Multiple monitors under Windoze...

D

Don Y

Guest
I typically run with 3 monitors on each workstation.
I arrange them so the center monitor is the monitor
driven by the BIOS at boot. Once Windows is up,
it acts as my primary desktop but extended to the
left and right to the adjoining monitors.

As sorting out \"which video output is which\" is
a crap shoot (they usually aren\'t labeled as such),
it\'s often the case that the boot monitor ends
up in the wrong place.

No problem: switch cables.

But, you would THINK that this would require you to
somehow tell Windows that you\'ve rearranged the
monitors. That doesn\'t appear to be the case!
(I\'ve done this on a couple of workstations already)

E.g., my boot monitor ended up as the rightmost
monitor. I had already configured windows for them
to be left-center-right. When I swapped the cables
for center and right (which moves the boot monitor into
the center, as desired), Windows still had the correct
placement. I.e., the right edge of the display surface
remained on the rightmost monitor -- even though the
previous \"right monitor\" had been electrically moved.

Does Windows \"identify\" the specific *monitors* on
each video port and arrange them accordingly (one would
think it just arranged video *ports*)...?
 
On Wednesday, July 19, 2023 at 10:27:03 PM UTC-4, Don Y wrote:
I typically run with 3 monitors on each workstation.
I arrange them so the center monitor is the monitor
driven by the BIOS at boot. Once Windows is up,
it acts as my primary desktop but extended to the
left and right to the adjoining monitors.

As sorting out \"which video output is which\" is
a crap shoot (they usually aren\'t labeled as such),
it\'s often the case that the boot monitor ends
up in the wrong place.

No problem: switch cables.

But, you would THINK that this would require you to
somehow tell Windows that you\'ve rearranged the
monitors. That doesn\'t appear to be the case!
(I\'ve done this on a couple of workstations already)

E.g., my boot monitor ended up as the rightmost
monitor. I had already configured windows for them
to be left-center-right. When I swapped the cables
for center and right (which moves the boot monitor into
the center, as desired), Windows still had the correct
placement. I.e., the right edge of the display surface
remained on the rightmost monitor -- even though the
previous \"right monitor\" had been electrically moved.

Does Windows \"identify\" the specific *monitors* on
each video port and arrange them accordingly (one would
think it just arranged video *ports*)...?

If Windows moved the boot monitor, why would it not have moved the monitors for other purposes as well?

--

Rick C.

- Get 1,000 miles of free Supercharging
- Tesla referral code - https://ts.la/richard11209
 
On 7/20/2023 10:00 AM, Bertrand Sindri wrote:
Don Y <blockedofcourse@foo.invalid> wrote:
On 7/20/2023 7:18 AM, legg wrote:
Are they the same make and model, operating from same video card
type output - or otherwise?

\"Identical\" monitors. I\'ve seen this with DP, DVI and VGA outputs
(though always three of the same, not mix and match)

One of the data items in the EDID stream is the monitor serial number.

Ah! OK, that would make sense.

But, that is surprising that MS would go to those lengths to
ensure the right *physical* monitor got placed in the correct
virtual portion of the display. *I* would have opted for
a simpler implementation (keep track of video output ports
and ignore what\'s connected to them).

So, they must have considered there to be some value to this
added effort (?)

[In my case, it added to the confusion as I assumed swapping
cables was equivalent to physically moving monitors... and was
surprised when the results were unchanged!]

Which, assuming the manufacturer burned an actual unique serial into
the ROM and assuming Win notes the serial numbers from the EDID data,
could be used to track where the monitors have moved when you swap the
cable connections.

Whether Win does this I do not know -- but your experiment implies it
might.

The next question would be how things work if you move a monitor
to a different type of interface on an entirely different type
of display adapter. I.e., just how \"special\" do they want to
treat the identity of each physical display?
 
Don Y <blockedofcourse@foo.invalid> wrote:
On 7/20/2023 10:00 AM, Bertrand Sindri wrote:
Don Y <blockedofcourse@foo.invalid> wrote:
On 7/20/2023 7:18 AM, legg wrote:
Are they the same make and model, operating from same video card
type output - or otherwise?

\"Identical\" monitors. I\'ve seen this with DP, DVI and VGA outputs
(though always three of the same, not mix and match)

One of the data items in the EDID stream is the monitor serial
number.

Ah! OK, that would make sense.

It is at least, assuming monitor makers *actually* burn a unique serial
into the ROM, a way that MS could identify that a monitor moved between
outputs.

