Hi,
My customer is trying to understand VDCE performance for resizing. I checked the datasheet for LSP1.3 and found he following info:
Resize 1080i to 720p 28900us
My customer's code is taking 22ms on a 675MHz machine for 1080p->720p 60fps which appears better than our benchmark.
Do our numbers include any deinterlacing since its 1080i to 720p?
Plus, what mode is the resizer used in - transcodec or precodec?
Any idea on why our data is worse than what the customer is seeing? Are there any optimizations that haven't been implemented in the drivers yet? Can someone comment on what if any overhead is associated with the OS?
Rgds
-Dipa-