Cheap old generic graphics card is a Nvidia knockoff?, was theculprit for BSOD

From another thread I have information below on what I thought was a problem (the high temperatures on an (Intel?) ASRock Conroe1333-D667 Winbond W83627EHF
Brand- American megatrends Bios motherboard from 2007), but I think it turns out the culprit was in fact not high mobo temps but the cheap original equipment (from Thailand, yr 2007) video graphics card simply failed after ten years of use. When I removed the suspect video card, the BSOD, black and then back and then black flickering and seeming 'overheating' and artifacts on the screen disappeared (for now, thankfully, so far so good, though there's always a chance a confounding factor in addition to the old videocard is perhaps a bad mobo)

Right now I am running graphics on the mobo "integrated port" graphics card (built into the mobo) with no problems, but before the cheap video card identified itself (falsely!?) as an NVIDIA GeForce 8400 GS

From Nvidia's info log program, when I had this card installed:

/////////////
[Display]
Operating System: Windows 10 Home, 32-bit
DirectX version: 11.0
GPU processor: GeForce 8400 GS
Driver version: 341.92
Direct3D API version: 10
CUDA Cores: 16
Core clock: 459 MHz
Shader clock: 918 MHz
Memory data rate: 800 MHz
Memory interface: 64-bit
Total available graphics memory: 1855 MB
Dedicated video memory: 256 MB DDR2
System video memory: 0 MB
Shared system memory: 1599 MB
Video BIOS version: 60.86.4E.00.00
IRQ: 16
Bus: PCI Express x16
Device Id: 10DE 0422 82541043

///////////

However, looking at the old card now, in my hand, it has a sticker from 2007 (when I bought this Core2Duo tower in Thailand--Flasherly! Thailand! :-)) and it says nothing on it that it's from NVIDIA, but it's clearly generic, made in China, "ASUS" is the only label visible on the green circuit board, the chips are "Hynix" (video memory) and everything else is unmarked except the industry standards logos like "CE" and such.

It's not a "NVIDIA GeForce 8400 GS" as Windows 10 identified it as, and as Windows 10 installed the drivers for (and, amazingly, the drivers worked on this old generic card! How is that?). Interesting that Nvidia drivers worked for this generic card.

What to buy to replace this card? I probably will get the cheapest c rap out there, maybe a $30 card, which is somewhat ironic since I code C# programs, fairly complex, and even read complex AutoCAD drawings on this system, but for me performance is not important as I don't code or do architect stuff professionally. I guess I should get a decent card maybe since AutoCAD can bog down your system if you don't have a good graphics card.

RL

PS--the Prime95 program is excellent for stress testing the processor (uP), as well as memory, and as found in their Wiki page, it actually is superior to other programs since it 'bypasses' the 'shut down uP on high temperatures' mechanism and in fact you have to watch it or it could destroy your system due to overheating (so says the Wiki page and Tom's Hardware), see also: https://en.wikipedia.org/wiki/Prime9...stress_testing

Of interest since years ago I posted here, to some disbelief, that I found this Prime95 program superior to anything else, and it was catching memory mistakes that the memtest86 program was not catching, even when memtest was run overnight. (turns out the memory chips were not seated properly, it was a subtle problem that only occurred once in a while when the mobo got hot after many hours of use).

On Tuesday, February 14, 2017 at 2:34:12 PM UTC-5, Paul wrote:



One of my pet peeves today would be "is the Vcore heatsink
big enough" ? I see a lot of fairly tiny heatsinks
out there. And my last purchased motherboard, that's
about the only thing holding it back from being
a great motherboard. Is a ****ty Vcore heatsink choice.
I never gave it a thought before I bought it, but once
it overheated, I could immediately see what a dope
I had been. For not reviewing that before purchase.
If I'd noticed that, I probably would have rejected
that one, and spent another $100 on a better motherboard.

