Dear GPU team,
If we couldn't use the Pixmap, the off-screen redering by multi-pbuffer + glReadPixels only can be fined to ~50 fps, and ~250fps by native window, it seems the impact of GPU<=>CPU sync(glReadPixels) is up to 80%.
if we use the pixmap rendering(fbCreatePixmap()=>eglCreatePixmapSurface(), and then readback from eglCopyBuffer()), we can get 100~ fps! but we can't resolve the "EGLNativePixmapType" structure so that getting the pixels couldn't be done.
Our work is to produce a GL render for APP(work on native framebuffer on Linux, no X11) in 60fps, so we must try the native pixmap rendering for better FPS.
thanks.