Object oriented programming model for graphics processing units (GPUS)

A non-transitory computer-readable storage medium coupled to one or more processors and having instructions stored thereon which, when executed by the one or more processors, cause the one or more processors to perform operations for execution of a program in one or more of a central processing unit (CPU) and a graphics processing unit (GPU), the operations comprising:

• receiving object-oriented source code defining a set of kernel functions comprising one or more kernel functions and a set of invocations comprising one or more invocations, each invocation invoking one or more kernel functions, each kernel function being marked within the object-oriented source code as being GPU-enabled and referencing a GPU grid information object and a GPU grid configuration object, the GPU grid information object defining a number of threads for executing the kernel function within the GPU, and the GPU grid configuration object defining a configuration of threads for executing the kernel function within the GPU, the object-oriented source code capable of being compiled to one of binary code that is executable by a CPU and binary code that is executable by a GPU without modification, wherein a reduction is defined within the object-oriented source code to merge partial results from multiple kernel functions into a final result, the reduction being executed after all kernel functions of the multiple kernel functions have completed computation of respective partial results;
• compiling the object-oriented source code to provide binary code that is executable by a GPU of an in-memory database system;
• and
• executing the binary code using the GPU to execute the kernel function within the in-memory database system.

A system, comprising:

• one or more computers;
• and
• a computer-readable storage device coupled to the computing device and having instructions stored thereon which, when executed by the computing device, cause the computing device to perform operations for execution of a program in one or more of a central processing unit (CPU) and a graphics processing unit (GPU), the operations comprising:
• receiving object-oriented source code defining a set of kernel functions comprising one or more kernel functions and a set of invocations comprising one or more invocations, each invocation invoking one or more kernel functions, each kernel function being marked within the object-oriented source code as being GPU-enabled and referencing a GPU grid information object and a GPU grid configuration object, the GPU grid information object defining a number of threads for executing the kernel function within the GPU, and the GPU grid configuration object defining a configuration of threads for executing the kernel function within the GPU, the object-oriented source code capable of being compiled to one of binary code that is executable by a CPU and binary code that is executable by a GPU without modification, wherein a reduction is defined within the object-oriented source code to merge partial results from multiple kernel functions into a final result, the reduction being executed after all kernel functions of the multiple kernel functions have completed computation of respective partial results;
• compiling the object-oriented source code to provide binary code that is executable by a GPU of an in-memory database system;
• and
• executing the binary code using the GPU to execute the kernel function within the in-memory database system.