RX 480 gives Error 43 upon installation?

Just installed an AMD RX 480 graphics card into my system (replacing the
previous GTX 750 Ti). After installation, I keep getting a message,
"Windows has stopped this device because it has reported problems. (Code
43)".

I've looked up the code #43 and there doesn't really seem to be a
solution, nobody seems to know how to fix.

I installed the latest Adrenalin drivers 18.11.1. When that didn't work,
I went back to the previous version 18.9.3. I've also used the DDU
utilitiy to completely remove the driver remnants, but it didn't work
any better.

Yousuf Khan

Yousuf Khan wrote:
Just installed an AMD RX 480 graphics card into my system (replacing the
previous GTX 750 Ti). After installation, I keep getting a message,
"Windows has stopped this device because it has reported problems. (Code
43)".

I've looked up the code #43 and there doesn't really seem to be a
solution, nobody seems to know how to fix.

I installed the latest Adrenalin drivers 18.11.1. When that didn't work,
I went back to the previous version 18.9.3. I've also used the DDU
utilitiy to completely remove the driver remnants, but it didn't work
any better.

Yousuf Khan

https://support.microsoft.com/en-ca/...-error-code-43

"This error occurs when your graphics device driver has
notified Windows that the device isn’t working properly.

This may mean that the device has a hardware problem,
or that the driver or driver software is failing.
"

In other words, it "means anything they want it to mean".

*******

I don't know about you, but I learned a lesson the hard way
many eons ago (Win2K). I'd been through three brands of video
cards. Being a noob at it, I didn't uninstall the NVidia driver
before shutdown and removal of the NVidia card. I just plugged
in the ATI card and installed the driver. I got away with
that. However, on the third occasion (a Matrox card), I didn't
get a new error code, no, I got a refusal by the card to
enter 3D mode. I couldn't play games on it.

I tried everything. Detonator Destroyer. Whatever the ATI
cleaner of the day was. *Nothing* would work. I ended up
doing a clean install. Then, my new video card worked.

*******

As a result, it doesn't matter now what OS I'm running.
I learned my lesson. I always uninstall the existing
video card driver, before the shutdown and card changeover.
Even if I was changing from a TNT2 to a FX5200. Sometimes,
going from NVidia to NVidia, I *could* have left the driver,
it mighta sorta worked. But I no longer want to take chances.

In Windows 10, you have several driver possibilities.

Microsoft Basic Display Adapter (equivalent to inbox VESA driver)
(fallback, when none available)

In-box NVidia driver (probably cannot prevent this from being used)

Externally added NVidia driver with ShadowPlay and junk === removable

While you could have two video cards in the box, one
card an NVidia, the other an ATI, the two drivers are
present, that's different than a driver which is sitting
there with missing hardware to talk to. The driver should
really unload, if its mate is missing, but who knows
how this stuff works.

While I don't know exactly what broke in your situation,
I can heartily recommend "good hygiene" when dealing
with video cards. I learned my lesson. It *really* ****ed
me off to have to redo that OS.

If you do a Repair Install, the migration logic will
see the installed NVidia package and try and reinstall it.
That's my best guess. I don't think the installer logic
is clever enough to "discount" any of the materials
which would otherwise install. If a package is "blocked"
on a particular edition of Windows 10, then the package
won't be moved/reinstalled. But if the package passes
a validity check, I doubt it would block the package
installation "because it no longer makes sense to do it".

You can try removing the NVidia package now. But again,
my experience is, there could be junk left behind. Third
parties invented driver cleaners for a reason. It implies the
installers aren't doing a good job when you select "Remove".

HTH,
Paul

On 2018-11-18 2:08 p.m., Yousuf Khan wrote:
Just installed an AMD RX 480 graphics card into my system (replacing the
previous GTX 750 Ti). After installation, I keep getting a message,
"Windows has stopped this device because it has reported problems. (Code
43)".

I've looked up the code #43 and there doesn't really seem to be a
solution, nobody seems to know how to fix.