But, that is surprising that MS would go to those lengths to ensure
the right *physical* monitor got placed in the correct virtual
portion of the display. *I* would have opted for a simpler
implementation (keep track of video output ports and ignore what\'s
connected to them).

I have no idea what MS did, or did not do. Your experiments imply they
may be tracking physical monitors. One way to do so could be the EDID
serial number value.

So, they must have considered there to be some value to this added
effort (?)

Or it is merely an unanticipated result of something else they did do
for some specific reason.

Which, assuming the manufacturer burned an actual unique serial into
the ROM and assuming Win notes the serial numbers from the EDID
data, could be used to track where the monitors have moved when you
swap the cable connections.

Whether Win does this I do not know -- but your experiment implies
it might.

The next question would be how things work if you move a monitor to a
different type of interface on an entirely different type of display
adapter. I.e., just how \"special\" do they want to treat the identity
of each physical display?

You\'ll have to perform those experiments and report your findings.
 
On 7/20/2023 12:11 PM, Bertrand Sindri wrote:
Don Y <blockedofcourse@foo.invalid> wrote:
On 7/20/2023 10:00 AM, Bertrand Sindri wrote:
Don Y <blockedofcourse@foo.invalid> wrote:
On 7/20/2023 7:18 AM, legg wrote:
Are they the same make and model, operating from same video card
type output - or otherwise?

\"Identical\" monitors. I\'ve seen this with DP, DVI and VGA outputs
(though always three of the same, not mix and match)

One of the data items in the EDID stream is the monitor serial
number.

Ah! OK, that would make sense.

It is at least, assuming monitor makers *actually* burn a unique serial
into the ROM, a way that MS could identify that a monitor moved between
outputs.

There has to be some way for MS to know that my \"identical\" monitors
are not, in fact, identical. A S/N would be the most obvious.

But, that is surprising that MS would go to those lengths to ensure
the right *physical* monitor got placed in the correct virtual
portion of the display. *I* would have opted for a simpler
implementation (keep track of video output ports and ignore what\'s
connected to them).

I have no idea what MS did, or did not do. Your experiments imply they
may be tracking physical monitors. One way to do so could be the EDID
serial number value.

But it begs the question: why would you want to track THAT endpoint
(of the cable) instead of the nearside endpoint (video output port)?

So, they must have considered there to be some value to this added
effort (?)

Or it is merely an unanticipated result of something else they did do
for some specific reason.

Possible as well.

The next question would be how things work if you move a monitor to a
different type of interface on an entirely different type of display
adapter. I.e., just how \"special\" do they want to treat the identity
of each physical display?

You\'ll have to perform those experiments and report your findings.

Ain\'t gonna happen withthe workstations -- they weigh ~70 pounds so
I don\'t drag them out of their respective parking spaces to poke around
with the back-panel cabling unless REALLY necessary.

But, I\'ve got a pair of 1U servers that are getting reheaded
that might provide some insight. Much fewer hardware options
(cuz servers aren\'t intended to have anything more than consoles!)
between the slot restrictions (slots are rarely \"adjacent\" in
the same way they are in a traditional PC so no two-wide GPUs;
x16 i/fs are infrequent as most HBAs are x4 or even x1), no
auxillary power for a high TDP device, etc.

OTOH, as most x1 or x4 video cards are two-headed, I\'ll be forced
to use the MB video in addition to an add-in card. So, almost
certain that they will have different drivers and low level hooks...

[Just installed a dual DVI-I so I can check that out this afternoon]
 
On Thursday, July 20, 2023 at 5:58:07 PM UTC-4, Don Y wrote:
There has to be some way for MS to know that my \"identical\" monitors
are not, in fact, identical. A S/N would be the most obvious.

Belarc Advisor will show you the monitor serial numbers. This is a tool to sow installed software and other information about a computer, to track licenses and equipment assets. I used to use it when I finished building a system, to be able to tell if it was tampered with, without opening the case.
https://www.belarc.com/products/belarc-advisor
 
On Thu, 20 Jul 2023 08:28:17 -0700, Don Y
<blockedofcourse@foo.invalid> wrote:

On 7/20/2023 7:18 AM, legg wrote:
E.g., my boot monitor ended up as the rightmost
monitor. I had already configured windows for them
to be left-center-right. When I swapped the cables
for center and right (which moves the boot monitor into
the center, as desired), Windows still had the correct
placement. I.e., the right edge of the display surface
remained on the rightmost monitor -- even though the
previous \"right monitor\" had been electrically moved.

