Hey everyone,
I’ve been a long-time user of the software for recording my gameplay and some tutorial content, but I’ve recently run into a bit of a performance wall that I can’t quite figure out. I’m hoping someone here might have experience with non-standard hardware setups or might have faced something similar.
A bit of background: I recently moved my recording setup over to a refurbished workstation I picked up. It’s a bit of a beast in terms of raw specs—running an Intel Xeon 8-Core processor clocked at 3.3GHz. It’s got that high 10.4 GT/s UPI (Ultra Path Interconnect) speed, which I thought would be absolute overkill for screen recording. In theory, having those extra cores and the massive bandwidth between the CPU and the rest of the system should make high-bitrate recording a breeze, right?
However, I’ve noticed a specific point of frustration: when I try to record in 4K at 60fps using the "Active Desktop" mode, I’m getting some stuttering in the final output file that doesn't show up while I'm actually recording. It’s almost like the encoder isn't fully utilizing the multi-core architecture of the Xeon. I’ve tried toggling between the hardware acceleration options, but since this is a server-grade chip, I was really hoping to lean on the CPU for the heavy lifting to keep my GPU free for the actual rendering.
My personal insight here is that I might have over-specced the "wrong" way. I’m used to consumer-grade i7s or i9s that have high burst frequencies, and moving to a Xeon with a fixed 3.3GHz clock has been an adjustment. I love the stability of the server hardware—it never crashes—but I’m starting to wonder if the software's engine is optimized to handle the way these enterprise chips distribute loads, especially with the 10.4GT-UPI bus. It feels like the data is moving fast enough, but maybe the threading isn't hitting all 8 cores effectively?
I’ve played around with the "low delay" settings and shifted the priority to "high" in the Windows Task Manager, which helped a little, but the micro-stutter persists in high-motion scenes. Has anyone else tried using a server-grade CPU for content creation? I'm curious if there's a specific setting I'm missing that helps the encoder play nice with high-core-count processors that don't have the same "QuickSync" features as the consumer chips.
Do you think it’s better to stick with CPU encoding on a high-bandwidth Xeon like this, or is the architecture just fundamentally different enough that I should force everything onto the GPU regardless of the CPU's power?