I installed the latest Adrenalin drivers 18.11.1. When that didn't work,
I went back to the previous version 18.9.3. I've also used the DDU
utilitiy to completely remove the driver remnants, but it didn't work
any better.

Is it even slightly possible that you received a GPU that was defective?


--
SilverSlimer
Minds: @silverslimer

SilverSlimer wrote:
On 2018-11-18 2:08 p.m., Yousuf Khan wrote:
Just installed an AMD RX 480 graphics card into my system (replacing
the previous GTX 750 Ti). After installation, I keep getting a
message, "Windows has stopped this device because it has reported
problems. (Code 43)".

I've looked up the code #43 and there doesn't really seem to be a
solution, nobody seems to know how to fix.

I installed the latest Adrenalin drivers 18.11.1. When that didn't
work, I went back to the previous version 18.9.3. I've also used the
DDU utilitiy to completely remove the driver remnants, but it didn't
work any better.

Is it even slightly possible that you received a GPU that was defective?

There's all sorts of separate test cases you could
run for that.

Grab a separate disk, put Win10 on it, test.
Or Win7. Or Linux and the ATI proprietary driver,
then some OpenGL based test.

I wonder if there's actually a log entry somewhere with
the details of the error ? (Like, what isn't working.)

Paul

On 11/18/2018 2:39 PM, Paul wrote:
I don't know about you, but I learned a lesson the hard way
many eons ago (Win2K). I'd been through three brands of video
cards. Being a noob at it, I didn't uninstall the NVidia driver
before shutdown and removal of the NVidia card. I just plugged
in the ATI card and installed the driver. I got away with
that. However, on the third occasion (a Matrox card), I didn't
get a new error code, no, I got a refusal by the card to
enter 3D mode. I couldn't play games on it.

I tried everything. Detonator Destroyer. Whatever the ATI
cleaner of the day was. *Nothing* would work. I ended up
doing a clean install. Then, my new video card worked.

Well, I did run the DDU util (Display Driver Uninstaller), which is
supposed to be able to uninstall all display drivers leftover on the
system. But are you saying maybe it is only uninstalling the current
driver (i.e. the AMD ones), and not touching the previous drivers?

Do you think an In-Place Repair install might fix this? Or do you think
putting the old 750 Ti back in, and then removing its drivers will be
the better option?

Also interestingly, I am not sure if this is related, but the hibernate
and sleep options are gone from the shutdown menu. shrug The machine's
power profiles aren't putting it into sleep either.

Yousuf Khan

On 11/18/2018 3:45 PM, SilverSlimer wrote:
Is it even slightly possible that you received a GPU that was defective?

I won't discount it, but the video card is working perfectly fine when
using the generic Microsoft drivers. The MS drivers are even working at
a 1080p resolution, I've never seen the generic drivers work at those
higher resolutions before. So the card seems to be working perfectly
fine at a basic level.

Yousuf Khan

Yousuf Khan wrote:
On 11/18/2018 2:39 PM, Paul wrote:
I don't know about you, but I learned a lesson the hard way
many eons ago (Win2K). I'd been through three brands of video
cards. Being a noob at it, I didn't uninstall the NVidia driver
before shutdown and removal of the NVidia card. I just plugged
in the ATI card and installed the driver. I got away with
that. However, on the third occasion (a Matrox card), I didn't
get a new error code, no, I got a refusal by the card to
enter 3D mode. I couldn't play games on it.

I tried everything. Detonator Destroyer. Whatever the ATI
cleaner of the day was. *Nothing* would work. I ended up
doing a clean install. Then, my new video card worked.

Well, I did run the DDU util (Display Driver Uninstaller), which is
supposed to be able to uninstall all display drivers leftover on the
system. But are you saying maybe it is only uninstalling the current
driver (i.e. the AMD ones), and not touching the previous drivers?

Do you think an In-Place Repair install might fix this? Or do you think
putting the old 750 Ti back in, and then removing its drivers will be
the better option?

