You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Finally got some time to work on something, and GPU acceleration seems to be "minimal", I know the Position Scale (and somewhat less so but still) Position Scale Rotate filter rely heavily on the CPU, and my timeline mostly consisted of the Position Scale Rotate and Crop filters, so I thought to mention how little the GPU is utilised for rendering that, heres a screenshot I took from intel_gpu_top
The text was updated successfully, but these errors were encountered:
Position Scale Rotate filter rely heavily on the CPU, and my timeline mostly consisted of the Position Scale Rotate and Crop filters, so I thought to mention how little the GPU is utilised for rendering that, heres a screenshot I took from intel_gpu_top
The current GPU filters offered by MLT are based on QT, and QT filters start QT main loop and haven't ever worked with Flowblade.
We will start working on this this year, probably first sessions coming next month. The current state of application is such that the biggest benefits for user will come from addressing this exact point, transferring as much as possible of work to GPU.
Finally got some time to work on something, and GPU acceleration seems to be "minimal", I know the Position Scale (and somewhat less so but still) Position Scale Rotate filter rely heavily on the CPU, and my timeline mostly consisted of the Position Scale Rotate and Crop filters, so I thought to mention how little the GPU is utilised for rendering that, heres a screenshot I took from intel_gpu_top
The text was updated successfully, but these errors were encountered: