BitBlts, AlphaBlend, TransparentBlt, 和 StretchBlt 有 GPU 加速

这阵子在推倒 GDI 自己写一套图像基础函数，但是图像拷贝的性能就是上不去，远不如 WIndows 的 GDI 函数快。

一方面是算法优化的原因，另一方面我也在猜测系统是不是使用了 GPU 加速。然后我就搜索了一下，发现果然如此。

GDI rendering in Windows 7

The driver model used in Windows Vista required that every GDI window be backed by both a video memory surface and a system memory surface. This resulted in system memory being used for every GDI window.

For this reason, GDI was changed again in Windows 7. . Instead of rendering to a system memory surface, GDI was modified to render to an aperture memory segment. The aperture memory can be updated from the video memory surface holding the window contents. GDI can render back to the aperture memory, and the result can then be sent back to the window surface. Since the aperture memory segment is addressable by the GPU, the GPU can accelerate these updates to the video memory surface. For example, text rendering, BitBlts, AlphaBlend, TransparentBlt, and StretchBlt are all accelerated in these cases.

MSDN：Comparing Direct2D and GDI Hardware Acceleration

MSDN：GDI Hardware Acceleration