a number of pixels, but this isn't really the same).  Most older
SUN, SGI and other workstation monitors were of this type.  Generally,
these monitors are limited in their applications, since they require
that the incoming video signal falls within narrow timing
specifications.

These type monitors also typically use composite video signals (with
sync on Green), so are not compatible with most of today's PC graphics
controllers. Also note that even if the composite video signal issue is
overcome, there are additional issues related to attempting to use such
monitors with a PC. Among these are DOS text mode support, and radiated
emissions compliance. See "How can I get a fixed frequency (RGB) monitor
to work on my PC?" below.

In part due to the desire to produce more flexible monitors (i.e. fewer
different models), the lack of PC SVGA/EVGA/etc video standards, and
in part due to recognition of an emerging trend toward higher pixel
addressability formats within the computer industry, along with a desire
to provide an upward migration path for new customers, vendors started
to produce monitors capable of syncing to video signals within a range
of frequencies.  Such monitors are called multisychronous, or Multisync.
Multisync is actually a trademark of NEC's, though it has become a
generic term for a monitor which is capable of syncing to more than one
video frequency.  The meaning of multisynchronous has become somewhat
muddled.  To truly be multisynchronous, a monitor should be able to sync
to any frequency of incoming video signal (within reason, of course).
However, many so-called multisynchronous monitors can only sync to a
number of discrete frequencies (usually 3 or 4).

If the video signal supplied to such a monitor is within the range of
it's deflection circuits, the image will be displayed; otherwise, the
image may be either not synchronized, or completely blanked. It is also
possible to harm some monitors of this type by applying a video signal
outside it's ranges, if protective measures were not put into place by
the design. Thus, such a monitor will usually operate at the most common
video modes, but may not operate at less common modes. This type of
monitor may be referred to as a 'banded' design. A continuous frequency
design should operate at any frequency within the specified range.



Q) How can I get a fixed frequency (RGB) monitor to work on my PC?
[From: Michael Scott ([e-mail]) and Bill Nott
([e-mail])]

There are plenty of old RGB monitors (possibly from old Sun, SGI, or
other workstations) around which are attached to outdated or non-
functional machines.  Most of these units are quality products by Sony,
Hitachi or other vendors.  You want an easy way to connect your VGA or
better card to the monitor.  It's may not be that easy, since many of
these monitors are only capable of displayed non-standard pixel
addressabilities, but read on..

The easiest solution (but not necessarily the cheapest) is a commercial
solution.  See the section from Declan Hughes, below.

Most of the old RGB monitors are fixed frequency, meaning that they are
intended to work at only specific horizontal and vertical scan rates.
This is in contrast to many newer models which are variations on the
multi-sync theme.  Multi-sync means that the monitor can sync to a
_range_ of scan rates, or a number of discrete scan rates, based on the
incoming video signal.  i.e. the monitor will detect the scan rates of
the video signal, and switch to the closest scan rate it is capable of.
Since a fixed frequency monitor can't do this, you have to make sure
that the video signal your video card is generating is compatible with
the monitor.

To hook up a VGA card to a fixed frequency monitor requires three things:

1)  A cable that connects the VGA output to the RGB (and possibly sync)
on the monitor.  This may be the easiest part.  Fixed frequency monitors
typically have BNC connections for video input, so you need a cable which
connects to your computer video output (typically a 15 pin D-Sub VGA) at
one end, and which has 3, 4, or 5 BNC's at the other end. The number of
BNC's depends on how you plan to resolve the sync signal issue
(see point 3).

2)  The horizontal and vertical scan rates must be compatible.  Some
video cards have adjustable scan rates, and if you can adjust the video
signal to be within the range that the monitor can handle, you might be
in business.  If you can't get the generated video signal scan rates to
within the monitor's specs, you need a scan converter, which is a very
expensive and complex device (read: not practical).  Even if you can
get your graphics controller to adjust to the monitor's unique frequencies,
you will have to figure out what to do when you first boot your PC, and how
to run DOS programs, if needed.  Just about every PC boots up in the DOS
character mode (720 by 400 pixels, 31.5 kHz Horizontal, 70 Hz Vertical).
No fixed frequency monitor will operate at this mode. If you never need to
use DOS, you may be able to set your autoexec.bat to start Windows
immediately upon boot-up.

3)  You must have a compatible electronic signal.  The problem is this:
VGA cards have separate channels for red, green, blue, vertical sync
and horizontal sync.  Most RGB monitors have 3 or 4 connectors, either
red, green (with sync) and blue, or red, green, blue and a separate
sync.  The sync signals from the VGA card must be combined to be fed
into the monitor.  This is not as simple as soldering the horizontal,
vertical sync and green wires together.  Some folks have been able to
get their monitors to work by building simple circuits.  However, keep
in mind that items 1 and 2 above _must_ also be satisfied.   One such
circuit that has been suggested is in Appendix C.  Unfortunately, I
don't know who designed this circuit, so I can't give credit but it's
part of the Apollo FAQ.  It is presented as-is, and there are no
guarantees that it will do what you need.  Be warned, if you don't have
considerable electronics experience you shouldn't attempt this.  If you
blow your monitor or video card, don't come crying to me.

[From: Declan Hughes ([e-mail]) and Michael Scott
([e-mail])]

A frequently asked question is how to use a fixed frequency monitor
(often a Sony or Sun monitor) with a PCAT. Some of the companies that
provide the required video cards are:

 1. Mirage Computer Systems
    4286 Lincoln Blvd.,
    Marina Del Rey, CA 90292
    tel: [phone]
    tel: [phone]
    fax: [phone]
    http://www.mirage-mmc.com/
    ftp://ftp.mirage-mmc.com/
    Contact: Emil Darmo ([e-mail])

 2. Software Integrators
    104 East Main st.,
    Suite 206,
    Bozeman, MT 59715
    tel: [phone]
    tel: [phone]
    fax: [phone]
    http://www.si87.com/

 3. PCG Corp. (Photon)
    tel: [phone]
    tel: [phone]
    fax: [phone]
    http://www.photonweb.com/
    ftp://ftp.photonweb.com/
    email: [e-mail]

 4. MaxVision Corporation
    2705 Artie Street, Suite 27
    Huntsville, Alabama  35805
    tel: [phone]
    http://www.maxvision.com/
    ftp://ftp.maxvision.com
    email: [e-mail]

 5. STB
    http://www.stb.com
    Some cards have limited fixed frequency support.

Mirage make video cards that support all single frequency/high
frequency monitors that operate between 28-35Khz, 47-52Khz, 60-65Khz
and 70-78Khz at specific VGA, EGA and DOS modes (various drivers are
included) with ISA, Vesa local bus & PCI local bus interfaces. They
also have fast drivers for specific software products such as Autocad,
3D Studio and Windows 3.1 etc.

