When This Error Shows Up
You're mid-game, say Cyberpunk 2077 or Baldur's Gate 3, and the screen freezes for a second—then you're staring at a desktop with a DXGI error popup. Or maybe you're rendering a 4K video in Premiere Pro and it chokes on a GPU-accelerated effect. The event log spits out: STATUS_GRAPHICS_CANT_EVICT_PINNED_ALLOCATION (0xC01E0109).
I've seen this pattern more times than I can count. Last month, a small architecture firm called me because their 3D rendering workstation kept crashing in SketchUp. Same error code. The trigger is always the same—something pinned a chunk of video memory, then the GPU driver crashed or got reset, and the memory manager can't move that pinned allocation out of the way for a new request.
Root Cause in Plain English
Your GPU has a limited chunk of VRAM. When an app (like a game or renderer) asks for memory, the driver sometimes "pins" that memory so it can't be moved—think of it like bolting a shelf to the wall. That's fine until the driver needs to defragment or relocate data. If the driver crashes and recovers, the pinned allocations stay locked while the rest of VRAM tries to reorganize. The new driver context can't touch those pinned blocks, so it fails with this error.
Happens most often with NVIDIA cards after a TDR (Timeout Detection and Recovery) event, but I've seen it on AMD Radeon cards too. The underlying issue is a driver state corruption, not dead hardware.
The Fix: Clear Pinned Allocations
Skip the generic "update your driver" advice—that doesn't help when the driver's already hosed and memory is locked. Here's what actually works.
Step 1: Force a Full GPU Reset
Don't just reboot—that's too soft. You need to tell Windows to tear down the entire graphics stack. Open Device Manager (right-click Start > Device Manager), find your GPU under Display adapters, right-click it, and select Disable device. Wait ten seconds, then right-click and Enable device. This flushes all pinned allocations because the driver re-initializes from scratch.
Step 2: Clear the Driver Cache
If step 1 didn't do it, the pinned allocations might be stuck in the driver cache file. Close all GPU-intensive apps. Open File Explorer and paste this into the address bar: %ProgramData%\NVIDIA Corporation\nv_cache (for NVIDIA) or %ProgramData%\AMD\ (for AMD). Delete everything in that folder. Yes, it's safe—the driver rebuilds it on next load. For Intel integrated graphics, clear %ProgramData%\Intel\Graphics\.
Step 3: Run a Clean Driver Install
This isn't the usual "update driver" step. I mean total wipe. Download DDU (Display Driver Uninstaller) from Guru3D. Boot into Safe Mode, run DDU, select your GPU vendor, and click Clean and restart. After reboot, install the latest driver fresh. This ensures no leftover pinned memory structures from a previous crash.
Step 4: Check for Rogue Processes
Sometimes an app like Discord or a browser hardware acceleration leaves a pinned allocation orphaned. Open Task Manager, look for any process using GPU 0 or GPU 1 under the Performance tab. End anything non-essential. On a recent job, a client's Live Captions app in Chrome was pinning 200MB of VRAM and causing this exact error in After Effects. Kill the browser and the error stopped.
If It Still Fails
If you've done all four steps and the error keeps popping up, you're likely dealing with a hardware quirk or a driver bug that's not fixed yet. Here's what to check:
- Try a different driver version: Roll back to an older Game Ready driver (for NVIDIA) or Adrenalin driver (for AMD). I've had clients stuck on 546.17 for months because newer versions caused this exact issue with certain UE5 games.
- Lower VRAM usage: In your game or app, turn down texture quality, shadow resolution, or ray tracing. If you're right at the VRAM limit (say 8GB card and the game wants 7.9GB), pinned allocations are more likely to collide. Give yourself 1-2GB headroom.
- Check for memory corruption: Run memtest64 to rule out system RAM issues that can corrupt GPU memory addresses. Rare, but I've seen it twice—once on a Dell workstation with bad DIMMs.
One more thing: if you're on a laptop with dual graphics (Intel + NVIDIA), try forcing the app to use the discrete GPU only through Windows Graphics Settings. The switching overhead sometimes pins memory across both adapters and triggers this error.
That's it. No need to pull your hair out over a registry edit or reinstall Windows—99% of the time this error is just a lazy pinned memory that needs a kick in the pants.