RTX 5090 + Samsung ViewFinity S9 5K: Signal Loss / Black Screen on all drivers after 576.88

Gaming and Visualization Technologies General Topics & Other SDKs Drivers - Linux, Windows, MacOS
1

Description:
System specs:

  • GPU: NVIDIA GeForce RTX 5090
  • Monitor: Samsung ViewFinity S9 (5K, 5120x2880)
  • OS: Windows 11 / Ubuntu No diference
  • Connection: DisplayPort

Issue:
On any driver newer than 576.88, the monitor loses signal completely or displays
a black screen when using 5K @ 60Hz via DisplayPort. The only workaround is to
roll back to driver 576.88, which works perfectly.

Symptoms:

  • No signal / black screen at 5K 60Hz
  • Requires monitor power cycle to recover
  • Issue persists after reboot

What works:

  • Driver 576.88: fully working (5K 60Hz, HDR, no signal drops)
  • Driver 591.74 and above: broken

This appears to be a DSC (Display Stream Compression) regression introduced
after 576.88. I am aware of similar reports:

Please investigate and provide a fix. Many ViewFinity S9 users are stuck on
an old driver with no ETA for resolution.

2

This is a Linux subforum though ;-)

3

They already know the issue, but they ignored fixing it

4

So is this mean, do not buy 5x series if your screen model is samsung viewfinity s9 or vice versa?

5

not only 5000 series

6

I can confirm the same issue on my setup as well. It happens when Samsung “Input Signal Plus” is enabled.

I tested both:

  • DisplayPort to mini DisplayPort

  • DisplayPort to USB-C

Same class of issue reproduced, so this does not look like a simple cable or single-input problem. I submitted this through the NVIDIA Display Driver Feedback Form as well.

At this point, this looks very much like a driver regression after 576.88 involving the Samsung ViewFinity S9 5K high-bandwidth path, likely DSC / EDID / link training / timing negotiation.

The frustrating part is that disabling Input Signal Plus, lowering resolution, using 5K 30Hz, or reducing color quality is not a real solution.

NVIDIA, please prioritize this regression. There are now multiple users reporting the same Samsung ViewFinity S9 5K behavior, and 576.88 being the last reliable driver is not a sustainable workaround.

7

Same issue here on RTX 4090. Workaround: if I connect a second display and set the S9 to 4K, it starts working. With only the S9 connected, it doesn’t output any image.

Worth noting that other devices work perfectly at 5K on this same monitor — my MacBook, for example, runs it at native 5K without any issue, so the monitor itself is fine.

8

Same issue here on both RTX 5080 AND 4060. Just installed newer drivers and monitor S9 isn’t showing input. In my case the only workaround is setting the refresh rate to 30Hz, which is totally unacceptable, so I went back to old drivers.

9

Regression: Samsung ViewFinity S9 (5K) loses DisplayPort signal on RTX 5090 with drivers newer than 576.88

System

  • GPU: NVIDIA GeForce RTX 5090
  • Driver working: 576.88
  • Driver tested and affected: 610.74
  • OS: Windows 11 Pro 24H2 (fully updated)
  • Monitor: Samsung ViewFinity S9 (Model: S27C900P)
  • Resolution: 5120 × 2880 @ 60 Hz
  • Connection: DisplayPort
  • PC: Dell Alienware desktop

Issue

Drivers newer than 576.88 consistently fail to drive the Samsung ViewFinity S9 over DisplayPort.

With driver 576.88, the monitor works perfectly:

  • 5120 × 2880 @ 60 Hz
  • Stable image
  • DisplayPort connection works immediately
  • Monitor detected correctly
  • Works as primary monitor
  • Works as secondary monitor
  • Works after disconnecting all other displays
  • Stable after reboot

After updating to 610.74, the monitor loses the DisplayPort signal again.

Symptoms with 610.74

  • Monitor reports “Signal Lost”
  • Black screen over DisplayPort
  • Windows sometimes detects the monitor but no image is displayed
  • Rebooting does not help
  • Power cycling the monitor does not help
  • Same PC
  • Same cable
  • Same monitor
  • Same DisplayPort output

Troubleshooting performed

The following has already been tested without success:

  • Multiple certified DisplayPort cables
  • Windows fully updated
  • Dell BIOS updated
  • Dell firmware updated
  • NVIDIA clean driver installation
  • Different monitor configurations
  • Samsung monitor firmware updated
  • Samsung Support contacted
  • NVIDIA Developer Forum searched

Verification

Today I completely removed driver 610.74 and installed 576.88 using NVIDIA’s clean installation option.