They have four models which use popular video chipsets, providing a
wide selection of processors and buses.
Drivers are available for Windows 3.1, Windows 95, OS/2 Warp, ACAD,
Linux, lots of CAD programs & VESA compatible applications.
All cards support separate sync and composite sync as well as sync
on green.  Software comes with the cards to adjust frequency, setup
and centering.

*** Mirage will give you a special discount if you mention that you
were referred from this FAQ  (No, I don't get anything in return :-( ).

For example, a STORM 1280/256 will drive a Sony GDM-1950 at 640x480,
800x600, 1024x768, 1280x1024 and DOS modes (this monitor is rated at
63.34Khz Horizontal sync. and the card runs at 64Khz Horizontal
sync.). This card uses an S3 graphics accelerator. See also PC
Magazine/April/13/1993.

Software Integrators make similar video cards (the MERCURY X1 Series)
that will also support CGA modes as well as DOS, EGA and VGA modes
using the S3 801 graphics accelerator and again they also make fast
drivers for specific products such as Autocad, 3D Studio and Windows
3.1 etc.

These cards work with all fixed scan monitors including, IBM,
Mitsubishi, Hitachi, Sony, Sun, HP, Verticom, DEC, Taxan, Philips,
Apollo, Silicon Graphics, Intergraph, Aydin, Amtron, Monotronix,
etc. and will replace old boards from, Artist, Number9, Nth,
Verticom, Photon, BNW, VMI, Matrox, Metheus, Mirage, Graphax,
Imagraph, TAT etc.

PCG provides several models, supporting ISA, VLB or PCI buses and their
cards can support 1, 2 or 4 MB of video RAM.  CL-5434 processors appear
to power all of the Photon cards.  They support sync-on-green (RGB)
monitors, Windows95, Win 3.1, Linux and provide a variety of video
drivers for various applications.  Cards automatically boot up with
a frequency compatible with your monitor.

I was also informed of a German manufacturer ELSA that makes similar
cards, but I do not know of their address or product range.

All companies can supply interface cables such as a 15 pin male VGA to
5 BNC connector.

MaxVision makes two cards, both based on Weitek P9x00 processors.
These are high-end cards that can drive monitors at up to 2048x1536!
MaxVision claims that their custom signal processing chip generates
high quality, high resolution video from low res. VGA modes.  Two
MaxVision adapters can be installed in one machine to have a two-
headed display.  Models are available with 1-4 Megabytes of VRAM.
Windows NT and Windows 3.x drivers are currently available, but
considering that Weitek has dropped support for Win95 and has stated
that they will not provide P9x00 drivers, it's possible that Win95
support will not become available for these cards.

For more information, refer to the Fixed Frequency PC Video FAQ at:
http://www.devo.com/video, Appendix C of this FAQ and the Fixed
Frequency circuits on the PC Video FAQ WWW site.

[From: Bill Nott ([e-mail])]
Nobody has dealt with the radiated emissions issue. All PC's are
marketed as FCC class B products. I recently learned that Mirage labels
their cards as Class B. However, most of the single frequency
workstation monitors becoming available are FCC Class A, and the FCC
rules say that attaching a Class A device to a Class B system degrades
it to a Class A system. This in itself is not necessarily a problem,
unless the system is actually found by the FCC to be causing radio
interference. My recollection is that the process of remedy for a
Class A system is more severe than for a Class B system, so users may
be opening themselves up to a higher level of risk.  Comments are
welcomed on this.



Q) What is a low emission monitor?

Generally, low emission monitors follow the Swedish government's
SWEDAC (Swedish Board for Technical Accreditation) MPR II spec's or
the stricter TCO (Swedish Confederation of Professional Employees)
standards.

Both limit the emissions of VLF (Very Low Frequency) and ELF (Extremely
Low Frequency) electric and magnetic fields.  Although new monitor
technology generates less radiation than older units, additional active
and passive shielding mechanisms are installed to further reduce emissions.
The majority of monitors produced now fall within the MPR II spec's, though
some still do not.  The price differential between regular and low-
emission units has declined substantially.

The basic specs are as follows:
[From www.noradcorp.com/swedish.htm]

Frequency Range                         MPR II          TCO

Electric Fields

Static Field                            +/-500 V        +/-500 V
ELF   5 Hz - 2 KHz (Band I)

Q) What does DPMS mean?

Display Power Management System, correctly referred to as VESA DPMS as
it is a VESA standard.  This is the replacement for screen savers.
Screen savers either blank the monitor, or continually change the images
painted on the screen.  They do this because prolonged continuous exposure
of the screen phosphor to an electron beam can cause degradation of the
phosphor.  This is especially true of older monitors, but can occur in
newer ones.  If the phosphor becomes damaged from prolonged exposure to
the same screen image (a login screen or your Windows background)
permanent ghosting patterns may result.  DPMS does one step better than
a screen saver and actually saves energy too by 1) turning off the video
signal to your screen after a period of inactivity which puts most newer
monitors into a standby mode or 2) shutting your monitor off.  For some
new monitors, 1 & 2 are the same, as these monitors are in a low-power
standby mode even when apparently turned off.

DPMS is often implemented with CPU power saving (i.e. Intel 486SL
enhanced CPU's) and hard drive power-down circuitry.  They are all
based on the same idea:  when a device is idle, slow it down or
shut it off entirely to save energy.  Sometimes this can be an
inconvenience, since the computer may be busy performing some operation
without any keyboard, mouse or screen activity.



