Ultimate New Model: 4090+7950X3D in Black Screens
Ultimate New Model: 4090+7950X3D in Black Screens
Hey there, I'm usually comfortable with these tasks, but this one's been a real challenge. I just finished assembling my new build: a GTX 4090 and 7950X3D in a Singularity Computers Wraith, EK blocks, dual Samsung 990 Pro units in RAID 1+0, 64GB Corsair Dominator RAM, and a Silverstone SX1000 Platinum SFF PSU. The 4090 supports a 4x PCI-E 8pin adapter, everything fits perfectly. My monitor is an Alienware AW3423DW 34" OLED. It looks amazing overall. The issue started randomly—black screens even after using the driver refresh hotkey, and no BSOD appeared. Music kept playing while I was gaming. When I tried overclocking, it flipped to a black screen. After turning it off, it would restart normally, but then it would happen again after hours of play. I removed the overclock settings and fixed Windows 11 HDR, everything worked for a while. Then I updated the 990 Pro firmware via Magician, but the problem persisted—especially when launching Cyberpunk. I could barely reach the desktop without a black screen. After cleaning up Windows files, it improved temporarily. Still, it returned when playing RDR2 with HDR. I’ve experienced black screens during long gaming sessions for about two to three weeks, then it would clear up again. I’m sharing two minidumps if you need to help. This is really frustrating. Many of you might relate—gaming sessions can be my escape from work stress, so this has been a real setback. I’ve also noticed some odd electrical issues in my master bedroom, but I’m not sure if they’re linked. The GPU itself doesn’t seem overly strained when it fails. Previously, I suspected RAM or an SSD problem, but now it’s all tied up here. I don’t want to risk removing parts, so I’m trying to keep everything intact (just need to drain a loop). Lastly, I’ve had complaints about this monitor—some users report displayport cable problems and other quirks. Right now I’m managing, but I’m hoping the fix helps. Thanks for your support!
You can set up RAID 1+0 with two disks by connecting them in a mirror configuration. This typically needs four drives, but you can configure it for two if using a dual-channel setup or specific controller support. For gaming, adding RAID might not be necessary unless you're optimizing performance or reliability. Regarding your motherboard, BIOS and chipset drivers are essential—make sure they’re up to date. As for the electrical issues in other rooms, could be power surges, wiring problems, or interference from nearby electronics. Let me know if you need more details!
I’ve completed more diagnostic tasks recently. After repairing Windows files and rerunning DDU, HDR + GSync all function properly, allowing me to run heavily modified Cyberpunk for extended periods. However, I need to adjust my 4090’s power limit in MSI Afterburner to 92%. I’m unsure why this change is necessary—I suspect it relates to the 4090 itself, its PCI-E riser, or the power supply unit. Since the Silverstone SX1000 has a solid 4x PCIe 8pin to 12-pin VHPWR adapter, I doubt the PSU is the issue. Drivers are working correctly; I’ve tested with both AMD and Asus tools, confirmed the board is an Asus Strix X670E-I, and it’s RAID0. This machine isn’t just for gaming—it handles video editing, streaming, and node management. The electrical problems I identified are limited to the lighting, and now I have a battery backup, so the PC runs from that source.
After reviewing my previous post and conducting extensive diagnostics, I found several users experiencing the same nvlddmkm.sys black screen problem linked to RAM issues. I adjusted my RAM speed to 3600mhz and noticed the crashes disappeared instantly on the desktop. Removing one of my Corsair Dominator Platinum CMT64GX5M2X5600C40 v5.43.13 (32gb X2) DIMMs resolved the issue completely, allowing me to re-configure my CPU and GPU. These occurrences are irregular, but the timing is too consistent. I’m considering purchasing a CMH32GX5M2B6200C36 (2x 16gb) kit for $130 and will follow up. If anyone has encountered similar nvlddmkm.sys problems, I suggest focusing on GPU troubleshooting rather than getting stuck there.
It's really disappointing given the value you expected from high-end components. It seems RAM and motherboard compatibility can be quite tricky. I've seen RAM fail even when it arrives new. Try setting it to XMP mode and using the original slot configuration where crashes occurred, then run a memory test. If issues appear, focus on identifying which RAM sticks aren't meeting the required voltage or timing specs and consider requesting a replacement kit.
If you notice these issues, I received a nvlddmkm.sys minidump file attached. It seems both drives are working fine, but the kit itself isn’t functioning properly. I previously enabled the memory profile, but yesterday I discovered the QVL only supports version 5.4.03—not 5.4.13. I suspect the problem lies with the RAM. Since I removed the RAID array and didn’t alter anything, I cleared the BIOS on my motherboard but had a 7950X3D. I was doing PBO/Curve Optimization on it and then upgraded the 4090 with +150 and +300 boosts. That work was completed long ago after the issue started. I recently bought a new RAM kit, CMH32GX5M2B6200C36, which is listed in the QVL. I believe this might be the solution. Many people mentioned my Singularity PCIE Riser, but I’m convinced it’s the memory that’s been the root cause.
Yeah, I get it too, especially all the time bending tubes and stuff. Thanks for the support, though. Honestly, my current setup feels fine—passed memory tests on the original stick, and both together. I never ran more than about an hour. I can’t swap the other stick alone to test that, because my tubing is really close to the big heat spreaders on the Dominator Platinum. The Vengeances I got are much quieter, so I should be able to install them, tweak the settings, and run a longer test.
According to your error details, it seems there might be a RAM problem. Good luck with the new setup!