Does Windows \"identify\" the specific *monitors* on
each video port and arrange them accordingly (one would
think it just arranged video *ports*)...?

Are they the same make and model, operating from same video card
type output - or otherwise?

\"Identical\" monitors. I\'ve seen this with DP, DVI and VGA
outputs (though always three of the same, not mix and match)

If you\'re using the motherboard output and some plug in cards,
they\'ll be handled differently in the bios and during/after boot.

In previous attempts, I\'ve been using a pair of dual-output GPUs.
E.g., most recently, after cabling the monitors to the GPUs,
I powered everything up and noticed the BIOS screen appeared
on the rightmost monitor:

\"OK, I\'ll have to make a note to fix that\" (But, because booting takes
so long with all the HBAs, etc. it\'s easier to let the boot proceed
to the OS instead of aborting it to \"fix\" the wiring)

Windows boots. Arrange the monitors and select the center one
for the desktop, extending left and right.

Shutdown.

Swap *cables* between right and center monitors. This ensures that
the BIOS will use the center monitor to display the boot screen
(press F2 for setup, etc.) because the BIOS is hardwired to pick
a specific video output connector as the \"primary monitor\".

Note that windows hasn\'t \"seen\" this action on my part because
it wasn\'t running when I unplugged/plugged. And, isn\'t running
*yet*!

Now Windows boots and, MAGICALLY, the monitors are still arranged
as I had intended when I previously selected the center
one for the desktop, etc.

I.e., despite the fact that the center monitor is no longer electrically
connected to the same video output port, windows has \"found it\" and
arranged the virtual displays the way I had previously specified.
(You would have expected the center and right content to be swapped!)

Last night, I opted to try a four-headed video card to see if
the results would be different. Label the physical monitors
A, B and C... arrange them (in Windows display settings) to be
in the ABC order. Now, power down and swap ANY two cables.
Power back up and windows still has them arranged ABC (but
the boot monitor may have \"physically moved\" if one of the
swapped cables happened to be connected to whichever port
the BIOS considers as \"console\")

I tried swapping AB, BC, AC and windows always sorted things out
to preserve the \"virtual display\" in the intended order.

So, there has to be something *instance* specific that is available
from the monitor so Windows can \"find\" the monitor that it
considers to be \"the left edge of the virtual display\" vs. the
\"right edge of the virtual display\" (make/model being a constant
in these cases).

I *guess* this makes sense -- but, you\'d have assumed swapping cables
(video outputs) would be the same as physically moving monitors.
It\'s not.

Set it up and don\'t swapo anything.

RL
 
On Wed, 19 Jul 2023 19:26:50 -0700, Don Y
<blockedofcourse@foo.invalid> wrote:

I typically run with 3 monitors on each workstation.
I arrange them so the center monitor is the monitor
driven by the BIOS at boot. Once Windows is up,
it acts as my primary desktop but extended to the
left and right to the adjoining monitors.

As sorting out \"which video output is which\" is
a crap shoot (they usually aren\'t labeled as such),
it\'s often the case that the boot monitor ends
up in the wrong place.

No problem: switch cables.

But, you would THINK that this would require you to
somehow tell Windows that you\'ve rearranged the
monitors. That doesn\'t appear to be the case!
(I\'ve done this on a couple of workstations already)

E.g., my boot monitor ended up as the rightmost
monitor. I had already configured windows for them
to be left-center-right. When I swapped the cables
for center and right (which moves the boot monitor into
the center, as desired), Windows still had the correct
placement. I.e., the right edge of the display surface
remained on the rightmost monitor -- even though the
previous \"right monitor\" had been electrically moved.

Does Windows \"identify\" the specific *monitors* on
each video port and arrange them accordingly (one would
think it just arranged video *ports*)...?

Is random truly random?

Does it make any difference if the the monitors remain
powered on between re-boots?

If they\'re powered down, or only your \'desktop\' of choice
is powered on at boot, the powered unit will be selected
as desktop.

Don\'t change any cables. Move monitors ;-?

RL
 
On 7/21/2023 4:57 AM, legg wrote:
On Thu, 20 Jul 2023 08:28:17 -0700, Don Y
blockedofcourse@foo.invalid> wrote:

On 7/20/2023 7:18 AM, legg wrote:
E.g., my boot monitor ended up as the rightmost
monitor. I had already configured windows for them
to be left-center-right. When I swapped the cables
for center and right (which moves the boot monitor into
the center, as desired), Windows still had the correct
placement. I.e., the right edge of the display surface
remained on the rightmost monitor -- even though the
previous \"right monitor\" had been electrically moved.