Q) How can I maximize the life of my monitor?
[From: Sam Goldwasser ([e-mail])]

Monitor Life, Energy Conservation, and Laziness:
-----------------------------------------------

A common misconception about the care and feeding of computer monitors is
that they should be left on all the time.  While there are some advantages to
this, there are many more disadvantages:

1. CRT Life: The life of a monitor is determined by the life of the CRT.
   The CRT is by far the most expensive single part and it is usually not
   worth repairing a monitor in which the CRT requires replacement.
   The brightness half-life of a CRT is usually about 10-15 K hours of on time
   independent of what is being displayed on the screen.  10 K hours
   is only a little more than a year.  By not turning the monitor off at
   night, you are reducing the life of the monitor by a factor of 2-3.
   Screen savers do not make any substantial difference especially with
   modern displays using X-Windows or MS Windows where the screen layout is
   not fixed.  With video display terminals, the text always came up in the
   same position and eventually burned impressions into the screen phosphor.

2. Component life: The heat generated inside a monitor tends to dry out parts
   like electrolytic capacitors thus shortening their life.  These effects
   are particularly severe at night during the summer when the air
   conditioning may be off but it is still a consideration year around.
   Note that the claim about electrolytic capacitors needing to
   used frequently only applies on a time scale of years, not hours.

3. Safety:  While electronic equipment designed and manufactured in accordance
   with the National Electrical Codes is very safe, there is always a small
   risk of catastrophic failure resulting in a fire.  With no one around,
   even with sprinklers and smoke alarms, such an failure could be much
   more disastrous.

4. Energy use:  While modern monitors use a lot less energy than their
   older cousins, the aggregate energy usage is not something to be ignored.
   A typical monitor uses between 60 and 200 Watts.  Thus at a $.10 per KWH
   electric rate such a monitor will cost between $48 and $160 a year
   for electricity.  During the night, 1/2 to 2/3 of this is wasted for
   every monitor that is left on.  If air conditioning is on during the
   night, then there is the additional energy usage needed to remove this
   heat as well - probably about half the cost of the electricity to run
   the monitor.

The popular rationalization for what is most often just laziness is that
power-on is a stressful time for any electronic device and reducing the
number of power cycles will prolong the life of the monitor.  With a properly
designed monitor, this is rarely an issue.  Can you recall the last time
a monitor blew up when it was turned on?  The other argument, which has more
basis in reality is that the thermal cycling resulting from turning a monitor
on and off will shorten its life.  It is true that such thermal stress can
contribute to various kinds of failures due to bad solder connections.
However, these can be easily repaired and do not effect the monitor's
heart - the CRT.  You wouldn't leave your TV on 24 hours a day, would you?

Some of the newest ('green') monitors have energy conserving capabilities.
However, it is necessary for the software to trigger these power reduction or
power down modes.  Few monitors in actual use and fewer workstations or PCs
are set up to support these features.  If you have such a monitor and computer
to support it, by all means set up the necessary power off/power down timers.
However, using the power saving modes of a 'green' PC with an older monitor
can potentially cause damage since some of the modes disable the sync signals.
A 'green' monitor which can detect a blank screen and use this as a trigger
can easily be used with a screen saver which can be set to display a blank
screen - on any PC or workstation.

My recommendation is at the very least to turn your monitor off at night.
Turning it off if you are not going to be using it for an hour or two is
fine as well.  This will extend the life of the monitor (and your investment)
and is good for the environment as well.

For workstations, there are good reasons to leave the system unit and
peripherals on all the time.  However, the monitor should be turned off
using its power switch.  For PCs, my recommendation is that the entire unit
be turned off at night since the boot process is very quick and PCs are
generally not required to be accessible over a network 24 hours a day.



Q) Is it important to use a screen saver?
[From: Sam Goldwasser ([e-mail])]

The quick answer is: it is not as important as it once was with
fixed format text displays.  I still recommend using a screen saver
as cheap insurance if it uses mostly dark colors or black.  In this
case, your power dissipation and heating will be slightly reduced
as well.  However, even completely blanking the screen with the video
signal does not significantly prolong the life of the CRT. (See the
section on Monitor Life.)

The best solution is to use a 'Green', monitor in its
full power down mode.  This saves energy and wear and tear on the
CRT and other monitor components.  Most newer PC BIOS's now support
energy saving modes in the CMOS setup.  These should be used but
only if your monitor specifies that it supports the relevant energy
saving modes.  Otherwise, you could damage the monitor.



Q) Should I be concerned about monitor emissions?
[From: Michael Scott ([e-mail]) and Bill Nott
([e-mail])]

In the November 8, 1996 issue of Science (Vol. 274, pg. 910) is an
announcement that the US National Research Council "seemed to deal a
mortal blow to one of the most polarised and long-running environmental
controversies -- whether electromagnetic fields (EMF's) from power lines
or household appliances pose a threat to human health.  After an
exhaustive, 3-year study, a 16-member panel said there is 'no conclusive
and consistent evidence' that ordinary exposure to EMF's causes cancer,
neurobehavioral problems, or reproductive and developmental disorders."
For those interested in what this is all about, a brief summary follows.

CRT's (Cathode Ray Tubes) direct a beam of electrons at a thin layer of
phosphor which coats the screen on your monitor.  When the electrons
strike the phosphor, shadow mask and other screen components, x-rays are
produced.  The amount and energy of the x-rays depends on the
accelerating voltage.  The relatively low voltages in CRT's (compared to
commercial x-ray machines) means that relatively low quantities of low
energy x-rays are produced and modern monitors are so well shielded, that
there is no concern of being irradiated over time.  Though it is possible
for a damaged monitor to emit x-ray radiation, it is unlikely that
harmful amounts will be released, and most x-rays would be directed
towards the back or sides of the monitor.  Any damage to the front of
the CRT severe enough to increase x-ray emission would cause the CRT to
implode.

All televisions and computer monitors must comply with various worldwide
standards for ionizing emissions. Information relating to this compliance
is typically included on the product label, or within the users manual.

Recently, concerns about low frequency (LF) and very low frequency (VLF)
electro-magnetic and electro-static emissions have been raised.  Many
studies have been established recently to determine if these concerns
are warranted.  None of the studies has concluded that there is any
correlation between the radiation and possible health risks. In Sweden,
a large study was undertaken and as a result, the Swedish government,
and the Swedish Workers Union (TCO) both established recommended limits
of radiation for office equipment, including Video Display Terminals
(VDT's). The same limits are applied to monitors; the Swedish Government
standard is referred to as MPR 1992, and the TCO standard is referred
to as TCO.  Many new monitors adhere to the Swedish emission regulations.

