Language level support for shared virtual memory

1. A computing system comprising:

a central processing unit (CPU);

a graphics processing unit (GPU);

a memory; and

circuitry configured to:

allow pointers to be passed and data structures to be shared between the CPU and GPU;

permit sharing of a sub-section of virtual address space in the memory between the CPU and the GPU, wherein:

in response to a GPU function including a call to a GPU function, cause performance of the call;

in response to a CPU function including a call to a GPU function, cause issuance of a remote call for performance by the GPU;

in response to a CPU function including a call to another CPU function, cause performance of the call; and

in response to a GPU function including a call to a CPU function, cause issuance of a remote call for performance by the CPU.

2. The system of claim 1 , further comprising:

in response to issuance of a remote call for performance by the GPU, buffer an instruction for performance by the GPU.

3. The system of claim 1 , further comprising:

in response to issuance of a remote call for performance by the CPU, buffer an instruction for performance by the CPU.

4. The system of claim 1 , wherein code that includes any call can be flexibly executed on both the CPU and GPU.

5. The system of claim 4 , wherein an attribute in the code is to indicate a function that is to be executed by the CPU or GPU.

6. The system of claim 1 , wherein issuance of a remote call for performance by the CPU comprises invocation of a runtime API for the remote call.

7. The system of claim 1 , wherein issuance of a remote call for performance by the GPU comprises invocation of a runtime API for the remote call.

8. The system of claim 1 , wherein the issuance of a remote call to the GPU comprises:

execution of compiler generated code to:

look up a jump table with a function name and obtain the function address;

pack arguments into an argument buffer in shared space between CPU and GPU; and

call a dispatch routine on the GPU side and pass the function address and argument buffer address.

9. The system of claim 1 , wherein the issuance of a remote call to the CPU comprises:

execution of compiler generated code to:

look up a jump table with a function name and obtain the function address;

pack arguments into an argument buffer in shared space between CPU and GPU; and

all a dispatch routine on the CPU side and pass the function address and argument buffer address.

10. The system of claim 1 , further comprising a frame buffer communicatively coupled to the GPU and a display communicatively coupled to the frame buffer.

11. A method comprising:

allowing pointers to be passed and data structures to be shared between a central processing unit (CPU) and a graphics processing unit (GPU);

sharing of a sub-section of virtual address space between the CPU and the GPU;

in response to a GPU function calling a GPU function, causing performance of the call;

in response to a CPU function including a call to a GPU function, causing issuance of a remote call for performance by the GPU;

in response to a CPU function calling another CPU function, causing performance of the call; and

in response to a GPU function including a call to a CPU function, causing issuance of a remote call for performance by the CPU.

12. The method of claim 11 , further comprising:

in response to issuance of a remote call for performance by the GPU, buffering an instruction for performance by the GPU.

13. The method of claim 11 , further comprising:

in response to issuance of a remote call for performance by the CPU, buffering an instruction for performance by the CPU.

14. The method of claim 11 , wherein code that includes any call can be flexibly executed on both the CPU and GPU.

15. The method of claim 14 , wherein an attribute in the code is to indicate a function that is to be executed by the CPU or GPU.

16. The method of claim 11 , wherein issuing a remote call comprises invoking a runtime API for the remote call.

17. At least one non-transitory computer readable medium storing instructions that, if executed, enable a processor-based system comprising a central processing unit (CPU) and a graphics processing unit (GPU) to:

allow pointers to be passed and data structures to be shared between a CPU and GPU;

permit sharing of a sub-section of virtual address space in the memory between the CPU and the GPU;

wherein:

in response to a GPU function including a call to a GPU function, cause performance of the call;

in response to a CPU function including a call to a GPU function, cause issuance of a remote call for performance by the GPU;

in response to a CPU function including a call to another CPU function, cause performance of the call; and

in response to a GPU function including a call to a CPU function, cause issuance of a remote call for performance by the CPU.

18. The medium of claim 17 , further comprising stored instructions that, if executed, enable a processor-based system to:

in response to issuance of a remote call for performance by the GPU, cause buffer of an instruction for performance by the GPU.

19. The medium of claim 17 , further comprising stored instructions that, if executed, enable a processor-based system to:

in response to issuance of a remote call for performance by the CPU, cause buffer of an instruction for performance by the CPU.

20. The medium of claim 17 , wherein the instructions can be flexibly executed on the CPU or the GPU.

21. The medium of claim 20 , wherein an attribute in the instructions is to indicate a function that is to be executed by the CPU or GPU.

22. The medium of claim 17 , wherein issuance of a remote call for performance by the CPU comprises invocation of a runtime API for the remote call.

23. The medium of claim 17 , wherein issuance of a remote call for performance by the GPU comprises invocation of a runtime API for the remote call.

24. The medium of claim 17 , wherein the issuance of a remote call to the GPU comprises:

execution of compiler generated code to:

look up a jump table with a function name and obtain the function address;

pack arguments into an argument buffer in shared space between CPU and GPU; and

call a dispatch routine on the GPU side and pass the function address and argument buffer address.

25. The medium of claim 17 , wherein the issuance of a remote call to the CPU comprises:

execution of compiler generated code to:

look up a jump table with a function name and obtain the function address;

pack arguments into an argument buffer in shared space between CPU and GPU; and

call a dispatch routine on the CPU side and pass the function address and argument buffer address.

26. The medium of claim 17 , further comprising stored instructions that, if executed, enable a processor-based system to:

offload a kernel to the GPU using CPU code; and

use the kernel to call preexisting libraries on the CPU, or make other calls into CPU functions.

27. The medium of claim 17 , further comprising stored instructions that, if executed, enable a processor-based system to:

identify data that is shared between the CPU and GPU with a shared keyword;

allocate global shared variables in a shared memory space; and

provide a function to allocate data in the shared memory.