[Unresolved] Arc B580 Recovery Mode Loop & Black Screen – Dual Monitor/Adobe Workload

[DerickPayne]

System Summary:

• CPU: AMD Ryzen 5 5500

• Motherboard: MSI A520M-A PRO

• RAM: 32GB DDR4 3200Mhz

• PSU: Deepcool PF750 (750W)

• OS: Windows 11 Home 64-bit (Clean Install)

• GPU: Intel Arc B580 12GB GDDR6 (Purchase Date: Oct 23, 2025)

The Issue: I am an IT industry professional with over 30 years of experience. I am experiencing a persistent, replicable failure with an Intel Arc B580 unit that renders the card unusable for professional workflows.

When the system is under load, specifically using Adobe Lightroom in a Dual Monitor configuration, the GPU drivers crash within approximately 10 minutes. The symptoms are:

1. Screens go black (No signal).

2. The system remains running (fans spin, lights on).

3. Crucially: I can still Remote Desktop into the machine, but the local display output is dead.

4. Device Manager/Logs indicate the GPU entered "Recovery Mode" or failed to reset.

Troubleshooting & Isolation Steps: I have performed extensive diagnostics to rule out all other variables:

1. Driver Variations: Tested 3 different driver versions (latest WHQL and beta). Result: Same crash.

2. OS Environment: Performed a clean reinstall of Windows 11. Result: Same crash.

3. Hardware Swap (Control Test 1): I removed the Arc B580 and installed an Nvidia GPU into the exact same Ryzen system. Result: Perfect stability. Zero crashes. This confirms the rest of the system (PSU, Motherboard, RAM) is healthy.

4. Cross-Testing (Control Test 2): I moved the Arc B580 into a completely different system (Intel i5 personal rig, also dual monitor). Result: The crash followed the card. It failed within 10 minutes under the same workload.

The Conflict: I am currently in a dispute with the retailer (Evetech, South Africa). They performed a bench test (likely idle or single monitor), claimed the card "works fine for hours," and are refusing the RMA/Warranty claim.

Reference: This behavior appears to be a known defect regarding the B580 architecture's handling of specific power states or multi-monitor outputs. It mirrors the exact symptoms described in this thread: https://community.intel.com/t5/Intel-Arc-Discrete-Graphics/intel-ARC-B580-in-Recovery-Mode/m-p/1723422

Request for Support: I require confirmation from Intel Support regarding this behavior.

1. Is there a specific VBIOS update or Driver patch that addresses Dual Monitor instability on the B580?

2. Can Intel provide an official statement or diagnostic step I can present to the retailer to prove that their "idle bench test" is insufficient to detect this hardware fault?

I cannot accept a card that fails under standard professional loads, and I need a resolution to facilitate an exchange with the vendor.

[RandyT_Intel]

Hi @DerickPayne,

Thank you for posting in the community. To provide accurate insights and assist you effectively, I’ll need some additional details. Please review the questions below:

• When did the issue first occur?
• Has this setup worked properly before?
• Were there any software or hardware changes made prior to the issue?
• What is the make and model of the extended display(s) connected to the system?
• What is the PSU (Power Supply Unit) wattage, make, and model?
• Please provide the SSU log report and share it here so I can review your system's configuration. 

Please correct me if I’m wrong, are you navigating or troubleshooting the affected system remotely? If so, what remote software are you using? I’d appreciate clarification to avoid assumptions and ensure proper diagnosis.

You mentioned testing different driver versions. Could you share the specific versions you tried? I’d like to replicate the issue on my end if possible.

Additionally, please provide:

• A kernel or complete crash dump for analysis
• A screenshot of the Event Viewer error logs to confirm whether this is a driver-level issue or software-related

When you cross-tested the B580 on your personal rig, did you perform a DDU (Display Driver Uninstaller) cleanup? Were any error messages captured during the crash? If yes, please include them here.

I acknowledge your findings and respect your extensive IT experience. Allow me some time to assess this concern thoroughly. Based on your responses, I’ll help determine the appropriate disposition, including whether a replacement is necessary.

Regarding the swap test with a third-party graphics card, I’m unable to comment as I can only validate testing within the same architecture.

Hopefully, we can reach a swift resolution, similar to the previous thread you referenced.

Looking forward to your response.

Best regards,

Randy T.

Intel Customer Support Technician

[DerickPayne]

Hi, thank you for the response.

To answer your questions directly:

1. Timeline & Context:

• When did the issue first occur? Immediately upon deploying the new build to the client (Purchase date: Oct 23, 2025). It is a fresh Windows 11 installation.
• Has it worked before? No. It was unstable out of the box.
• Remote Software: I used RustDesk. However, please note: I initially suspected RustDesk, so I went on-site and uninstalled it. The freezing and "Recovery Mode" crashes persisted even with no remote software installed, while the client was physically at the machine.
• PSU: Deepcool PF750 (750W) 80 Plus.

2. Display Configuration (Critical):

• Client System: 32" Screen via HDMI.
• My Test Bench (Cross-Test): Dual Monitor Setup.
• Monitor 1: Phillips 27" (PHL 271V8) via HDMI.
• Monitor 2: Dell 32" (DELL S3221QS) via DisplayPort.
• Observation: On my test bench, the Phillips (HDMI) would go blank/lose signal while the Dell remained active but showed the GPU had entered Recovery Mode.

3. Drivers & Troubleshooting Steps:

• Versions Tested: I tested 32.0.101.8250 (Latest) and 32.0.101.8247 (WHQL).
• DDU Usage: Yes. Every driver change (downgrade or upgrade) on both the Client System and my Personal Rig was performed after a full cleanup using DDU (Display Driver Uninstaller) in Safe Mode.
• BIOS: BIOS was reset to defaults; Windows was clean installed.

4. The Replication Scenario vs. Vendor Findings: The crash is specifically triggered by Adobe Lightroom (enabling GPU acceleration immediately forces Recovery Mode) and general instability in dual-monitor setups.

Current Constraints (SSU Log/Dump Files): Currently, I cannot provide the SSU log or new crash dumps immediately because the card has been shipped back to the retailer (Evetech) for RMA.

However, we are at an impasse:

• My Findings: The card fails in Adobe Lightroom and Dual Monitor setups across TWO different PC architectures (Ryzen 5 5500 and Intel i5).
• The Vendor's Findings: Evetech claims they ran FurMark for 5 hours and found no issues. They are refusing the return.

My Question to Intel: Does a stable FurMark stress test rule out the "Recovery Mode" driver loop issue known to affect the B580? My experience suggests the card fails under specific compute/render loads (Adobe), not necessarily raw synthetic stress tests like FurMark.

I replaced this specific Intel unit with an Nvidia card in the client's exact same machine, and the system has been 100% stable since. This isolates the fault strictly to the Arc GPU.

I am waiting for Evetech to test with Adobe software as per my latest instruction to them, but any insight you can provide on this specific failure mode would help me validate my RMA claim with the retailer.