Epidemiologists have suggested that the risk factors for some childhood
cancers (particularly leukemia) are as high as two for some populations
exposed to low frequency EMI.  A risk factor of two means that the odds
of being afflicted with a disease is twice as likely in the exposed
population than in a control population.  In general, a risk factor of
less than six is not considered significant (cigarette smoking has a risk
factor of 10-20).  As a result, several groups have publicly stated that
there is no significant health risk from EMI radiation levels experienced
by people from home appliances or nearby high voltage lines.

Critics of the Swedish study suggest that it was simply too huge.
According to a television documentary, over 800 comparisons were made
for correlation between exposure and pathologies.  Statistics would
suggest that given enough completely random and uncorrelated measures,
the odds are that some of them will display a high correlation.  As a
result, any study that is large enough will produce correlations
between _some_ of the measured quantities.  Because of this criticism,
and the fact that only correlation, and no causation was proved in the
study, the Swedish government has since reversed their decision to
mandate maximum EMI emissions.

Studies in the U.S. to determine if EMI could cause cancer or other
illness, birth defects or any other health problems in rats have
come up negative.  The rats were exposed to 0-10000 mG (milli-Gauss)
magnetic fields (the earth's magnetic field is ~500 mG), and their
skeletal and visceral organs, reproduction, frequency of cancer and
immunology all came up normal.  The chronic studies that were
undertaken by the same group will be completed in 1996.  Other studies
showed that EMI had no effect on the growth of cancer cells.

So, you have to make your own decisions, but the overwhelming majority
of experts agree that there is no cause for concern.

For more information, contact:

EMF INFOLINE
Tel: [phone] (in the USA)
In Washington, DC, call: 484-1803



Q) How do I calculate the minimum bandwidth required for a monitor?

[Paraphrased from Richard Trueman ([e-mail])  by Michael Scott
([e-mail]) ]

The bandwidth is a measure of the amount of data that a monitor can
handle in one second.  It is measured in MHz.  The maximum bandwidth
of a monitor should be matched as closely as possible to the dot clock
of the video controller.  If there is a mismatch, then capacity of
either the controller or monitor may be wasted.  It is not as serious
for the monitor to lack video bandwidth as it is for a graphics
controller to lack the dot clock rate needed for a given video mode.
The maximum bandwidth of a monitor cannot be directly calculated
without detailed timing information, but often this information is
provided by the manufacturer.  In fact, the exact bandwidth required
in a monitor at a given pixel addressability and vertical refresh
frequency is also dependent on internal timing of the monitor itself.
To calculate an approximation of the required bandwidth for a given
pixel addressability and vertical refresh frequency: [This
approximation tends to overestimate the actual bandwidth frequency
from my experience -Mike]

The bandwidth is dependent on the number of vertical and horizontal
pixels and the vertical refresh rate.  This approximation grossly
simplifies the calculation:

Given that the vertical pixel addressability is Y, horizontal pixel
addressability is X and refresh rate is R:

To account for the additional time required for the vertical blanking
interval, Y is multiplied by 1.05.  The additional time required for
the horizontal blanking interval is about 30% of the scan time, so use
1.3X.  Note that 30% is very conservative with most new monitors.  In
order to do an exact calculation, you would have to know the vertical
and horizontal blanking intervals for the mode in question, as well
as the horizontal scan frequency.  So the resulting approximation is:

bandwidth = 1.05Y*1.3X*R

i.e. for 1280x1024 at 60 Hz,

approx. bandwidth required = 1.05*1024*1.3*1280*60 = 107 MHz

Clearly these are gross simplifications, so use this equation for
approximations only.



Q) How do I calculate how much VRAM/DRAM I need?

This discussion only deals with calculating the minimum amount of RAM
you will require _on your video card_ and is not related to main system
RAM.  The following calculations will tell you the minimum amount of
RAM necessary, but some video cards do not use all of their RAM for
the frame buffer (area that stores screen information).  In particular,
some Windows accelerator cards use some of their memory to store font
or other graphical information.  As a result, some cards with 2 Megs
of video memory will not be able to display the higher pixel
addressabilities and colour depths that you might expect.

There are two things that have to be decided in order to determine how
much video RAM is required for a given pixel addressability.  The first
is the screen addressability in pixels and the second is the colour
depth in bits.  Before you go out and purchase a video card and/or
extra RAM, make sure that the card is capable of the pixel
addressability and number of colours that you want.  Often cards are
advertised as 1280x1024 and up to 16.7 million colours, _not_ 1280x1024
_at_ 16.7 million colours.

Standard pixel addressabilities available are:
640x480, 800x600, 1024x768, 1280x1024 & 1600x1200

Less commonly, 1152x864 and 1600x1280 are supported.

For an idea of pixel addressabilities appropriate for your monitor, see
"What pixel addressabilities are best for my monitor".
Colour depth information is provided in "How does colour depth relate
to the number of colours?".

To calculate the amount of video memory you need, simply multiply:

(horizontal addressability) * (vertical addressability) * (pixel depth)/8

So, for 1024x768 and 256 colours (that's 8 bit):
1024 * 768 * 8/8 = 786432 bytes i.e. a 1 Meg card will suffice

and for other configurations:
640x480x24 bit colour = 921600 (min. 1 Meg card)
800x600x16 bit colour = 960000 (min. 1 Meg card)
800x600x24 bit colour = 1440000 (min. 2 Meg card)
1024x768x16 bit colour = 1572864 (min. 2 Meg card)
1024x768x24 bit colour = 2359296 (min. 4 Meg card)
1280x1024x8 bit colour = 1310720 (min. 2 Meg card)
1280x1024x24 bit colour = 3932160 (min. 4 Meg card)
1600x1200x24 bit colour = 5760000 (min. 6 Meg card)

Note that many truecolour implementations (24 bit colour) use 32 bit
long words.  For these chipsets/modes you will have to use a pixel
depth of 32 in the above calculation  i.e. 24 bit colour may not be
available at 1280x1024 with some 4 Meg cards.



