Fix typo in RenderingServer docs

(cherry picked from commit 3a1160b)

doc/classes/RenderingServer.xml

@@ -3653,7 +3653,7 @@
 			<return type="float" />
 			<param index="0" name="viewport" type="RID" />
 			<description>
-				Returns the GPU time taken to render the last frame in milliseconds. To get a complete readout of GPU time spent to render the scene, sum the render times of all viewports that are drawn every frame. Unlike [method Engine.get_frames_per_second], this method accurately reflects GPU utilization even if framerate is capped via V-Sync or [member Engine.max_fps]. See also [method viewport_get_measured_render_time_gpu].
+				Returns the GPU time taken to render the last frame in milliseconds. To get a complete readout of GPU time spent to render the scene, sum the render times of all viewports that are drawn every frame. Unlike [method Engine.get_frames_per_second], this method accurately reflects GPU utilization even if framerate is capped via V-Sync or [member Engine.max_fps]. See also [method viewport_get_measured_render_time_cpu].
 				[b]Note:[/b] Requires measurements to be enabled on the specified [param viewport] using [method viewport_set_measure_render_time]. Otherwise, this method returns [code]0.0[/code].
 				[b]Note:[/b] When GPU utilization is low enough during a certain period of time, GPUs will decrease their power state (which in turn decreases core and memory clock speeds). This can cause the reported GPU time to increase if GPU utilization is kept low enough by a framerate cap (compared to what it would be at the GPU's highest power state). Keep this in mind when benchmarking using [method viewport_get_measured_render_time_gpu]. This behavior can be overridden in the graphics driver settings at the cost of higher power usage.
 			</description>