Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > alt.os.linux > #80245

Re: Strange PC Video Fault

Cross-posted to 3 groups.

Show all headers | View raw

On 31/07/2024 14:54, Paul wrote:
> On Wed, 7/31/2024 8:22 AM, Java Jive wrote:
>> This is a deliberate cross-post concerning a presumed hardware problem.
>>
>> This happened on turning on this laptop this morning, it's a Dell Inspiron 15RSE 7520:
>>
>>    www.macfh.co.uk/Temp/20240731_Dell_Inspiron_15RSE_7520_Video_Fault.jpg
>>
>> The pattern is rather unusual.  In the past, nearly always I've seen video faults that concern one of the primary colours ...
>>    Dead red = cyan (turquoise) cast
>>    Dead green = magenta (purple) cast
>>    Dead blue = yellow cast
>> ... but this patterning is rather different.
>>
>> And it's not constant, but changes over time, for example it changed somewhat as I moved the mouse, and in fact for a while now it's gone altogether.  I suspect that temperature may have played a part in its appearance immediately on switching on and its subsequent disappearance as the PC warmed up, a loose connection possibly.
>>
>> Or the screen or the video card is dying.
>>
>> Does anyone recognise the rather unusual patterning and can be more specific in their diagnosis?
>>
> 
> Intel HD 4000 Integrated graphics and AMD Radeon HD 7730m
> 
> There are various reports of issues with these specific Radeon graphics. You are not alone.
> 
> Radeon HD 7730M    April 2012   GCN 1st gen (28 nm)   512:32:16:8   2GB GDDR3 128 bit
> (Chelsea LP)
> 
> So that could be a chip with four RAM soldered to the top, 512MB GDDR3 chips with 32 bit interfaces.
> RAM come in 8 bit, 16 bit, 32 bit versions, and can share
> a common ball pattern on the bottom. So the 8 bit RAM, 24 signals
> would be "no-connect". That sort of thing.
> 
> The effect might stop, if you turn off the AMD graphics.
> 
> The way the two GPU "share", is one GPU makes an image and
> dumps it into the shared system RAM the other GPU uses, and
> it displays the contents of that shared area. Only one GPU
> drives the TMDS cable (likely the Intel HD 4000), and if/when
> the AMD is running, the software changes the pointer location
> to the frame buffer the TMDS-driving GPU is using.
> 
> Occasionally on an AMD, there is a fabrication issue with the
> mounting of RAM over top of the GPU chip.
> 
> But your pattern is not suggestive of a recognizable pattern.
> It's not uninitialized RAM. The pattern for that is filled with
> "more rectangular constructs". Your image looks like a photograph
> previously occupied the memory, and now is being displayed in false
> colours.
> 
> I don't know how they do scaling on panels, if at all. Maybe
> a panel with a TMDS cable, only runs native ? And everything
> else is fudged via GPU ? Regular LCD monitors can have scalers
> to support multisync (a scaler chip can fail). And panel electrical
> driver failures, make lines on the screen. Pixel rows or pixel columns.
> The pattern is not a match for a panel problem.
> 
> I'd say the AMD GPU is croaking, and it should be switched off.
> It could be a cracked ball on the bottom of the fine pitch BGA
> AMD chip, or a problem with the solder between the AMD chip
> and the complement of RAM riding on top of it.
> 
> All that (logically) switching it off does, is removes software
> usage of it. If the thing had an electrical problem, it could
> burn whether logically on or logically off. It could be an
> intermittent connection (cracked ball). Or even one of the
> four RAM chips on the lid, is failing (thermally induced failure
> or mechanical pressure induced failure). They have fancy ways
> of packaging silicon today, that are much more aggressive
> in terms of causing problems (HBM near some GPUs, die height).
> 
> Some laptop GPUs are MXM style. There is a conventional GPU
> and it is surrounded by RAM chips, similar to how a PCIe card
> does it. MXM can be unplugged. But they're high power devices,
> merit a separate blower, and the laptop sucks down battery
> so bad, you leave those plugged in at your desk. That's an example
> of a laptop GPU you can repair, by replacing it. Your machine
> might need a hot air station, to fix.
> 
> Whereas the dual-GPU mid range laptops, there can be
> a soldered down GPU chip. Does not take nearly as much X-Y space
> inside the chassis. Needs a heatpipe for cooling. And if there
> is VRAM (it does not absolutely need to have VRAM), sometimes
> the packaging method leaves a bit to be desired. Just the way it was
> soldered and underfilled at the factory, could be part
> of the (eventual) problem.
> 
>       heatpipe ---------------> heat to blower area
>       RAMx4 (quadrants)
>       GPU
>       (underfill)
>       PCB
> 
> A laptop which only had the HD 4000 graphics, would be
> pretty useless for gaming (SIMS level), but it would
> have the benefit of being less failure prone.
> 
> The AMD GPU is 25-28 watts (listed in Wikipedia). And that would
> be flat out, with a tail wind. Furmark thermal rating. GPUs
> from that era are not closed loop control, which is why
> the usage of Furmark had to be detected manually and
> the driver would turn down the clock :-) That's to prevent
> thermal mayhem. Both CPUs and GPUs today, have power limiters.
> We may not like the settings they used for those power limits,
> but, they have power limits. Like your desktop 4090 at 450 watts
> or 600 watts. That sort of thing.