Q) What is the difference between VRAM and DRAM?
        (or, Should I buy a VRAM or DRAM based video card?)

This is one of the most commonly asked question in this group, and is
usually answered more or less correctly, though often for the wrong
reasons.

DRAM (Dynamic RAM) used on video cards is the same technology as the
main system RAM on most computers.  The 'dynamic' part refers to the
fact that this type of memory must be refreshed several times per
minute or it will 'forget' the data it is storing.  This means that
DRAM has a duty cycle (a period during which the RAM is being refreshed
and can't respond to external requests like reads/writes), unlike SRAM
(Static RAM) which does not require refreshing, and thus is available
at all times.  DRAM, however, requires fewer discrete components for
each bit stored, so physically takes less silicon, and thus is cheaper
to manufacture.

An additional limitation of DRAM is that it can do only one thing at a
time - it can either be read from or written to.  There are two data transfer
steps occurring on your video card.  The first is to transfer data from the
CPU to video RAM.  The second is to transfer the video RAM data to the
RAMDAC, which produces the video signal you see on your screen.  The maximum
amount of data which you can pump in and out of your video memory in one
second is your 'video bandwidth'.  Thus, the read and write operations must
share the available video bandwidth, which means that the DRAM has to
service both read requests from the RAMDAC and write requests from the CPU.
At high pixel addressabilities and colour depths, an enormous amount of
extra data has to be moved to and from the video memory, and as a result,
DRAM boards may run out of bandwidth.  This means that you may not be able
to refresh your monitor fast enough to avoid flicker.

VRAM is a special type of DRAM which is dual-ported.  It still has a duty
cycle, but it can written to and read from at the same time.  In practice,
this means that you get double the bandwidth out of 60 ns VRAM as you
would out of 60 ns DRAM (if implemented correctly on the video card).

The long and the short of this is that VRAM cards are capable of higher
screen refresh rates at high pixel addressabilities and colour depths, while
DRAM cards are not.  Some VRAM cards provide marginally better performance
than comparable DRAM versions at lower addressabilities, but this will not
affect the majority of users significantly.  It will affect you if you run
your monitor at high pixel addressabilities _and_ colour depths.  Typically
VRAM based cards perform better where DRAM cards drop off; noticeably at
pixel addressabilities and colour depths greater than or equal to:
800x600 x 24 bit colour (16.7 M colours)
1024x768 x 16 bit colour (64k colours)

DRAM cards may be unable to provide high vertical refresh rates (>70Hz)
at higher addressabilities.  Most people aren't bothered by refresh rates
>=60 Hz.

For techies who are looking for more detail:
[From: Sam Goldwasser ([e-mail])]

Both types are used to store video information.  However, VRAM is not just
fast DRAM.  In fact, the random access times for typical VRAM is worse than
for similar DRAM:

VRAM is a special type of DRAM which includes a shift register that
can be loaded in parallel from an entire row in the DRAM array in approximately
the same time as a single read cycle.  The shift register (typically
256-2048 stages depending on the organization of the memory array) can be
clocked independently of the normal random access to the chip.

The original intended use was for refreshing raster scan displays - thus
the 'V' for video.  Since the shift register is clocked independently,
the percentage of time that the VRAM random access port is busy servicing
video refresh is reduced from 'very high' to almost insignificant.
For example, using DRAM, a typical design may require 50% of the
random access port bandwidth for video refresh with DRAM but only .5% with
VRAM.  You load the shift register only once or twice per video line rather
than having to access the memory array for every pixel.  Some designs
have a split shift register which provides even more flexibility in
shift register load timing.

VRAM is slightly more expensive on a $/MB basis and is usually about 1
generation behind in terms of common chip densities.  4 Mbit
VRAM chips are just now becoming commonplace.

There are a number of variations on this basic theme including some
triple port varieties as well.

In addition to video, VRAM finds application in high performance printers
and plotters, communications, signal processing, image capture using
the shift register for input), and many other areas.

Also, from: [e-mail] (Harm Hanemaaijer):

Each resolution takes up a certain amount of bandwidth for monitor
refresh. If this takes up most of the available bandwidth, performance
goes down steeply.

With VRAM the bandwidth for drawing is basically unaffected by monitor
refresh.

Resolutions where this happens are
                                                bandwidth left
1024x768x256 NI on a 1Mb DRAM card               45 Mb/s
800x600x16bit on a 1Mb DRAM card                 20 Mb/s
1Mb VRAM card (all resolutions)                 100 Mb/s

(1Mb DRAM card has 60 MHz MCLK yielding 120Mb/s of memory bandwidth)

1024x768x256 on a 64-bit 2Mb DRAM card          165 Mb/s (good)
1024x768x32K on a 64-bit 2Mb DRAM card           90 Mb/s
800x600x32bit on a 64-bit 2Mb DRAM card          40 Mb/s
64-bit 2Mb VRAM card (all resolutions)          200 Mb/s

(2Mb DRAM card has 60 MHz MCLK yielding 240Mb/s of memory bandwidth)

It follows that so called 64-bit DRAM cards with only 1Mb are a pretty
bad idea.

It can also be seen that 2Mb 64-bit DRAM cards can be faster than
VRAM in very low resolutions that take up little bandwidth since the
total bandwidth of the DRAM card may be a bit higher (e.g. 240 vs.
200Mb/s).

I'm not sure about the typical bandwidth of VRAM-based cards, but
as far as DRAM cards are concerned most aggressively timed S3-864
based cards it is 120 Mb/s (1Mb) or 240 Mb/s (2Mb) while for more
conservatively timed cards (which may imply better stability)
it is about 100 / 200 Mb/s (this also goes for most CL-GD5434 based
cards). You might imagine the performance vs. stability dilemma faced
by manufacturers on this issue (the conservative 1Mb model has only
25Mb/s bandwidth at 1024x768x256 -> bonehead tester thinks it sucks).



Q) What types of video RAM are available (or coming soon)?

