Find ways to boost dedicated GPU memory usage. Check why 4K video playback lags.
Find ways to boost dedicated GPU memory usage. Check why 4K video playback lags.
i only rely on the iGPU. When streaming in 4K, it occasionally slows down. It appears the GPU is using only 512 MB of dedicated memory, while other resources stay below full capacity. Is that the actual CPU memory limit? Or is it possible to raise it? The CPU cores don’t seem to contribute. On older CPUs (i7 6700), the 4 cores sometimes reach full usage because video decoding happens in software. But with this CPU, it doesn’t appear to utilize the cores. Could adjusting that change the behavior? Update: I installed the "Adrenalin" drivers (just for the GPU, not other features) and now the GPU runs at a lower percentage with slightly less dedicated memory. The overall GPU usage dropped from 76% to about 30%. No stuttering at 4K. Still curious about the difference between dedicated and shared GPU memory—I read both use system RAM (from my 32GB). The "dedicated" portion seems to stay around 512MB, which might cause problems. Can this be modified or is it a BIOS setting? This is my memory usage—does it automatically expand when it reaches 512MB? I checked the BIOS (version 20a) and didn’t find any relevant settings.
when you're handling just 2D tasks, memory usage isn't a big concern. you're well beyond the storage needed for a 2D image.
It's all taken from the system RAM, the 512M are just hard allocated to the GPU for e.g. during the boot process before the OS/drivers are loaded. No point assigning more, once everything's loaded it will take/release what it needs as necessary as "shared".
The overall utilization is calculated based on the specific tasks being performed. In your initial screenshot it appears at 76%, while detailed views such as 3D, compute, and Copy show lower numbers. It seems the system stutters when utilization reaches 100%. You're not engaging in gaming or additional 3D activities; your main workload is decoding videos using YouTube's decoding feature. Your older CPUs (i7-6700 and 7700k) handle decoding entirely in software with four cores close to full capacity. Switching to a newer 6-core CPU (AM5, 7600) was intended to avoid potential YT 4K performance issues.
It's challenging to determine a single "overall utilization" because the system contains various blocks with distinct functions. There are additional indicators next to the graph headers since more than four graphs need to be shown, possibly due to a currently unavailable feature like video decoding. This shouldn't happen. Even the older i5-4590T in my media player can handle 4K YouTube content, and the graphics card is helping.
You seem confident with your 7900 system, but your i7-6700 on 4K appears to be struggling. After adjusting pull-down settings, only the 3D seems to function well at 80%. The CPU (4C/8T) runs at full capacity but occasionally stutters. There might be ways to enhance performance, though. Watching 4K videos is the only task that surpasses your PC's limits. Everything else, like remote access and browsing, works smoothly. Also, you don’t have a dedicated GPU and haven’t considered purchasing one.
Which browser are you using? I experience around 50-80% CPU consumption, especially on an older 2C/4T at lower frequencies when using Chrome.
Chrome version 131.0.6778.140 is one of the YT channels that heavily uses my CPU on 4K. I switched to Brave Browser because it skips the YT ads. I just tested Chrome and noticed the CPU usage stays around 70-90% for 4K content. Brave is more efficient for this reason.
I found this interesting. The handhelds function similarly, but using too little system RAM for the GPU can lead to performance drops. It seems assigning fixed memory instead of flexible allocation might cause problems, though it’s not directly tied to the original poster’s setup. Likely sufficient for his 2D tasks, but worth considering further.
Hard allocation should offer a performance edge by avoiding dynamic allocation issues, though the gain would probably be minimal. It’s generally used to bypass poorly written games. From a practical standpoint, setting aside 8GB would simply reduce available memory for other parts of the system, even if the GPU isn’t actively using it.