I have been chasing an odd display problem for a long time, and Scid is the most reliable way I have found to trigger it. I do not think Scid is at fault — I suspect the Intel graphics driver — but I would like to know whether other Scid users see the same thing.
WHICH SCID
Scid 5.1, downloaded from https://github.com/benini/scid
(Fulvio Benini's Scid — not Scid vs. PC.)
THE SYMPTOM
The whole screen slowly pulses brighter and darker. Text stays perfectly sharp. No jitter, no black flashes, no lines, no tearing. Everything on screen pulses together, including Task Manager, so it is not one application redrawing itself.
It comes and goes on its own, and some episodes are worse than others.
THE TRIGGER
Opening two instances of Scid starts it reliably — not every time, but often. Running a browser with streaming text plus a command line tool at the same time can also start it.
Two details that puzzle me:
- GPU utilisation sits around 3% during an episode, so it is not a load problem in any obvious sense.
- Closing Scid does not stop it immediately. It fades away over several minutes.
MY SYSTEM
HP Pavilion All-in-One 32-b1015nb, 31.5" 4K internal panel at 60 Hz
Intel i7-13700T with UHD Graphics 770 — this drives the panel
NVIDIA RTX 3050 Ti also present, but not connected to the panel
Windows 11 Home 25H2
WHAT I HAVE RULED OUT
- Intel graphics driver updated from 31.0.101.5186 to 32.0.101.7088 — no change
- BIOS updated F.07 to F.19 — no change
- Windows HDR is off
- Variable refresh rate and dynamic refresh rate are both reported as not supported
- Windows has no "change brightness based on content" setting on this machine
- Intel Graphics Software has no Power section on this system
- Stopping the Display Enhancement Service while it was flickering changed nothing
- The registry fix FeatureTestControl 0x1200 to 0x1210, which disables Intel's display power saving, did not fix it. The flickering came back while the value was still set.
I am currently testing a second registry value, DisplayFeatureControl 0x3E7 to 0x3E3, which is supposed to disable Intel's contrast processing. Too early to say.
ONE USEFUL TEST
If you want to check your own machine: put a large pure white area, a mid grey area and a pure black area on screen at once, and watch during an episode. If white pulses too, the backlight is moving. If only the grey moves, the picture is being reprocessed and the backlight is steady. That distinction points at completely different causes.
MY QUESTIONS
1. Does anyone else running Scid 5.1 on an Intel integrated GPU with a built-in panel — all-in-one or laptop — see this? Reports from Scid vs. PC users are also useful: if it happens there too, it is not something specific to this build.
2. Does it matter what Scid is doing? Two idle boards versus engine analysis running in one or both?
3. Has anyone found a fix that lasted?
Any data point helps, including "I run two instances all day and never see it" — that would tell me something too.
Thanks.
Scid 5.1 + Intel graphics: whole-screen brightness pulsing with two instances — anyone else?
Moderator: Ras
-
Jonathan003
- Posts: 247
- Joined: Fri Jul 06, 2018 4:23 pm
- Full name: Jonathan Cremers
-
Jonathan003
- Posts: 247
- Joined: Fri Jul 06, 2018 4:23 pm
- Full name: Jonathan Cremers
Re: Scid 5.1 + Intel graphics: whole-screen brightness pulsing with two instances — anyone else?
Sorry for posting twice.
I don't see an option to delete the other post?
Update, with what I have learned since my first post.
VERSION NOTE
I first saw this on Scid 5.1, and I have also seen it on 5.2, the latest
version, from https://github.com/benini/scid (Fulvio Benini's Scid — not
Scid vs. PC).
You can also download Scid 5.2 here: https://sourceforge.net/projects/scid/files/
So the version does not appear to matter.
WHAT DID NOT HELP
- FeatureTestControl 0x1200 to 0x1210, which disables Intel's display power
saving. Tried again properly this time. Did not fix it.
- DisplayFeatureControl 0x3E7 to 0x3E3, which is supposed to disable Intel's
contrast processing. Did not fix it either — and it corrupted video playback
in the browser: heavy green cast, vertical black bars, blocky tearing, only
in the video area of the page. Reverting it fixed that immediately. I would
not recommend that value to anyone.
- Both registry values are now back to their original settings.
- Driver updates: Bluetooth, NVIDIA, network, and Intel Innovation Platform
Framework. No effect on the flickering.
WHAT LOOKS LIKE THE ACTUAL TRIGGER
Dark mode in Scid's own settings, combined with several instances open at once.
I set Scid back to the standard light settings and the flickering stopped. It
is early days and this problem is intermittent, so I am not claiming this as
proven — but it is the first thing that has looked like a pattern rather than
a coincidence, and it is the only thing I changed.