Video cards use their on-board RAM in different ways, but for this
example we will only consider it as a framebuffer.  This means that it
is used to store a digital 'snapshot' of what appears on the computer
monitor.  This framebuffer is used in two different ways.  The video
processor writes data to the framebuffer, and the RAMDAC reads data
from the framebuffer, converts it to an analog signal, then sends that
signal to the monitor.  These two operations, reading and writing,
must share the available bandwidth of the video RAM.  At high pixel
addressabilities and refresh rates, the RAMDAC can be quite demanding.
At the same time, if more pixels have to be updated by the video
processor (because of the higher pixel addressabilities) then there
may not be enough video memory bandwidth available.  As today's users
move to higher and higher pixel addressabilities (1024x768 and up)
and want to display more colours simultaneously (16.7 million and up)
traditional type of RAM are becoming inadequate.

All video cards currently use some form of DRAM (Dynamic Random Access
Memory) because of its high price/performance ratio.  This section
discusses some of the current and upcoming types of DRAM.  There are two
basic classifications used here;  single-ported and dual-ported.  See
the section "What is the difference between VRAM and DRAM?" for a
detailed explanation of single verses dual-ported RAM.  The quick
explanation is that single-ported RAM has only one data path that has
to be shared between read and write operations.  So, if the RAM has
a bandwidth of, say, 80 Mbyte/s, then the read bandwith plus the write
bandwidth must be less than 80 Mbytes/s.  The read and write operations
can share this bandwidth in any ratio, 50-50 or otherwise.  Dual-ported
RAM has separate read and write data paths, meaning that if it operates
at the same speed as the RAM in the single-ported example, 80 Mbytes/s
bandwidth is available for _each_ of the reading and writing operations.
This effectively doubles total bandwidth of the RAM.

If that's the case, why don't all video cards use VRAM?  Dual-ported
RAM requires more discrete silicon components to store each bit of
digital information than traditional DRAM, so the manufacturing cost
goes up.  Recent efforts have been focussed on trying to produced
alternatives to VRAM which are as fast or faster, but at lower cost.
Some of the new contenders are very fast, and not much more expensive
than DRAM.

   Standard DRAM (also called FPM DRAM for Fast-Page Mode) is the least
expensive memory used in video cards.  It is still the most popular
type of memory used in video cards.  DRAM typically runs at an i/o bus
frequency in the 25-33 MHz range and provides a net bandwidth of
~90 Mbyte/s.  Note that for DOS/VGA games, DRAM provides loads of
bandwidth, and that faster video memory will _not_ give better
performance.

   VRAM (Video RAM) is one of the most expensive types of RAM, and is
the most common type used in high-end graphics cards.  It is dual-
ported, providing double the effective bandwidth of DRAM running at
the same speed.

   WRAM (Window RAM) is dual-ported, and can be clocked at up to
50 MHz providing up to 50% more bandwidth than conventional VRAM.
Due to its design, WRAM requires fewer silicon components than
VRAM and as a result is ~20% cheaper.  It has also been optimized to
provide fast text and colour fills and aligned BitBLT's.  For more
info, refer to:
http://www.matrox.com/mgaweb/wram.htm   Benefits of WRAM Memory

   EDO DRAM (Extended Data Out DRAM) is being used for both video
cards and main system RAM due to it's improved performance and only
marginally higher cost over FPM DRAM.  It can be clocked at a
higher i/o bus frequency (40-50 MHz) and provides higher bandwidth
(~105 MHz).

   Other types of RAM are under development, or have been implemented
to a lesser extent.  Most of these are single-ported designs which
try to reduce memory latency through tricks like higher bus frequency,
interleaving of multiple banks (MDRAM) and wider memory buses.

Summary:

Type            Means                   I/O Bus         Total   Net
                                        Frequency       Latency Bandwidth
                                        (MHz)           (ns)    (Mbyte/s)
FPM DRAM        Fast Page-Mode RAM      25-33           80      80
VRAM            Video RAM
WRAM            Window RAM                                      120
EDO DRAM        Extended Data Out DRAM  40-50           100     105
SDRAM           Synchronous DRAM        66-100          102-75  166-253
RDRAM           Rambus DRAM             250             108     206
MDRAM           Multibank DRAM          125-166         22-19   405-490
SGDRAM          Synchronous Graphics DRAM
EDRAM           Enhanced DRAM

Note that these numbers come from data published in Electronic Engineering
Times and Tseng Labs.



Q) What is the EEPROM, EPROM, PROM on my video card?

A PROM (Programmable Read Only Memory) chip can be used to store video BIOS
code and/or video configuration information.  All modern video cards have a
video BIOS, while many (but not all) store video configuration data right
on the video card.

Usually, EPROM's or PROM's (the 'E' indicating that the PROM is erasable or
'reprogrammable') are used to store the video BIOS.  The BIOS consists of
code which the computer uses to communicate with the video card.
Occasionally when a bug is discovered in the video BIOS of a card, the card
manufacturer will supply updated ROM's that contain corrected video BIOS code.
Swapping of the new chip for the old is quite easy and safe, as long as the
machine is powered off and appropriate electrostatic precautions are taken.

EEPROM's (Electronically Erasable PROM's) are used on some video cards to
store video mode configuration information.  These cards usually require
the user to run an installation program initially to determine the correct
screen refresh rate for the monitor at each screen pixel addressibility used.
Whenever the video card receives a request from the computer to switch video
modes, the card checks its EEPROM so that it can generate the video signal
at the correct vertical refresh rate.  Information such as horizontal and
vertical centering and size can also be stored by some cards.



Q) How does colour depth (bit planes) relate to the number of colours?
[From: Michael Scott ([e-mail])]

To understand this (and it isn't that difficult) you have to know what
the binary (or base 2) number system is.  Instead of each digit in a
number varying between 0 and 9, the values can only be 0 or 1.  This
means that for a given digit, there are only two possible options.
So, for say a 4 digit binary value, there are 4^2 (or 2x2x2x2) or 16
unique values.  Now it becomes easy to translate between the number
of bit planes (that's number of binary digits) and number of colours.

number of colours = 2^(# of bit planes) resulting in:
1 bit = 2 colours, 2 bit = 4 colours, 4 bit = 16 colours
8 bit = 256 colours, 15 bit = 32k, 16 bit = 64k, 24 bit = 16.7M

Note that the maximum colour depth at a given pixel addressability is
limited by the video controller, not the monitor, since almost all
modern monitors are analog.



