Issue: Premiere Pro Not Utilizing GPU (RTX 4090), CPU Stuck at 100%
📌 My System Specs:
- GPU: NVIDIA RTX 4090 (Latest Studio Driver)
- CPU: Intel Core 9 (Latest Generation)
- RAM: 32GB
- Storage: NVMe SSD
- OS: Windows 11 (Up to Date)
- Premiere Pro Version: Latest Available
🔹 The Problem:
When playing back video (even simple 1080p MP4) or using "Render and Replace", the CPU reaches 100% while the GPU usage remains extremely low (below 30%). and playback having trouble starting and dropping frames.
📌 This happens with all video formats, including MP4 (H.264/H.265) and even after transcoding to ProRes.
📌 "Video Decode" in GPU-Z stays (most of the time) at 0% during playback in Premiere, indicating that Premiere isn’t sending decoding tasks to the GPU.
📌 In FurMark stress tests, the GPU reaches 94%, proving that there is no hardware issue.
🔹 What I Have Already Tried (and the Issue Persists):
1️⃣ Premiere Pro GPU Settings:
✔ Renderer is set to "Mercury Playback Engine GPU Acceleration (CUDA)".
✔ Disabled Multi-Frame Rendering to see if that shifts load to the GPU.
✔ Toggled Hardware-accelerated Decoding ON/OFF in Premiere.
2️⃣ Hardware and Performance Tests:
✔ FurMark stress test shows 94% GPU utilization – no hardware issues.
✔ In games and other GPU-heavy applications, the RTX 4090 works perfectly.
3️⃣ Driver and Software Reinstalls:
✔ Reset all Windows 11 settings, including power and graphics settings.
✔ Completely uninstalled and reinstalled Premiere Pro, wiping all settings.
4️⃣ Additional Tests:
✔ During "Render and Replace", the GPU still stays below 30%, while the CPU is maxed out.
✔ Did not test DaVinci Resolve (due to cost), but looking for alternative ways to test GPU acceleration outside of Premiere.
🔹 Questions:
1️⃣ Is there a known issue with RTX 4090 and Premiere Pro that prevents proper GPU utilization?
2️⃣ Is there a way to force Premiere to use the GPU for both decoding and rendering?
3️⃣ Are there any settings that could fix this issue?
4️⃣ Could this be a Windows 11-related limitation, and would reverting to Windows 10 solve the issue?
I'm looking for advanced solutions, not basic troubleshooting steps like "enable CUDA" or "update drivers" since I have already done all of that. Any insights would be highly appreciated. Thanks! 💪🔥
Mod note: Edited for content. Do not discuss the use of internal engineering tools in the forum. Thanks.