Paul

Well Paul saved me the effort of a separate thread. I logged in today to 'complain' about this old 2007 vintage ASRock mobo:

//
asrock conroe1333-d667 Winbond W83627EHF
Brand- American megatrends Inc
Version- P1.80
Date- 12/10/2007

The Board is Asrock Conroe1333-D667
Chipset- Intel i945G/GZ Rev. A2
Southbridge-Intel 82801GB(ICH7/R)
LPCIO Winbond- W83627EHF

http://www.asrock.com/mb/Intel/Conro...p?cat=Download
//

My complain is somewhat trivial --high temperatures reported for the mobo--but may relate to what Paul is saying about the heatsink (I'm not sure, but I noticed somebody else complaining about this ten year old mobo in that it had high temperatures on Tom's Hardware site, about 9 years ago):

when I run the Open Hardware Monitor 0.8.0 beta version (freeware, works good) it has as a reported temperature as follows (note the --!!!!!! temp) :


/////////////////////
Operating System: Microsoft Windows NT 10.0.14393.0
Process Type: 32-Bit

--------------------------------------------------------------------------------

Sensors

|
+- ConRoe1333-D667 (/mainboard)
| |
| +- Winbond W83627EHF (/lpc/w83627ehf)
)
| | +- Voltage #10 : 1.6 1.584 1.608 (/lpc/w83627ehf/voltage/9)
| | +- CPU : 32 30.5 35.5 (/lpc/w83627ehf/temperature/0)
| | +- Auxiliary : 44.5 42 96 --!!!!!!

/lpc/w83627ehf/temperature/1)
| | +- System : 33 30 35 (/lpc/w83627ehf/temperature/2)
| | +- CPU Fan : 2136.08 2136.08 2163.46

////////////////////////

Apparently this is 'normal" (from replies to a guy on Tom's Hardware), but it confused me since I was having problems, the system was acting funny on the screen, flickering, going black, then back to normal, and over and over, until after a few days I started getting BSOD, more and more (and annoyingly Windows 10 was asking if I wanted to reset the PC, which is bad advice IMO if you have a hardware problem). Of course I had backed up on an external drive my data and mirror imaged my drives. It turns out it was a old bad graphics card from 2007, a counterfeit "Nvidia", which I'll post separately on.

But what confused me was the high temps on this mobo.

Question for Paul: what would high temps do on a mobo? Give an unstable system? This "Thai" cheap core 2 duo always had quirks in it, and I blame the cheap mobo (and cheap components, which over the years I have replaced, not just the cheap mechanical HDD with SSDs, but also several power supplies, but so far no 'blown capacitors' as far as I can visually inspect).

RL

RayLopez99 wrote:
From another thread I have information below on what I thought was a problem (the high temperatures on an (Intel?) ASRock Conroe1333-D667 Winbond W83627EHF
Brand- American megatrends Bios motherboard from 2007), but I think it turns out the culprit was in fact not high mobo temps but the cheap original equipment (from Thailand, yr 2007) video graphics card simply failed after ten years of use. When I removed the suspect video card, the BSOD, black and then back and then black flickering and seeming 'overheating' and artifacts on the screen disappeared (for now, thankfully, so far so good, though there's always a chance a confounding factor in addition to the old videocard is perhaps a bad mobo)

Right now I am running graphics on the mobo "integrated port" graphics card (built into the mobo) with no problems, but before the cheap video card identified itself (falsely!?) as an NVIDIA GeForce 8400 GS

From Nvidia's info log program, when I had this card installed:

/////////////
[Display]
Operating System: Windows 10 Home, 32-bit
DirectX version: 11.0
GPU processor: GeForce 8400 GS
Driver version: 341.92
Direct3D API version: 10
CUDA Cores: 16
Core clock: 459 MHz
Shader clock: 918 MHz
Memory data rate: 800 MHz
Memory interface: 64-bit
Total available graphics memory: 1855 MB
Dedicated video memory: 256 MB DDR2
System video memory: 0 MB
Shared system memory: 1599 MB
Video BIOS version: 60.86.4E.00.00
IRQ: 16
Bus: PCI Express x16
Device Id: 10DE 0422 82541043

///////////

However, looking at the old card now, in my hand, it has a sticker from 2007 (when I bought this Core2Duo tower in Thailand--Flasherly! Thailand! :-)) and it says nothing on it that it's from NVIDIA, but it's clearly generic, made in China, "ASUS" is the only label visible on the green circuit board, the chips are "Hynix" (video memory) and everything else is unmarked except the industry standards logos like "CE" and such.

