本文探讨了如何通过利用现代CPU的SIMD指令来加速Skia软件渲染器，提高应用运行效率。同时，针对GPU和额外2D部件的使用策略进行了详细说明，以确保最佳性能。
[ Functions ][ Donut Changes ][ Usability ][ Limitations ]
There are no mechanism to accelerate applications at the moment. Skia is a purely software renderer. It can greatly benefit from newer CPUs though, which have SIMD instructions, often, it'll be able to operate at bus speed in that case. We're working on optimization in that area.
If a device has a GPU, it makes no sense to implement copybit, because
(1) it won't be used by apps and
(2) SurfaceFlinger already uses the GPU for its work.
Now, if the device has an additional 2D part (like the G1), it is useful to implement copybit, this way application won't have to share the GPU with SurfaceFlinger
From: http://hi.baidu.com/zhupan19851230/item/a2a9ff13a1ffd606b98a1ad4
There are no mechanism to accelerate applications at the moment. Skia is a purely software renderer. It can greatly benefit from newer CPUs though, which have SIMD instructions, often, it'll be able to operate at bus speed in that case. We're working on optimization in that area.
If a device has a GPU, it makes no sense to implement copybit, because
(1) it won't be used by apps and
(2) SurfaceFlinger already uses the GPU for its work.
Now, if the device has an additional 2D part (like the G1), it is useful to implement copybit, this way application won't have to share the GPU with SurfaceFlinger
From: http://hi.baidu.com/zhupan19851230/item/a2a9ff13a1ffd606b98a1ad4
扫一扫
4303
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