Also interestingly, I am not sure if this is related, but the hibernate
and sleep options are gone from the shutdown menu. shrug The machine's
power profiles aren't putting it into sleep either.

Yousuf Khan

Well, a question I've got, is what is the OS doing
when the driver shows that 43 code ? How is it able
to run the screen ? Does it flip back to Basic Display
Adapter or something ? Maybe the reason the power schema
are damaged, is because of whatever the OS is doing
so it can have a screen to display on ?

You have Safe Mode, as an option for doing some work.
Maybe then, the driver won't be running, and the
equivalent of the Basic Display Adapter will be in charge.
Safe Mode might use a VESA method for screen display.

I'd approve of a Repair Install, if I didn't think
the computer would repeat all the stupid moves
it's made so far. You know how computers are.
Whatever "preference" got you where you are now,
the Repair Install will likely follow down the
same rat-hole.

When is the last time you made a backup image
of the C: on this machine ?

Paul

On 11/18/2018 6:35 PM, Paul wrote:
Well, a question I've got, is what is the OS doing
when the driver shows that 43 code ? How is it able
to run the screen ? Does it flip back to Basic Display
Adapter or something ? Maybe the reason the power schema
are damaged, is because of whatever the OS is doing
so it can have a screen to display on ?

This is actually a bit unclear to me. When I explicitly choose the
generic driver, GPU-Z and other utilities can't detect the GPU status.
When I'm using the AMD drivers, even with the Code 43, then the GPU
status is visible to these utils. Now how is this working if it says
there was an error? My assumption was that it just uses the generic
driver after it gets the error, but if the status is visible then it
must be using the AMD driver?

You have Safe Mode, as an option for doing some work.
Maybe then, the driver won't be running, and the
equivalent of the Basic Display Adapter will be in charge.
Safe Mode might use a VESA method for screen display.

Yeah, well I ran the DDU in Safe Mode, as that was what was suggested by
DDU.

I'd approve of a Repair Install, if I didn't think
the computer would repeat all the stupid moves
it's made so far. You know how computers are.
Whatever "preference" got you where you are now,
the Repair Install will likely follow down the
same rat-hole.

Doesn't a Repair install just replace all of the drivers with generic
drivers initially, until you update all of the drivers again?

When is the last time you made a backup image
of the C: on this machine ?

I make one every day, except weekends. Keep about 2 full images (each
month) and 10 incrementals.

Yousuf Khan

On 2018-11-18 4:57 p.m., Yousuf Khan wrote:
On 11/18/2018 3:45 PM, SilverSlimer wrote:
Is it even slightly possible that you received a GPU that was defective?

I won't discount it, but the video card is working perfectly fine when
using the generic Microsoft drivers.

Considering that the generic drivers don't push a GPU in any way, this
shouldn't be shocking in any way. They're more or less as basic as the
VESA drivers were in early versions of Linux and required to work no
matter what.

The MS drivers are even working at
a 1080p resolution, I've never seen the generic drivers work at those
higher resolutions before. So the card seems to be working perfectly
fine at a basic level.

Keyword: basic. I don't know but unless AMD is really bad at making
drivers, I'd return the card and get a new one. I know some
manufacturers have batches of GPUs which end up being defective, NVIDIA
had this happen a few years ago, so AMD might also be affected.


--
SilverSlimer
Minds: @silverslimer

On 11/18/2018 8:58 PM, SilverSlimer wrote:
On 2018-11-18 4:57 p.m., Yousuf Khan wrote:
On 11/18/2018 3:45 PM, SilverSlimer wrote:
Is it even slightly possible that you received a GPU that was defective?

I won't discount it, but the video card is working perfectly fine when
using the generic Microsoft drivers.

Considering that the generic drivers don't push a GPU in any way, this
shouldn't be shocking in any way. They're more or less as basic as the
VESA drivers were in early versions of Linux and required to work no
matter what.

