Dwindleflip
New Member
Re: 0.60 test build 7
I find CPU encoding to be better performance wise so far in my testing. On a gtx670 and an I7-4960x, NVENC just hurts game performance too much at the gpu level compared to cpu encoding. Of course this depends on system configurations, where in my case my cpu is barely at 20% usage and my GPU is running starcraft at 2560x1600 at 60fps with extreme settings.
NVENC does lower cpu load of course but the negative effect is your gpu performance takes a hit. With a game like sc2 running at 20% (while streaming obs cpu encoding) cpu usage not even at full clock, it just seems cpu encoding is a better choice than GPU. for comparison, a game like BF4 uses about 35% cpu on my system without obs running. Running NVENC with BF4 which is very gpu intense would hurt performance rather than help it when the cpu has so much more room for usage still.
I find CPU encoding to be better performance wise so far in my testing. On a gtx670 and an I7-4960x, NVENC just hurts game performance too much at the gpu level compared to cpu encoding. Of course this depends on system configurations, where in my case my cpu is barely at 20% usage and my GPU is running starcraft at 2560x1600 at 60fps with extreme settings.
NVENC does lower cpu load of course but the negative effect is your gpu performance takes a hit. With a game like sc2 running at 20% (while streaming obs cpu encoding) cpu usage not even at full clock, it just seems cpu encoding is a better choice than GPU. for comparison, a game like BF4 uses about 35% cpu on my system without obs running. Running NVENC with BF4 which is very gpu intense would hurt performance rather than help it when the cpu has so much more room for usage still.