It's not a "NVIDIA GeForce 8400 GS" as Windows 10 identified it as, and as Windows 10 installed the drivers for (and, amazingly, the drivers worked on this old generic card! How is that?). Interesting that Nvidia drivers worked for this generic card.

What to buy to replace this card? I probably will get the cheapest c rap out there, maybe a $30 card, which is somewhat ironic since I code C# programs, fairly complex, and even read complex AutoCAD drawings on this system, but for me performance is not important as I don't code or do architect stuff professionally. I guess I should get a decent card maybe since AutoCAD can bog down your system if you don't have a good graphics card.

RL

PS--the Prime95 program is excellent for stress testing the processor (uP), as well as memory, and as found in their Wiki page, it actually is superior to other programs since it 'bypasses' the 'shut down uP on high temperatures' mechanism and in fact you have to watch it or it could destroy your system due to overheating (so says the Wiki page and Tom's Hardware), see also: https://en.wikipedia.org/wiki/Prime9...stress_testing

Of interest since years ago I posted here, to some disbelief, that I found this Prime95 program superior to anything else, and it was catching memory mistakes that the memtest86 program was not catching, even when memtest was run overnight. (turns out the memory chips were not seated properly, it was a subtle problem that only occurred once in a while when the mobo got hot after many hours of use).

The card manufacturer would be Asus.

The chip (not the card) is made by NVidia.

The card would be considered Asus-branded, and the
driver for it should be listed on the Asus site. Of course,
they would have stopped putting up newer drivers long
ago. Then, you'd go to the NVidia site for a basic
driver.

It's probably a perfectly valid Asus card.

*******

List the new graphics cards by price on Newegg. and you
should be able to see what currently passes for a card.

Here's a GT 710 for $35. I selected this, so your
driver support would last a tiny bit longer. There should
be a Win10 driver.

https://www.newegg.com/Product/Produ...82E16814487222

There is a PNY version for the same price, only this
one has a fan, rather than being passively cooled.

https://www.newegg.com/Product/Produ...82E16814133622

The high end on those cards is $45. Note - since you're
using Win10 x32, a card like this with 2GB of video memory,
that's going to chew into the 4GB address space big time,
leaving less room for system memory. Stick with a 1GB
card, leaving more room for system memory in the address space.
They probably don't make them with smaller memory setups than 1GB
but you can shop around and see.

https://www.newegg.com/Product/Produ...82E16814487219

If you were on a 64 bit OS, we wouldn't have to worry about
"too much RAM on the video card".

Paul

On 2/18/2017 1:43 PM, RayLopez99 wrote:
From another thread I have information below on what I thought was a problem (the high temperatures on an (Intel?) ASRock Conroe1333-D667 Winbond W83627EHF
Brand- American megatrends Bios motherboard from 2007), but I think it turns out the culprit was in fact not high mobo temps but the cheap original equipment (from Thailand, yr 2007) video graphics card simply failed after ten years of use. When I removed the suspect video card, the BSOD, black and then back and then black flickering and seeming 'overheating' and artifacts on the screen disappeared (for now, thankfully, so far so good, though there's always a chance a confounding factor in addition to the old videocard is perhaps a bad mobo)

Any bulging or "popped" capacitors on that bad card?

On Saturday, February 18, 2017 at 6:09:36 PM UTC-5, Paul wrote:


The card manufacturer would be Asus.

The chip (not the card) is made by NVidia.

The card would be considered Asus-branded, and the
driver for it should be listed on the Asus site. Of course,
they would have stopped putting up newer drivers long
ago. Then, you'd go to the NVidia site for a basic
driver.

It's probably a perfectly valid Asus card.

*******

Interesting, I did not know that Nvidia sells just chips and not the entire card, but I should have inferred that since I've seen cheaper cards imply just that.