Does Windows \"identify\" the specific *monitors* on
each video port and arrange them accordingly (one would
think it just arranged video *ports*)...?

Are they the same make and model, operating from same video card
type output - or otherwise?

\"Identical\" monitors. I\'ve seen this with DP, DVI and VGA
outputs (though always three of the same, not mix and match)

If you\'re using the motherboard output and some plug in cards,
they\'ll be handled differently in the bios and during/after boot.

In previous attempts, I\'ve been using a pair of dual-output GPUs.
E.g., most recently, after cabling the monitors to the GPUs,
I powered everything up and noticed the BIOS screen appeared
on the rightmost monitor:

\"OK, I\'ll have to make a note to fix that\" (But, because booting takes
so long with all the HBAs, etc. it\'s easier to let the boot proceed
to the OS instead of aborting it to \"fix\" the wiring)

Windows boots. Arrange the monitors and select the center one
for the desktop, extending left and right.

Shutdown.

Swap *cables* between right and center monitors. This ensures that
the BIOS will use the center monitor to display the boot screen
(press F2 for setup, etc.) because the BIOS is hardwired to pick
a specific video output connector as the \"primary monitor\".

Note that windows hasn\'t \"seen\" this action on my part because
it wasn\'t running when I unplugged/plugged. And, isn\'t running
*yet*!

Now Windows boots and, MAGICALLY, the monitors are still arranged
as I had intended when I previously selected the center
one for the desktop, etc.

I.e., despite the fact that the center monitor is no longer electrically
connected to the same video output port, windows has \"found it\" and
arranged the virtual displays the way I had previously specified.
(You would have expected the center and right content to be swapped!)

Last night, I opted to try a four-headed video card to see if
the results would be different. Label the physical monitors
A, B and C... arrange them (in Windows display settings) to be
in the ABC order. Now, power down and swap ANY two cables.
Power back up and windows still has them arranged ABC (but
the boot monitor may have \"physically moved\" if one of the
swapped cables happened to be connected to whichever port
the BIOS considers as \"console\")

I tried swapping AB, BC, AC and windows always sorted things out
to preserve the \"virtual display\" in the intended order.

So, there has to be something *instance* specific that is available
from the monitor so Windows can \"find\" the monitor that it
considers to be \"the left edge of the virtual display\" vs. the
\"right edge of the virtual display\" (make/model being a constant
in these cases).

I *guess* this makes sense -- but, you\'d have assumed swapping cables
(video outputs) would be the same as physically moving monitors.
It\'s not.

Set it up and don\'t swapo anything.

It\'s not trivial to know WHICH connector on a multi-output
video card will be:
- the \"console\"
- #2
- #3
etc.

(And, some cards change how they behave based on what they
*sense* on their outputs)

So, you take a stab at cabling it. Then, turn on machine to
see which monitor displays the boot prompts (Press X for Setup).
Then, watch to see where \"Starting Windows\" appears. And,
finally, where the initial main desktop is located.

You *must* swap cables (or physically move the monitors)
to get the \"boot\" display where you want it.

The rest can usually be done rearranging virtual monitors
under Windows.

Then, you have to ensure that any other OSs that run on that
hardware follow a compatible convention.
 
On 7/21/2023 5:07 AM, legg wrote:
On Wed, 19 Jul 2023 19:26:50 -0700, Don Y
blockedofcourse@foo.invalid> wrote:

I typically run with 3 monitors on each workstation.
I arrange them so the center monitor is the monitor
driven by the BIOS at boot. Once Windows is up,
it acts as my primary desktop but extended to the
left and right to the adjoining monitors.

As sorting out \"which video output is which\" is
a crap shoot (they usually aren\'t labeled as such),
it\'s often the case that the boot monitor ends
up in the wrong place.

No problem: switch cables.

But, you would THINK that this would require you to
somehow tell Windows that you\'ve rearranged the
monitors. That doesn\'t appear to be the case!
(I\'ve done this on a couple of workstations already)

E.g., my boot monitor ended up as the rightmost
monitor. I had already configured windows for them
to be left-center-right. When I swapped the cables
for center and right (which moves the boot monitor into
the center, as desired), Windows still had the correct
placement. I.e., the right edge of the display surface
remained on the rightmost monitor -- even though the
previous \"right monitor\" had been electrically moved.

