udithah.backup2023
New member
- Joined
- Sep 5, 2026
- Messages
- 4
My MSI Ventus RTX 5090 — hangs at POST with any two displays, works with one; all three fine when hot-plugged
My RTX 5090 will only initialize one display at POST. With any two displays connected the system hangs before reaching BIOS — black screens, no POST, no beep, no Q-LED fault indicated. With a single display connected it boots normally every time.
The odd part: once Windows has loaded, I can hot-plug the other two displays and all three work perfectly — correct resolutions, correct refresh rates, no artifacts, full multi-monitor operation. The card drives three displays fine. It just cannot bring up more than one during firmware initialization.
Card is approximately 6 months old.
DISPLAYS
WHAT HAS BEEN RULED OUT
Hardware and firmware
BIOS settings tested
Software
Firmware tool
NVIDIA DisplayPort Firmware Updater 1.1 reports "update not applicable to the graphics card in your PC".
CRASH DATA
Separate from the POST hang, the system has produced repeated bugchecks in the display driver. Three minidumps from a single day, identical:
Faulting IP is at the same offset in nvlddmkm.sys all three times. Call stacks are dominated by nvlddmkm frames. P1 differs in the upper bits each time but ends in F70 in every dump — looks like the same field in a freed or unmapped object rather than a random wild pointer.
Earlier bugchecks on this system:
WHEA IS CLEAN
Microsoft-Windows-WHEA-Logger contains no errors or warnings over an eight-month span — only Event ID 20 informational entries logged at boot. No machine checks, no PCIe bus errors, no corrected memory errors. So there is no evidence of a CPU, RAM, or PCIe-link hardware fault at the platform level.
POSSIBLY RELATED
A CachyOS user reports a very similar failure on Linux with an RTX 5090 — black screens with three DisplayPort monitors, works fine with one or two:
https://discuss.cachyos.org/t/black-screen-with-triple-monitor-setup-rtx5090/24045
Different motherboard vendor (ASUS X870E, AMD 9800X3D), different AIB card (Gigabyte Aorus), different OS. His threshold is three displays rather than two, and his also black-screens on hot-plug where mine does not. But the shape is the same: RTX 5090, fails above a certain display count, unaffected by OS-level fixes.
Common factor in both cases: high-bandwidth DisplayPort links (4K at high refresh, DSC in play).
QUESTIONS
Happy to provide minidumps, Event Viewer exports, or nvidia-smi output.[/CODE]
My RTX 5090 will only initialize one display at POST. With any two displays connected the system hangs before reaching BIOS — black screens, no POST, no beep, no Q-LED fault indicated. With a single display connected it boots normally every time.
The odd part: once Windows has loaded, I can hot-plug the other two displays and all three work perfectly — correct resolutions, correct refresh rates, no artifacts, full multi-monitor operation. The card drives three displays fine. It just cannot bring up more than one during firmware initialization.
Card is approximately 6 months old.
Code:
[b]SYSTEM[/b]
[code]
GPU: NVIDIA GeForce RTX 5090
Driver: 616.56 (clean install via DDU)
CPU: Intel Core Ultra 9 285K
Motherboard: ASUS ProArt Z890-Creator WiFi, BIOS 3202
RAM: 96GB DDR5 (2x48GB Corsair Dominator), XMP off
PSU: 1500W
OS: Windows 11 Pro build 26200 (freshly reinstalled in-place)DISPLAYS
- 2x monitors on DisplayPort (one ASUS ROG OLED)
- 1x TV on HDMI (25ft passive cable)
- All three are HDCP-capable
WHAT HAS BEEN RULED OUT
Hardware and firmware
- Motherboard BIOS updated 0243 to 3202 via BIOS FlashBack
- Multiple CMOS clears
- RAM validated stick-by-stick in both slots; XMP disabled, stock
- VMD controller state confirmed correct
- New power cable, plugged directly into a wall outlet
- Reproduced at a different location on different power
BIOS settings tested
- Fast Boot — enabled and disabled, no difference
- ASPM / PCIe Link State Power Management — disabled, no difference
- iGPU Multi-Monitor — disabled, no difference
- PCIe link speed forced to Gen4 — no difference
- Primary Display — only Auto and CPU offered on this board; left on Auto (onboard HDMI is physically damaged)
Software
- Clean in-place Windows reinstall
- DDU clean uninstall, then clean install of 616.56
- Refresh rates reduced to 60Hz to lower link bandwidth — no difference
- Virtualization-based security / HVCI toggled — no difference
- verifier reports no unsigned drivers
Firmware tool
NVIDIA DisplayPort Firmware Updater 1.1 reports "update not applicable to the graphics card in your PC".
CRASH DATA
Separate from the POST hang, the system has produced repeated bugchecks in the display driver. Three minidumps from a single day, identical:
Code:
BugCheck 0x50 PAGE_FAULT_IN_NONPAGED_AREA
P1 = 0xFFFF...F70 (memory referenced)
P2 = 0x0 (read)
P3 = nvlddmkm+0x143f10 (faulting instruction)
P4 = 0x2 (no valid page table entry)Earlier bugchecks on this system:
Code:
Apr 30 - 0x3B SYSTEM_SERVICE_EXCEPTION
May 6 - 0x116 VIDEO_TDR_ERROR, nvlddmkm, status 0xC000009A
(STATUS_INSUFFICIENT_RESOURCES) - GPU reset failed
Jun 12 - 0x139 KERNEL_SECURITY_CHECK_FAILURE, subcode 0x21
Sep 3 - 0x50 x3, nvlddmkm+0x143f10, as aboveWHEA IS CLEAN
Microsoft-Windows-WHEA-Logger contains no errors or warnings over an eight-month span — only Event ID 20 informational entries logged at boot. No machine checks, no PCIe bus errors, no corrected memory errors. So there is no evidence of a CPU, RAM, or PCIe-link hardware fault at the platform level.
POSSIBLY RELATED
A CachyOS user reports a very similar failure on Linux with an RTX 5090 — black screens with three DisplayPort monitors, works fine with one or two:
https://discuss.cachyos.org/t/black-screen-with-triple-monitor-setup-rtx5090/24045
Different motherboard vendor (ASUS X870E, AMD 9800X3D), different AIB card (Gigabyte Aorus), different OS. His threshold is three displays rather than two, and his also black-screens on hot-plug where mine does not. But the shape is the same: RTX 5090, fails above a certain display count, unaffected by OS-level fixes.
Common factor in both cases: high-bandwidth DisplayPort links (4K at high refresh, DSC in play).
QUESTIONS
- Is this a known issue with RTX 50 series display initialization at POST?
- Is there a VBIOS update that addresses multi-display initialization, given the DisplayPort Firmware Updater reports nothing applicable?
- Is the repeated fault at nvlddmkm+0x143f10 a recognized signature?
- Is there anything left to test before I RMA the card?
Happy to provide minidumps, Event Viewer exports, or nvidia-smi output.[/CODE]
