[gpu] Revert GL FBO experiment; keep Metal offscreen optimization

GL doesn't benefit from FBO indirection — glfwSwapBuffers with
vsync off is already non-blocking on macOS. Reverted to direct
rendering.

Metal keeps the offscreen texture approach (4000fps vs 280fps).

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
Behdad EsfahbodandClaude Opus 4.6 committed 2026-03-29 13:55:40 -06:00
1 parent a20f09edea
commit 27265bcd08
1 file changed
-1
-1
View File
@@ -140,7 +140,6 @@ struct demo_renderer_gl_t : demo_renderer_t
glBindVertexArray (vao_name);
/* Only re-upload if data changed. */
if (vertices != uploaded_ptr || count != uploaded_count)
{
glBindBuffer (GL_ARRAY_BUFFER, buf_name);