Q) What are true color and high color?
[From: Ralph Valentino ([e-mail]), Mike Scott
([e-mail]) and corrections by Ethan Royael Nicholas
([e-mail])]

The color of a pixel is formed by mixing three colors: Red, Green and
Blue.  The number of discrete intensities that the video card is
capable of generating for each color determines the maximum number of
colors that can be displayed.  For most graphics cards, the intensity
of each of these colors ranges from 0 to 255, an 8 bit value.  So, the
total number of unique colors available is 16.7 million (2 ^ 24).

Depending on the implementation, a subset of these colours may be
available for display at a given pixel addressability.  The original
VGA controllers had three 6-bit DAC's (Digital to Analog Converter)
allowing up to 256 colours to be simultaneously displayed (in certain
video modes - others could display fewer colours) from a colour space
of 2^18 = 262144 unique colours.  Because of the 6-bit DAC, up to
2^6 or 64 shades of gray could be displayed.  With some newer VGA cards
and many SVGA video cards that have 8-bit DAC's, you can pick any 256
colours from a palette of 2^24 = 16.7 million, though these modes are
not available through the VGA controller - usually they are achieved
via a graphics accelerator in an environment like Windows or OS/2.
If a picture that you want to display has more than 256 unique colors,
various methods can be used to come up with the 256 closest colors,
and when combined with dithering reasonable results can be achieved for
some images.  Images displayed with 8 bit color often look grainy or
patchy.

An improvement on this was high color.  This provides either 15 or 16
bits of colour depth by using 5 bits for each of red, green and blue
or 5 bits of red and blue and 6 bits of green, respectively.  As a
result, up to 32768 (15 bit) or 65536 (16 bit) colours can be
simultaneously displayed.  This provides an enormous visual improvement
over 8 bit color, and can be noticed immediately when viewing most
images.  For many folks (me included) the differences between high color
and true color (explanation below) are almost indiscernable unless you
look closely.

Most newer cards are capable of displaying the full 16.7 million
colors simultaneously at certain (usually lower) pixel addressabilities.
The ability to display all 16.7 million colors at a time instead of a
limited palette of those available is called 24-bit or true color.
Since it has a larger gamut of colors to choose from, it can display
colors much closer to the true picture colors.  The true color label
refers to the belief that most people can not perceive more than 16.7
million different colors, and so a 24 bit representation of an image
will look as good as the original with respect to color reproduction.
Others disagree, feeling that 32, 48 or even 64 bit colour is necessary.

Since it doesn't have to use close colors, it displays the true
picture colors, thus the name 'true color'.  Note that many picture
formats, including GIF, also have the 256 color limitation, so a true
color card won't improve the picture viewing at all.  More recent
picture formats, such as JPEG, support 24 bit color.

When used in truecolor mode, some video cards actually operate in a 32
bit mode.  This is due to the fact that the video processor is often
optimized to move 32 bit words around, and that the memory bus is
often 32 bits wide to each RAM bank.  In almost all cases, the result
is that while the card is effectively operating in 32 bit mode, only
24 bit color is displayed.  On cards that run this way, the calculation
of the amount of video RAM required is different than you might think
since instead of 24 bits, the card actually uses 32 bits for truecolor.
To provide 24 bit color while minimizing video RAM requirements, many
video processors implement a packed-pixel mode.  This results in the
card operating in 24 bit mode, but may have an associated performance
penalty.



Q) Can I use a 64/128+ bit card in on an ISA/EISA/VLB/PCI bus?
[From: Ralph Valentino ([e-mail])]

In this case, the 64/128+ refers to the amount of data that can be
handled by the video processor in one clock cycle.  This affects
the speed of operations on the video card, including transfers to
and from video RAM and to the RAMDAC which generates the video
signal.

This is completely independent of the width of the data bus which
runs from the main CPU to the video card.  As a result, a 192 bit
video card can run on any of the buses available:  ISA (16 bit),
EISA & VLB (32 bit) or PCI (currently implemented as 32 bit - future
64 bit).



Q) Will my video speed up enormously with a VLB/PCI upgrade?

This is another question which can be answered by "it depends on your
application".  If you require high bandwidth on your system bus because
of intensive video demands, then VLB (for 486's) or PCI (for Pentium/586's)
is a must.  Examples of high bandwidth applications include most VGA
(sometimes incorrectly called DOS) games, full-motion video or other
VGA-intensive operation.

You may think that a pseudo-OS like Windows 3.1 would require a video card
with high VGA speed, but that isn't necessarily the case.  The problem
with VGA is that almost all of the work must be done by the CPU.  This,
coupled with the fact that each refreshed pixel must be transported
across the system bus means that VGA is slow for OS's like Windows or
OS/2.  Fortunately many operations in GUI environments (like move a
window, for example) can be implemented right on the video card, and
are handled by the video coprocessor rather than the system CPU.  This
reduces both the time required to complete such tasks and the amount
of data that has to flow over the system bus.  An implementation like
this is often called video acceleration (see "How does a video
accelerator work, and will one help me?").

With this background, we can now see that most general operations
within a GUI environment can be handled right on the video card.  As
a result, it's possible to have a card that is fast for Windows
operations, but runs on the ISA bus.  In fact, for most operations,
it's quite possible that the ISA-based card will respond just as
quickly as an equivalent VLB or PCI card.

To summarize, high-bandwidth applications like full motion video or
VGA games will benefit from a fast bus like VLB or PCI.  For most
GUI operations (like in Windows 3.1, Windows 95, OS/2, XFree86, etc)
an accelerated ISA card might be the most economical upgrade path.



Q) How can an 8/16/32/64/128+ bit video card work on my 16/32/64
       bit system?

There are four different things at work here:  the CPU, system or
memory bus, the peripheral bus and the video coprocessor.  In general,
the number of bits that each can handle in one clock cycle is
independent of the others, with some exceptions.

Dealing with these in order, CPU's are available in 16 (8088, 8086,
80186 & 80286) and 32 (386, 486, 586, Pentium & Pentium Pro, 6x86) bit
versions.  This is the number of bits that the CPU can process
_internally_ per clock cycle.

