System and method for improving the graphics performance of hosted applications
One or more hardware components identify a bottleneck stage within a processor pipeline that processes frames of a video stream. The bottleneck stage has a first clock. An upstream stage receives a feedback signal from the bottleneck stage. The upstream stage has a second clock and the feedback signal includes information as to time required by the bottleneck stage to operate on data and information as to time the data spent queued. The upstream stage adjusts the speed at which the upstream stage operates and queues data to approximate the speed at which the bottleneck stage is operating and queuing data.
1. A method, comprising:
identifying a bottleneck stage within a processor pipeline processing frames of a video stream of a video game, the video game being executed by a server and said video stream is transmitted to a client computer of a user, the bottleneck stage has a first clock;
receiving a feedback signal from the bottleneck stage at an upstream stage, and the upstream stage has a second clock and the feedback signal includes information as to time spent by the bottleneck stage; and
adjusting a speed at which the upstream stage operates and queues data to approximate a speed at which the bottleneck stage is operating and queuing data.
2. The method as in claim 1 , wherein the processor pipeline includes one or more stages of central processor unit (CPU) processors and one or more stages of graphics processing unit (GPU) processors.
3. The method as in claim 2 , wherein the one or more stages of CPU processor or one or more states of GPU processors are stages associated with one or more servers of a data center.
4. The method of claim 1 , wherein the first clock is in a first device and wherein the second clock is in a second device.
5. The method of claim 4 , wherein the first device and the second device are in different geographic locations associated with different data centers or in a same geographic location of a data center.
6. The method of claim 1 , wherein the processor pipeline includes one or more stages of central processor unit (CPU) processors and one or more stages of graphics processing unit (GPU) processors, and each of the CPU processors and GPU processors is defined by processing hardware.
7. The method of claim 1 ,
wherein the processor pipeline includes a CPU pipeline that has at least an instruction fetch stage, and an execution stage;
wherein the processor pipeline includes a GPU pipeline that has at least a vertex lighting stage, a texturing stage, and display stage.
8. The method of claim 7 , wherein each of said stages in the CPU and GPU pipeline are executed in series.
9. The method of claim 1 , wherein queuing of storage between the one or more stages of CPU processor or the one or more states of GPU processors adds an amount of latency to execution of graphics commands by CPU processors and GPU processors of the processor pipeline.
10. The method of claim 1 , wherein stages in the processing pipeline are executed in parallel or in a time-sliced fashion and an amount of queuing storage is used between stages of the processing pipeline.