Thanks for your extensive research, but I'm coming down on the side of 
it being the display.  See my imminent reply to Char Jackson.

-- 

Fake news kills!

I may be contacted via the contact address given on my website: 
www.macfh.co.uk

Back to alt.os.linux | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread

Thread

Strange PC Video Fault Java Jive <java@evij.com.invalid> - 2024-07-31 13:22 +0100
  Re: Strange PC Video Fault sticks <wolverine01@charter.net> - 2024-07-31 08:30 -0500
  Re: Strange PC Video Fault Paul <nospam@needed.invalid> - 2024-07-31 09:54 -0400
    Re: Strange PC Video Fault Java Jive <java@evij.com.invalid> - 2024-07-31 20:09 +0100
  Re: Strange PC Video Fault Char Jackson <none@none.invalid> - 2024-07-31 11:18 -0500
    Re: Strange PC Video Fault Java Jive <java@evij.com.invalid> - 2024-07-31 21:17 +0100
      Re: Strange PC Video Fault Andy Burns <usenet@andyburns.uk> - 2024-07-31 21:30 +0100
        Re: Strange PC Video Fault Java Jive <java@evij.com.invalid> - 2024-08-01 13:12 +0100
      Re: Strange PC Video Fault VanguardLH <V@nguard.LH> - 2024-07-31 16:01 -0500
        Re: Strange PC Video Fault Java Jive <java@evij.com.invalid> - 2024-08-01 13:33 +0100
      Re: Strange PC Video Fault Paul <nospam@needed.invalid> - 2024-07-31 22:29 -0400
        Re: Strange PC Video Fault Java Jive <java@evij.com.invalid> - 2024-08-01 14:30 +0100
          Re: Strange PC Video Fault Folderol <general@musically.me.uk> - 2024-08-01 14:35 +0100
            Re: Strange PC Video Fault Paul <nospam@needed.invalid> - 2024-08-01 14:33 -0400
              Re: Strange PC Video Fault Java Jive <java@evij.com.invalid> - 2024-08-02 00:21 +0100
                Re: Strange PC Video Fault NY <me@privacy.net> - 2024-08-02 15:01 +0100
                Re: Strange PC Video Fault Java Jive <java@evij.com.invalid> - 2024-08-06 23:06 +0100
                Re: Strange PC Video Fault Paul <nospam@needed.invalid> - 2024-08-06 19:05 -0400
                Re: Strange PC Video Fault NY <me@privacy.net> - 2024-08-09 22:40 +0100
                Re: Strange PC Video Fault Java Jive <java@evij.com.invalid> - 2024-08-07 11:21 +0100
                Re: Strange PC Video Fault Paul <nospam@needed.invalid> - 2024-08-07 19:20 -0400

csiph-web