List the new graphics cards by price on Newegg. and you
should be able to see what currently passes for a card.

Here's a GT 710 for $35. I selected this, so your
driver support would last a tiny bit longer. There should
be a Win10 driver.

https://www.newegg.com/Product/Produ...82E16814487222


What is your view about driver support? Why would driver support be a factor, if you have a Win10 driver? To get bug fixes maybe? Assuming Win10 does not 'morph' into a radically different OS, I don't see why driver support is an issue. If it works, don't fix it...



There is a PNY version for the same price, only this
one has a fan, rather than being passively cooled.

https://www.newegg.com/Product/Produ...82E16814133622

I like this idea, I don't like noise. Your views on passive cooling for low-performance cards? I think it's OK (my intuition).


The high end on those cards is $45. Note - since you're
using Win10 x32, a card like this with 2GB of video memory,
that's going to chew into the 4GB address space big time,
leaving less room for system memory. Stick with a 1GB
card, leaving more room for system memory in the address space.
They probably don't make them with smaller memory setups than 1GB
but you can shop around and see.

Good point, I think the old card was 512 MB video RAM. I will keep this in mind, I didn't know video RAM is now 2GB for 'standard' video cards, I thought only high-end cards had that much VRAM.


https://www.newegg.com/Product/Produ...82E16814487219

If you were on a 64 bit OS, we wouldn't have to worry about
"too much RAM on the video card".


I got a 32 bit OS since I figured that in practice (though not in theory) Windows 64-bit OS is a problem supporting some applications ported for 32-bit OS; since I run an old Windows 95 program for accounting I didn't want to take a chance (true, you can run in 'compatibility mode' which I guess).

RL

On Saturday, February 18, 2017 at 6:45:34 PM UTC-5, Bob F wrote:
On 2/18/2017 1:43 PM, RayLopez99 wrote:
From another thread I have information below on what I thought was a problem (the high temperatures on an (Intel?) ASRock Conroe1333-D667 Winbond W83627EHF
Brand- American megatrends Bios motherboard from 2007), but I think it turns out the culprit was in fact not high mobo temps but the cheap original equipment (from Thailand, yr 2007) video graphics card simply failed after ten years of use. When I removed the suspect video card, the BSOD, black and then back and then black flickering and seeming 'overheating' and artifacts on the screen disappeared (for now, thankfully, so far so good, though there's always a chance a confounding factor in addition to the old videocard is perhaps a bad mobo)

Any bulging or "popped" capacitors on that bad card?

Not that I can see. I have seen pictures of old popped capacitors and I do understand the top does not have to be blown off for them to be popped, but insofar as I can tell, the old card capacitors are not bulged at all, none of the 'scored' lines at the top appear in any way distorted.

And the temperature logging program is indicating even now the mobo is running a bit hot in spikes (it must be the design, as I said about the thread on Tom's Hardware that implied just that, this version mobo gives high temperatures at times--a temperature surge).

RL

RayLopez99 wrote:

What is your view about driver support?
Why would driver support be a factor, if you have a Win10 driver?
To get bug fixes maybe? Assuming Win10 does not 'morph'
into a radically different OS, I don't see why driver support
is an issue. If it works, don't fix it...

If a new OS comes along, and Microsoft decides to exclude the
old driver, then your screen runs in VESA mode. This is 1024x768
and may not agree well with the aspect ratio and capabilities
of your monitor (i.e. it will look ugly or fonts slightly fuzzy).

For example, Windows 10 is a rolling release. Say they decide
to bump the version of WDDM one notch higher. That might invalidate
the driver you currently are using. Changing the version of
software DirectX would be less of an issue.

I like this idea, I don't like noise. Your views on
passive cooling for low-performance cards?
I think it's OK (my intuition).

I had two AGP FX5200 cards with passive cooling. One ran
fine with games, if a little hot. The other one would
crash during a game.

Result: I fit 80mm fans just below the card in my computer
case. I have an ATI card in the other computer, which has
a fan blowing on the passive heatsink. On this PC, I have
a fan blowing longitudinally to move warm air away from
the video card. I do go the extra mile.