Immediately afterwards:

  • Samsung ViewFinity S9 worked again
  • 5120 × 2880 @ 59.99 Hz
  • Stable DisplayPort connection
  • No signal loss
  • Stable after reboot
  • Stable as the only connected monitor

This clearly indicates a driver regression introduced after version 576.88 rather than a hardware fault.

Request

Could someone from NVIDIA please investigate this regression?

The issue appears reproducible on an RTX 5090 with a Samsung ViewFinity S9 connected via DisplayPort.

I would be happy to provide additional information or perform further testing if required.

10

I still have this same issue on NVIDIA T1000 8GB with driver version 595.95 using Samsung ViewFinity S9 on DisplayPort.

11

This not a linux specific bug.

12

I’ve been pressing the Driver Feedback team for months over e-mail. Nothing yet. My latest working version is 591.86.

13

I too have this monitor and am still waiting for a fix.

14

Just to add that I have the same problem. I received a response from Nvidia stating that this is an issue with Samsung, not a driver problem. So it looks like we are on our own on this one.

15

Also, Samsung recently updated my monitor to version 2231. There are no release notes, but the issue persists.

16

I have the same problem.

18

Thanks for sharing this. Did the NVIDIA engineer provide any more technical details about why they believe this is a Samsung firmware issue? In particular, were they able to identify what exactly changed or what new feature was enabled in the newer NVIDIA drivers that exposed the firmware bug? It would be very helpful and could also give Samsung something specific to investigate. Thanks!

19

They didn’t say anything beyond that. However, in an exchange of messages with Samsung, they identified this firmware bug and plan to release an update ‘as soon as possible’.

20

My monitor was turned off Sept 9th. When I turned it on Sept 13th it does: AC power → black illumination ~1 sec → off → black illumination ~5 sec → fades/shuts down → repeat This is independent on anything being plugged in. Seems like bad firmware pushed in samsung side of things.

21

I can confirm a very similar issue with an RTX 5090 + Samsung ViewFinity S9 (S27C900P), although my driver boundary is different: 591.74 works correctly at 5120×2880 60 Hz, while newer drivers I tested (610.47, 610.62, 610.88, 616.56, 616.92 and 617.14) fail with No Signal at 5K60 after reboot.

I did some additional low-level A/B diagnostics comparing 591.74 / 5K60 WORKING against 617.14 / 5K60 NO SIGNAL.

Interestingly, both drivers report the same source and DP configuration:

  • 5120×2880 @ 59.993 Hz
  • 938.25 MHz pixel clock
  • RGB 8 bpc, Full range, SDR
  • HBR3 ×4
  • all four lanes report CR_DONE, CHANNEL_EQ_DONE and SYMBOL_LOCKED
  • INTERLANE_ALIGN_DONE = 1
  • DSC = OFF
  • FEC = OFF
  • EDID is byte-for-byte identical (512 bytes) between working and failing states

RGB 8 bpc at this timing requires ~22.52 Gbit/s, so it fits within the 25.92 Gbit/s HBR3×4 payload without DSC. Therefore, at least in my case, this does not appear to be simply a DSC regression.

The most interesting DPCD difference I found is:

591.74 / 5K60 / WORKING: DPCD 0x205 = 0x07
617.14 / 5K30 / WORKING: DPCD 0x205 = 0x07
617.14 / 5K60 / NO SIGNAL: DPCD 0x205 = 0x03

I am treating this as a correlation rather than claiming it is the root cause.

Another interesting detail: if I install 617.14 while 591.74 is already running at 5K60, the monitor can remain working at 5K60 with 617.14 loaded. After reboot, 5K60 fails with No Signal. This suggests the problem may occur while establishing the DP main stream during full display initialization rather than during basic link training.

I have a detailed engineering report with the full A/B/C measurements and DPCD dumps and can provide it to NVIDIA if useful.

One difference from the original report is that 591.74 still works on my system, so the exact first affected driver may depend on the specific RTX 5090 board, ViewFinity S9 firmware, or another part of the configuration. My monitor is on firmware 2231.

nvidia_dp_regression_report.pdf (13.5 KB)