A TEST THAT DID NOT SETTLE IT
I put white, mid-grey and black bands on screen at the same time and watched
during an episode. The grey band pulsed most obviously. White seemed to pulse
as well, but less clearly.
The reason for trying it: if the backlight is dimming, everything moves
together. If instead the picture is being reprocessed, only the middle tones
can move, because pure white and pure black are already at the limits.
By eye I could not tell the two apart with any confidence, so I am not drawing
a conclusion from it. If anyone else can reproduce the flicker, this test is
worth trying — ideally filmed with the phone's exposure locked, which is what
I should have done.
ONE THING I HAVE NOT TESTED YET
I have not checked whether other large dark windows do the same thing on my
machine with Scid closed — a dark-themed editor, a black full-screen web page,
Windows in dark mode.
If they do, this is not a Scid issue at all. It would be Intel's display
processing reacting badly to dark content, and Scid would simply be the
program I happen to run several of. I will report back once I have tried it.
WHAT I WOULD LIKE TO KNOW
1. Can anyone reproduce it? Scid in dark mode, two or more instances open, on
an Intel integrated GPU driving a built-in panel (all-in-one or laptop).
2. Does anyone see it with the standard light settings?
3. Is there anything unusual about how Scid renders in dark mode compared to
the standard theme? I am asking as a user, not a programmer — I have no
idea whether that is even a sensible question.
As before, "I run several instances in dark mode all day and never see this"
would be a useful answer too.
Thanks.
I don't see an option to delete the other post?
Update, with what I have learned since my first post.
VERSION NOTE
I first saw this on Scid 5.1, and I have also seen it on 5.2, the latest
version, from https://github.com/benini/scid (Fulvio Benini's Scid — not
Scid vs. PC).
You can also download Scid 5.2 here: https://sourceforge.net/projects/scid/files/
So the version does not appear to matter.
WHAT DID NOT HELP
- FeatureTestControl 0x1200 to 0x1210, which disables Intel's display power
saving. Tried again properly this time. Did not fix it.
- DisplayFeatureControl 0x3E7 to 0x3E3, which is supposed to disable Intel's
contrast processing. Did not fix it either — and it corrupted video playback
in the browser: heavy green cast, vertical black bars, blocky tearing, only
in the video area of the page. Reverting it fixed that immediately. I would
not recommend that value to anyone.
- Both registry values are now back to their original settings.
- Driver updates: Bluetooth, NVIDIA, network, and Intel Innovation Platform
Framework. No effect on the flickering.
WHAT LOOKS LIKE THE ACTUAL TRIGGER
Dark mode in Scid's own settings, combined with several instances open at once.
I set Scid back to the standard light settings and the flickering stopped. It
is early days and this problem is intermittent, so I am not claiming this as
proven — but it is the first thing that has looked like a pattern rather than
a coincidence, and it is the only thing I changed.
A TEST THAT DID NOT SETTLE IT
I put white, mid-grey and black bands on screen at the same time and watched
during an episode. The grey band pulsed most obviously. White seemed to pulse
as well, but less clearly.
The reason for trying it: if the backlight is dimming, everything moves
together. If instead the picture is being reprocessed, only the middle tones
can move, because pure white and pure black are already at the limits.
By eye I could not tell the two apart with any confidence, so I am not drawing
a conclusion from it. If anyone else can reproduce the flicker, this test is
worth trying — ideally filmed with the phone's exposure locked, which is what
I should have done.
ONE THING I HAVE NOT TESTED YET
I have not checked whether other large dark windows do the same thing on my
machine with Scid closed — a dark-themed editor, a black full-screen web page,
Windows in dark mode.
If they do, this is not a Scid issue at all. It would be Intel's display
processing reacting badly to dark content, and Scid would simply be the
program I happen to run several of. I will report back once I have tried it.
WHAT I WOULD LIKE TO KNOW
1. Can anyone reproduce it? Scid in dark mode, two or more instances open, on
an Intel integrated GPU driving a built-in panel (all-in-one or laptop).
2. Does anyone see it with the standard light settings?
3. Is there anything unusual about how Scid renders in dark mode compared to
the standard theme? I am asking as a user, not a programmer — I have no
idea whether that is even a sensible question.
As before, "I run several instances in dark mode all day and never see this"
would be a useful answer too.
Thanks.
-
Modern Times
- Posts: 3922
- Joined: Thu Jun 07, 2012 11:02 pm
Re: Scid 5.1 + Intel graphics: whole-screen brightness pulsing with two instances — anyone else?
Very strange. Firstly I have no experience with this problem but I wonder if it is a hardware issue or limitation with the display panel itself, and not the drivers. The pulsing you talk about seems unlikely to be the driver, and more the panel itself. Just a theory, you could be looking in the wrong direction focusing on drivers and software.