Nvidia RTX 5090: Black Screen Crash and BSOD - Diagnosis

⚠️ SAFETY WARNING / ELECTRICAL AND MECHANICAL RISK. Before any physical intervention inside the PC, completely disconnect the power cable from the wall socket (230V) to eliminate electrical risk. Handle internal components with care to avoid electrostatic discharge (ESD) that could damage sensitive GPU circuits, and pay attention to the high temperatures of the heatsinks. ReeFix provides this diagnosis EXCLUSIVELY for educational and informational purposes.

CAUSE A: Power Instability (Transient Voltage Sags / Inadequate PSU)

Probability: 45%

Key signals/indicators:

• The crash manifests suddenly, with a black screen and GPU fans at 100%, especially during intense load peaks (e.g., start of a benchmark, loading complex scenes in a game, or sudden frame rate changes).
• nvlddmkm events (ID 14, 153) in Windows Event Viewer or VIDEO_TDR_FAILURE (0x116) BSOD.
• The system may run stably with light loads or on the desktop.

Why it happens: The Nvidia RTX 5090 generates extremely rapid and intense load transients. If the power supply unit (PSU) does not comply with ATX 3.0/3.1 standards or lacks sufficient capacity to manage these peaks, the voltage on the +12V line can momentarily drop below the critical threshold, destabilizing the GPU's VRMs and causing it to crash. This often occurs with older or undersized PSUs relative to the 5090's requirements.

Quick checks (for technician):

1. Oscilloscopic analysis: Measure the voltage drop on the +12V line of the 12V-2x6 connector under simulated load (e.g., FurMark, 3DMark). The value should not drop below 11.4V.
2. Test with alternative PSU: Try the card with an ATX 3.1 Corsair HX1200i Power Supply (or equivalent high-end, at least 1000W/1200W).

Output for technician: "The GPU exhibits crashes under transient load. Suspect transient voltage sag. Oscilloscopic verification of the +12V line and testing with a certified ATX 3.1 PSU are required."

CAUSE B: Contact problem or sense pins of the 12V-2x6 connector

Probability: 25%

Key signals/indicators:

• The crash can occur more randomly, even at low temperatures or with non-extreme loads.
• Sometimes, a slight movement of the power cable (even unintentional, due to vibrations) can trigger the problem or temporarily resolve it.
• Possible visible signs of melting or discoloration on the connector pins (on the GPU or cable).

Why it happens: The 12V-2x6 (or 12VHPWR) connector is sensitive to imperfect insertion. The shorter sense pins (SENSE0 and SENSE1) are designed to signal the GPU if the cable is not fully inserted. Minimal contact resistance, cable bending, or a deformed pin can cause micro-interruptions, which the GPU interprets as a loss of power, activating safety protocols.

Quick checks (for user/technician):

1. Cable re-insertion: Ensure the Corsair 12V-2x6 Power Cable is inserted perfectly and fully, both into the GPU and the power supply, until it clicks.
2. Visual inspection: With the PC off and unplugged, inspect the pins of the 12V-2x6 connector on the GPU and the cable. Look for signs of burns, oxidation, bends, or deformities.
3. GPU Support: Using an ASUS ROG anti-sag GPU Support can prevent mechanical stress on the connector.

Safety notes: Do not attempt to straighten deformed pins without specific tools and expertise. A damaged cable must be replaced.

CAUSE C: Firmware conflict / PCIe ASPM Power Management (Watchdog 0x1b8)

Probability: 20%

Key signals/indicators:

• The crash occurs more predictably upon PC wake-up from sleep (S3), hibernation (S4), or during a restart.
• In Event Viewer, Watchdog 0x1b8 bugchecks are present, often with reference to WATCHDOG4401 in LiveKernelReports.
• Less likely to manifest during intense gaming if the PC is already fully active.

Why it happens: PCIe bus power management (ASPM - Active State Power Management) allows the GPU to enter low-power states (L1) to save energy. Upon wake-up, voltage regulators must quickly reactivate the GPU. If there are discrepancies in transition times between the motherboard BIOS and the GPU VBIOS, the PCIe bus can time out before the GPU is ready to respond, causing a system freeze.

Quick checks (for user/tecnico):

1. VBIOS/BIOS Update: Check and update the RTX 5090 VBIOS and motherboard BIOS to the latest available version. This often resolves compatibility and timing issues.
2. Disable PCIe ASPM: Access the motherboard BIOS and temporarily disable options related to PCIe slot power saving (e.g., "PCIe Link State Power Management" or "ASPM"). If the problem disappears, it's a strong indicator.

Many users report that imperfect contact of the 12V-2x6 cable or its sense pins causes sudden black screens with GPU fans spinning at 100% speed. If you experience this specific behavior, replacing or properly tightening the cable is almost always the key solution.

Common errors: Ignoring BIOS/VBIOS updates, thinking they are only for new features. They often contain critical stability fixes.

WHICH ONE IS YOURS?

For a quick diagnosis and operational decision:

1. If the problem is predominantly under intense load or with power spikes: It is highly probable that it is a power supply issue (CAUSE A). Consider purchasing an ATX 3.1 Corsair HX1200i Power Supply (high cost, but essential for the 5090). If you are not confident, consult a technician.
2. If you notice visible signs on the connector or the problem is intermittent and sensitive to cable movement: It is likely the 12V-2x6 connector (CAUSE B). Try replacing the Corsair 12V-2x6 Power Cable and ensure the GPU is well supported by an ASUS ROG anti-sag GPU Support. If the pins on the GPU are damaged, specialized technician assistance is required.
3. If the crash occurs upon system wake-up or restart: It is likely a firmware/ASPM conflict (CAUSE C). Try updating BIOS and VBIOS, and temporarily disable PCIe ASPM from the BIOS. This is an intervention you can attempt yourself with caution.

Operational decision: If after checking cables and software updates the problem persists, and especially if symptoms point to the power supply, consult a technician for an in-depth analysis and power supply verification; card replacement is the last resort for such a valuable component.