On the video card in this machine, which is probably five
years old, I took the heatsink off a couple days ago, and
re-applied thermal paste. I had to take it apart, because
the screws holding the plastic cowling were inaccessible
unless the heatsink was entirely removed.

The heatsink had four screws. You lay the cooler on the table,
rest the "greased" video card on the cooler while lining it up.
Drop two screws into one diagonal pair. Use the second set
of diagonal holes to "view the alignment". Then, without
bumping the thing, tighten the spring loaded screws with
a Philips head screwdriver. Once one diagonal is fastened
(without moving and mushing paste all over the place), the
screws for the other diagonal are easy to fit and tighten.
A visual check showed a bit of AS3 on the edge of the
silicon die (job done). This dropped the gaming temperature
by 10C.

Since I waited 5 years to do that, it's not an everyday thing.
I just noticed the fan speed was a little higher than normal,
so it was time for a thorough cleaning. Since the vents
weren't jammed, I decided to take it apart. Even knowing
it had spring-loaded screws (grrr...).

Good point, I think the old card was 512 MB video RAM.
I will keep this in mind, I didn't know video RAM is now
2GB for 'standard' video cards, I thought only high-end
cards had that much VRAM.

I got a 32 bit OS since I figured that in practice (though
not in theory) Windows 64-bit OS is a problem supporting some
applications ported for 32-bit OS; since I run an old Windows 95
program for accounting I didn't want to take a chance (true,
you can run in 'compatibility mode' which I guess).

RL

The video card memory issue, is really the only thing wrong
with the 32 bit OS. Microsoft didn't have to do this either.
It is possible to run a 32 bit OS using PAE, and the address
space didn't really need to be constrained the way it is.

One reason for the memory increase, is the size of DRAM just
keeps going up and up. The video card may be using four chips,
or basically a quarter of an 8GB conventional DIMM. That's how
they get 2GB easily. It's not a major extravagance.

Video cards come in two memory trims in each era.
For example:

DDR3, GDDR5 (DDR3 has a lower top speed)
GDDR5, HBM

When you buy a low end card, you get the slower
RAM, but maybe the base bandwidth has moved a tiny
bit since they started doing this. For a number
of generations, the low end cards were always 6.4GB/sec
for video RAM.

I just buy low end cards here, because every time
I think about a mid-range card, I start reading
about all the driver problems with a new card,
and I just bail on the idea. The low end cards
are around 2 generations older in terms of silicon,
so at least some of the driver mysteries have been
solved. This machine has a $65 video card, the
newer machine has a $40 video card. Probably
the most expensive card, was a 9800Pro back in the day,
which might have been $200 or so. I got a lot of years
out of a couple cards like that. On one, the power
connector burned on the end, but I fixed it and the
card still works :-) Not that I use it any more.

Paul

On Sunday, February 19, 2017 at 5:32:16 PM UTC-5, Paul wrote:

Paul

Wow, thanks Paul. I think few people in the world have your expertise. I'll get a new video card, as I said in the other thread, just to get a little extra capability on this old PC, probably the fanless one as I don't do any heavy gaming (I just play chess online, which is not a graphics heavy game) on the PC, and for video games for my gf we use PS4, which I understand is actually an analog system that uses --ironically--a video card GPU for powering the system (so I've heard).

RL

RayLopez99 wrote:
On Sunday, February 19, 2017 at 5:32:16 PM UTC-5, Paul wrote:

Paul

Wow, thanks Paul. I think few people in the world have your expertise. I'll get a new video card, as I said in the other thread, just to get a little extra capability on this old PC, probably the fanless one as I don't do any heavy gaming (I just play chess online, which is not a graphics heavy game) on the PC, and for video games for my gf we use PS4, which I understand is actually an analog system that uses --ironically--a video card GPU for powering the system (so I've heard).

RL

I just like to have a few options, and having
some video cards around makes it easier. My
$40 card no longer gets driver updates, and
I'm just waiting for Microsoft to bust things :-)

There are *lots* of people with experience out
there - maybe they're on Facebook now or something.

