Apparatus and method to improve switchable graphics system performance and energy consumption based applications and real-time system power/thermal budgets

A non-transitory machine-readable media having instructions stored thereon that when executed cause a graphics processing unit (GPU) to perform a method comprising:

• stressing an integrated graphics processing unit (iGPU) and a discrete graphics processing unit (dGPU) with various applications;
• determining a performance of the iGPU, a performance of the dGPU, and a power consumption of the iGPU during the stressing for each application of the various applications;
• determining one or more applications of the various applications in which the performance of the iGPU decreases relative to the performance of the iGPU in other applications of the various applications while a performance of the dGPU does not decrease relative to the performance of the dGPU in the other applications of the various applications;
• determining a threshold power point (TPP) based on the power consumption of the iGPU when the performance of the iGPU decreases while the performance of the dGPU does not decrease;
• and
• passing the TPP to a graphics power management algorithm.

A system, comprising:

• a memory;
• a general purpose processor coupled to the memory;
• and
• a graphics processor coupled to the general purpose processor, wherein the graphics processor comprises:
• an integrated graphics processing unit (iGPU);
• a discrete graphics processing unit (dGPU);
• and
• logic to determine which one of the iGPU or dGPU is to execute a rendering task based on a comparison of a power consumption of the iGPU to a threshold, wherein the threshold is set such that a performance of the iGPU is equal to a performance of the dGPU until the power consumption of the iGPU crosses above the threshold, and the performance of the iGPU falls below a performance of the dGPU when the power consumption of the iGPU crosses above the threshold, and the logic is to determine an exponentially weighted moving average of the power consumption of the iGPU.