The system bus for these processors covers a wide range;
8 bit (8088), 16 bit (8086, 80186, 80286, 386SX), 32 bit (80386DX,
80486, 586) and 64 bit (Pentium, Pentium Pro, 6x86)

Third is the width of the peripheral bus, which is the
number of bits per cycle that can be moved between the CPU and an
add-in card.  The ISA bus is 8 or 16 bit, EISA is 32 bit, VLB is 32 bit
and the current PCI bus is 32 bit, but in future will be expanded
to 64 bits.

The last is dependent on the graphics coprocessor, which are available
in 8-192 (wow!) bit models.  To complicate things further, the maximum
effective memory bandwidth of the video card is limited by its memory
configuration.  i.e. a 64 bit video processor needs 2 32 bit banks
of video memory to operate as a true 64 bit system.

So how can a 32 bit processor use a 16 bit bus to talk to a 64 bit
video coprocessor?  Let's say we want to move a 24 bit (Truecolour)
colour value from the CPU to the screen.  Since the bus is limited
to 16 bits, we have to pad the 24 bit number to make it 32 bits
(an even multiple of 16) and then split it into the high and low
16 bit words.  Each of these 16 bit words is then pushed onto the
bus, and moved to the video card.  The video card accepts one, then
the other word, then recombines them into a 32 bit word, which
contains the 24 bit data.  In this case, the 16 bit bus is the
bottleneck (discounting clock rates for simplicity).

The advantage to having a 64+ bit video processor attached to a
32 bit (or smaller) bus/CPU is that the video card can perform many
operations without input from the CPU.  These operations include
screen refreshes, pixmap painting, moving of windows, etc.  Since
the video processor moves more bits per cycle (i.e. 64), it can
complete video tasks very quickly (i.e. move 3 24 bit truecolour
values at once).



Q) How does memory interleaving work to increase the speed of a
        video card?
[From: Sam Goldwasser ([e-mail]) (with a bit by M. Scott)]

Memory interleaving using multiple banks of memory results in increased
video speed because the video processor is able to access data much
faster than possible with a single bank of video memory.

Interleaving means that the processor staggers memory read/writes between
2 or more banks of RAM, effectively multiplying the bandwidth of video
memory.  If the processor requests a read/write at a given memory
location, the amount of time required to complete the operation is
limited by the memory's speed.  I.e. 70 ns RAM will complete a read/write
request in no less than 70 ns.  If however, the processor can handle data
at, say, twice that rate (i.e. 35 ns per clock cycle or 29 MHz) then it's
effectively working at half speed for memory transfers.  If the processor
_interleaves_ memory accesses between two banks of RAM, then it can read/
write to bank 0 during one clock cycle, then instead of waiting for that
read/write cycle to finish before sending the next read/write request, it
immediately accesses bank 1 in the next clock cycle.

Interleaving is usually based on the low order (word) address bits.  Two-
way interleaved 32 bit memory will thus select between 1 of 2 banks based
on address bit 2 (bits 0 and 1 select a byte within the 32 bit word).
Four-way interleaved 32 bit memory will use both bits 2 and 3 to select
one of four banks.

Note that interleaved memory is most easily implemented for write cycles
since these can be 'posted' - issued and forgotten about.  Reads, on the
other hand, require that the processor keep track of the fact that one or
more read requests will be outstanding while it is issuing new ones.
This is a form of pipelining and not all processors are capable of
dealing with the necessary timing.

As an example of interleaving consider the following case of a processor
with a 33 MHz system bus accessing 70 ns memory:

Each clock cycle takes about 30 ns.  If the processor wants to write to
a block of RAM, it will have to insert 2 wait states between consecutive
writes, meaning that for this example it will take 90 ns for each write.
However, if each of 2 banks of RAM can be accessed separately, then
instead of inserting 2 wait states and leaving the processor idle, it
will instead access the second bank of memory during the next clock
cycle and insert only one wait state.  This has effectively doubled
memory throughput.

For this to work, the memory logic must latch the address, data, and
control signals so that as the processor moves on, the memory still
knows what to do.  In addition, since not all memory accesses will be
to alternate banks, the system must know to insert wait states if
successive accesses are to the same bank.

Since frame buffer writes are often to large blocks (pixblts and fills),
interleaving can achieve almost an ideal n:1 speedup where n is the
interleave factor.  The maximum practical value of n is limited by the
duration of a video processor clock cycle compared to RAM speed.  The cost
in terms of hardware depends on the memory organization since the n banks
must be in separate memory chips. The efficiency of the system depends on
the memory technology as well as the size of the frame buffer.  This is
one reason why a frame buffer which is not fully populated with memory
chips may not be able to take full advantage of its accelerated
capabilities.  In addition to the memory, some modest amount of additional
logic is required for controlling each bank of memory and generating the
timing including the insertion of wait states where needed.

Other approaches which may be used by themselves or in conjunction with
interleaving include 'page mode' accesses and the use of VRAM instead of
DRAM.



Q) Should I get 1 MB or 2 MB of video memory?

To determine the amount of video RAM you will need, use the method
outlined in "How do I calculate how much VRAM/DRAM I need?"

There is another issue which is important to consider if the video
card in question is a 64 bit card or is 32 bit but uses memory
interleaving.  Standard DRAM is addressed 32 bits at a time and
typically one 1 MB bank is 32 bits wide and 256 kbytes deep, so
if 2 1 MB banks are installed, each can be addressed separately.
In practice, a 64 bit video controller can move 64 bits in or out
of video RAM in each clock cycle, 32 bits to/from each bank.  If
only one bank of DRAM is installed (i.e. 1 MB) then the effective
bandwidth of the card is halved.  In a similar fashion, a 32 bit
controller that supports memory interleaving like the ET4000W32i/p
can move twice as much data per clock cycle if it has 2 MB of DRAM
installed (see "How does memory interleaving work to increase
the speed of a video card?"  This means that 64 bit cards that
have only 1 MB of video RAM will operate much more slowly that if
a full 2 MB is installed.  This may result in lower refresh rates
and more sluggish performance.



**********************************************************************
END of comp.sys.ibm.pc.hardware.video FAQ - Part 2/4
**********************************************************************





--
Michael J. Scott                       R.R.I.,  U of Western Ontario
[e-mail]                 'Need a good valve job?'
PC Video Hardware FAQ:  http://www.heartlab.rri.uwo.ca/videofaq.html
###############  Illegitimus non tatum carborundum.   ##############