I wouldn't think even a full-bore AMD driver would need to push the card
for just basic 2D tasks in the Windows desktop, until you actually start
using 3D apps like games or something.

The MS drivers are even working at a 1080p resolution, I've never seen
the generic drivers work at those higher resolutions before. So the
card seems to be working perfectly fine at a basic level.

Keyword: basic. I don't know but unless AMD is really bad at making
drivers, I'd return the card and get a new one. I know some
manufacturers have batches of GPUs which end up being defective, NVIDIA
had this happen a few years ago, so AMD might also be affected.

Well, it was a used card, but I won't blame it on that fact alone. It
seems that plenty of other people have had this problem too, even with
new cards. I'm more inclined to believe that perhaps it was an Nvidia
vs. AMD conflict, for right now. Another possibility, according to this
page is that perhaps there is something corrupt in the BIOS, and to
reflash the BIOS.

Error 43 with RX 480 graphics card - I feel lik... | Community
https://community.amd.com/message/2882838#2882903

Yousuf Khan wrote:
On 11/18/2018 8:58 PM, SilverSlimer wrote:
On 2018-11-18 4:57 p.m., Yousuf Khan wrote:
On 11/18/2018 3:45 PM, SilverSlimer wrote:
Is it even slightly possible that you received a GPU that was
defective?

I won't discount it, but the video card is working perfectly fine
when using the generic Microsoft drivers.

Considering that the generic drivers don't push a GPU in any way, this
shouldn't be shocking in any way. They're more or less as basic as the
VESA drivers were in early versions of Linux and required to work no
matter what.

I wouldn't think even a full-bore AMD driver would need to push the card
for just basic 2D tasks in the Windows desktop, until you actually start
using 3D apps like games or something.

The MS drivers are even working at a 1080p resolution, I've never
seen the generic drivers work at those higher resolutions before. So
the card seems to be working perfectly fine at a basic level.

Keyword: basic. I don't know but unless AMD is really bad at making
drivers, I'd return the card and get a new one. I know some
manufacturers have batches of GPUs which end up being defective,
NVIDIA had this happen a few years ago, so AMD might also be affected.

Well, it was a used card, but I won't blame it on that fact alone. It
seems that plenty of other people have had this problem too, even with
new cards. I'm more inclined to believe that perhaps it was an Nvidia
vs. AMD conflict, for right now. Another possibility, according to this
page is that perhaps there is something corrupt in the BIOS, and to
reflash the BIOS.

Error 43 with RX 480 graphics card - I feel lik... | Community
https://community.amd.com/message/2882838#2882903

On a modern UEFI BIOS, you might expect the NVRAM
holding the UEFI variables to be corrupt. UEFI variables
are too big to be stored in CMOS (256 byte) area, so
the storage is elsewhere.

As with a legacy BIOS, the main BIOS code block should be
checked with a checksum, before it is actually executed.
The check method might not be very good. And I don't
know if the boot block of the BIOS can beep an error for
that, or make a visual representation when it happens.
There have been cases, with legacy BIOS, where bit rot
occurred and the checksum didn't catch it. That's how
bad checksums were back then.

If you were to decide to "archive" a BIOS image, be aware
that portions of the BIOS are "volatile" and are altered
after you flash up. So attempting to compare an archive
against the original BIOS image of the same version,
you have to mask off the areas which are known-volatile.
The DMI/ESCD area or equivalent, would be an example
of a BIOS chip area which is writable and gets updated
every time your CPU changes or the amount and composition
of DIMMs changes. DMI is an "inventory" scheme, used by
corporations for "remote reporting" of hardware
inventories.

I'm willing to believe all sorts of things, before
I believe the BIOS main code block is getting damaged.
But the UEFI variable storage ? Open season. And I
don't know of a way off hand, to reset that. It would
be a research topic.

Paul

Yousuf Khan wrote:

Well, it was a used card, but I won't blame it on that fact alone. It
seems that plenty of other people have had this problem too, even with
new cards. I'm more inclined to believe that perhaps it was an Nvidia
vs. AMD conflict, for right now. Another possibility, according to this
page is that perhaps there is something corrupt in the BIOS, and to
reflash the BIOS.