Does Windows \"identify\" the specific *monitors* on
each video port and arrange them accordingly (one would
think it just arranged video *ports*)...?

Is random truly random?

Nothing is random. But, video cards often don\'t CLEARLY state:
\"This is output #1\", etc.

And, often there are two or more connectors that can be alternatives
for a particular output (pick one).

Does it make any difference if the the monitors remain
powered on between re-boots?

I leave the monitors on all the time (let them sleep when they
need to). In general, it\'s good practice to ensure monitor is
on before computer.

[And, most of my monitors are tied to two or more computers]

If they\'re powered down, or only your \'desktop\' of choice
is powered on at boot, the powered unit will be selected
as desktop.

Don\'t want the desktop to change. Want it to be *here*, always.

> Don\'t change any cables. Move monitors ;-?

Once things are cabled (and configured) correctly, everything stays
that way. The issue is in the initial cabling/configuration.

[Moving monitors is a lot harder than swapping cables. I label
each cable-end with the monitor to which it is attached so I
can sort through cable-ends without having to trace cables to
their destinations]
 
On 7/21/2023 4:18 AM, Michael Terrell wrote:
On Thursday, July 20, 2023 at 5:58:07 PM UTC-4, Don Y wrote:

There has to be some way for MS to know that my \"identical\" monitors
are not, in fact, identical. A S/N would be the most obvious.

Belarc Advisor will show you the monitor serial numbers. This is a tool to sow installed software and other information about a computer, to track licenses and equipment assets. I used to use it when I finished building a system, to be able to tell if it was tampered with, without opening the case.
https://www.belarc.com/products/belarc-advisor

Thanks. I have a similar tool -- but with a long, fancy name -- that queries
all hosts on a network for that sort of information. So, it takes a while
to run and, as such, I don\'t use it often. I\'ve just never really paid
attention to the specifics that it provides (too much detail is almost
as bad as too little).
 
On Friday, July 21, 2023 at 12:38:49 PM UTC-4, Don Y wrote:
On 7/21/2023 4:18 AM, Michael Terrell wrote:

Belarc Advisor will show you the monitor serial numbers. This is a tool to sow installed software and other information about a computer, to track licenses and equipment assets. I used to use it when I finished building a system, to be able to tell if it was tampered with, without opening the case.
https://www.belarc.com/products/belarc-advisor
Thanks. I have a similar tool -- but with a long, fancy name -- that queries
all hosts on a network for that sort of information. So, it takes a while
to run and, as such, I don\'t use it often. I\'ve just never really paid
attention to the specifics that it provides (too much detail is almost
as bad as too little).

Its output is a HTML page, so you can use your browser\'s search function to find what you need. It even diplays the serial number of your RAM modules.. It is free for personal use.
 
On Fri, 21 Jul 2023 09:35:17 -0700, Don Y
<blockedofcourse@foo.invalid> wrote:


Don\'t want the desktop to change. Want it to be *here*, always.

Only on boot. Press the monitor \'off\' button on non-desktop
displays, first.

RL
 
On 7/21/2023 12:07 PM, legg wrote:
On Fri, 21 Jul 2023 09:35:17 -0700, Don Y
blockedofcourse@foo.invalid> wrote:



Don\'t want the desktop to change. Want it to be *here*, always.

Only on boot. Press the monitor \'off\' button on non-desktop
displays, first.

Or, swap cables to get the boot monitor where you want it
and forget about it!
 
On 7/21/2023 9:57 AM, Michael Terrell wrote:
Thanks. I have a similar tool -- but with a long, fancy name -- that
queries all hosts on a network for that sort of information. So, it takes
a while to run and, as such, I don\'t use it often. I\'ve just never really
paid attention to the specifics that it provides (too much detail is
almost as bad as too little).

Its output is a HTML page, so you can use your browser\'s search function to
find what you need. It even diplays the serial number of your RAM modules.
It is free for personal use.

I will have to check what mine accumulates. I selected it because it also
was capable of gleaning information about non-windows hosts (as most of
my hosts fall into that category)
 
On 7/20/2023 2:57 PM, Don Y wrote:
OTOH, as most x1 or x4 video cards are two-headed, I\'ll be forced
to use the MB video in addition to an add-in card.  So, almost
certain that they will have different drivers and low level hooks...

[Just installed a dual DVI-I so I can check that out this afternoon]

Inconclusive as no driver available for the card. <frown>
 

Welcome to EDABoard.com

Sponsor

Back
Top