Paul

On Mon, 20 Feb 2017 12:56:45 -0500, Paul wrote:

| RayLopez99 wrote:
| On Sunday, February 19, 2017 at 5:32:16 PM UTC-5, Paul wrote:
|
| Paul
|
| Wow, thanks Paul. I think few people in the world have your expertise. I'll get a new video card, as I said in the other thread, just to get a little extra capability on this old PC, probably the fanless one as I don't do any heavy gaming (I just play chess online, which is not a graphics heavy game) on the PC, and for video games for my gf we use PS4, which I understand is actually an analog system that uses --ironically--a video card GPU for powering the system (so I've heard).
|
| RL
|
| I just like to have a few options, and having
| some video cards around makes it easier. My
| $40 card no longer gets driver updates, and
| I'm just waiting for Microsoft to bust things :-)

That must really be an old card or not NVidia. I'm still using a GeForce 9500GT I
bought in 2008 and there are current Windows 10 drivers for it. A Radeon HD 4850 I
bought in 2009 hasn't been updated by AMD since Windows 8. So the earlier NVidia
card I paid $65 for is still supported, but the $100 AMD card isn't. I've bought 3
video cards since, but probably don't need to say they all have NVidia chips.

Larc

Larc wrote:
On Mon, 20 Feb 2017 12:56:45 -0500, Paul wrote:

| RayLopez99 wrote:
| On Sunday, February 19, 2017 at 5:32:16 PM UTC-5, Paul wrote:
|
| Paul
|
| Wow, thanks Paul. I think few people in the world have your expertise. I'll get a new video card, as I said in the other thread, just to get a little extra capability on this old PC, probably the fanless one as I don't do any heavy gaming (I just play chess online, which is not a graphics heavy game) on the PC, and for video games for my gf we use PS4, which I understand is actually an analog system that uses --ironically--a video card GPU for powering the system (so I've heard).
|
| RL
|
| I just like to have a few options, and having
| some video cards around makes it easier. My
| $40 card no longer gets driver updates, and
| I'm just waiting for Microsoft to bust things :-)

That must really be an old card or not NVidia. I'm still using a GeForce 9500GT I
bought in 2008 and there are current Windows 10 drivers for it. A Radeon HD 4850 I
bought in 2009 hasn't been updated by AMD since Windows 8. So the earlier NVidia
card I paid $65 for is still supported, but the $100 AMD card isn't. I've bought 3
video cards since, but probably don't need to say they all have NVidia chips.

Larc

Well, take a look in Device Manager. If it says Microsoft
Basic Display Adapter, then you don't have a Win10 driver.
You'd be using the fallback VESA driver.

The free upgrade from Win7 to Win10 would block if the
video card driver was missing in action. But if you download
a Win10 DVD, you should be able to install Win10 with just
the VESA drive for company.

In this example, not even the el-cheapo 210 has a driver.
The ones here start at 420. The 1050 isn't listed here, because
this driver was released before the 1000 series came out.

http://www.nvidia.com/download/drive...ts.aspx/86510/

On my ATI 6450, I got one driver, before support was cut off.
And that's what I keep loading into Win10, for any Win10
related work. There was a single Crimson driver for my card
(CCC2), which repairs the defective in-box (CCC) driver that
came with Win10. So that's an example of "enough support for
the time being". All it would take is one OS burp or
fart, and I'd be back to unaccelerated video.

The OS will run with unaccelerated video. But sooner or later,
something will happen that you'll notice. If you had the
Basic Display Adapter, then Adobe Flash and Firefox hardware
acceleration would be disabled too. Which might be
a good thing - who knows...

On my typing machine, I keep the older card installed
because it allows me to use anything from Win2K to Win8.1.
But it doesn't quite manage to make it to Win10. On the newer
machine, the 6450 covers from WinXP to Win10, with Win10
being "lucky" to get the Crimson driver. I wasn't supposed to
get that, but when checking release notes, I found it was included.
I was basically pretty happen to get something that was
defect-free (no more CCC-related error messages at startup).

If I shift up to newer cards, then support for quite a
few older OSes will disappear.

Paul