One other thing.

Do you know the underclocking mechanism they use
for mining ? Is any information stored on the card
for that ? I would think mining would have the
underclock controlled by something like MSI Afterburner
or equivalent.

When mining, some flavor of mining is mainly about
"video card RAM bandwidth". You can turn down the GPU
portion, and turn up the RAM on the card. Perhaps
Ethereum needs this (Ethereum is RAM bandwidth limited).
Whereas for Bitcoin (which isn't mined with video cards
any more), that's all about the fastest SHA256 you can do,
so the RAM could be turned down, and the GPU portion
turned up.

I doubt the info is in the card - unless people
are distributing video card "BIOS" images. There
is a (typically) 64KB or 128KB EEPROM on the video
card, and one thing mine had was a declaration
that the DDR on the video card was "CAS 3". If
a person where to hack a BIOS and back off on the
settings, the power used by the video card RAM
could be reduced. But changing the clock with
Afterburner would be easier.

With a used card, there might be some other
possibilities. You'd really need to research the
depths a miner would go to, to understand what
to check for.

I would be using a separate disk, a clean OS
install (which would take 15 minutes), and see
if that's enough to make things work. As my first
step. Dark thoughts come after that test fails...

Paul

On 11/18/2018 6:35 PM, Paul wrote:
I'd approve of a Repair Install, if I didn't think
the computer would repeat all the stupid moves
it's made so far. You know how computers are.
Whatever "preference" got you where you are now,
the Repair Install will likely follow down the
same rat-hole.

Alright so I downloaded the latest image of Win 10 from the Microsoft
site. After running the In-Place installer, it said that it can only
work by deleting all programs and files, because my User folder is "in
an unsupported place"! So it looks like whatever is causing Microsoft's
infamous 1809 upgrade bug to delete everything is not fixed, and that MS
won't even try anymore.

Yousuf Khan

On 11/18/2018 2:08 PM, Yousuf Khan wrote:
Just installed an AMD RX 480 graphics card into my system (replacing the
previous GTX 750 Ti). After installation, I keep getting a message,
"Windows has stopped this device because it has reported problems. (Code
43)".

I've looked up the code #43 and there doesn't really seem to be a
solution, nobody seems to know how to fix.

I installed the latest Adrenalin drivers 18.11.1. When that didn't work,
I went back to the previous version 18.9.3. I've also used the DDU
utilitiy to completely remove the driver remnants, but it didn't work
any better.

So I reinstalled the 750 Ti again, and reinstalled the Nvidia drivers,
and it's working like normal before. Even the Sleep and Hibernate came
back. So it looks like Nvidia's drivers are causing the issues, it's
gotten itself intertwined into the power management in some way.

Yousuf Khan

Yousuf Khan wrote:
On 11/18/2018 6:35 PM, Paul wrote:
I'd approve of a Repair Install, if I didn't think
the computer would repeat all the stupid moves
it's made so far. You know how computers are.
Whatever "preference" got you where you are now,
the Repair Install will likely follow down the
same rat-hole.

Alright so I downloaded the latest image of Win 10 from the Microsoft
site. After running the In-Place installer, it said that it can only
work by deleting all programs and files, because my User folder is "in
an unsupported place"! So it looks like whatever is causing Microsoft's
infamous 1809 upgrade bug to delete everything is not fixed, and that MS
won't even try anymore.

Yousuf Khan

https://www.windowscentral.com/how-m...ive-windows-10

"Although relocating the user's default folders one by one means
extra steps, we do not recommend moving the main account folder ===
as it may cause unexpected problems."

This implies though, that there is a way to do that. I
can see an option in lusrmgr.msc labeled "profile path"
but the interface doesn't populate it with the
current value. Which makes it difficult to verify if
that's what the purpose is (move user entire folder,
not just components, blowing up the ability to update).

Paul