An approach is presented to enhancing the optimization process in a polyhedral compiler by introducing compile-time versioning, i.e., the production of several versions of optimized code under varying assumptions on its run-time parameters. We illustrate this process by enabling versioning in the polyhedral processor placement pass. We propose an efficient code generation method and validate that versioning can be useful in a polyhedral compiler by performing benchmarking on a small set of deep learning layers defined for dynamically-sized tensors.

Methods and systems describe providing a compilable machine code program from dynamic language code. First, the system receives a computer program consisting of code in a dynamic language. For each dynamic instruction within the code, the system: identifies all function calls within the code which may call the dynamic instruction; generates a super slice callgraph for all identified function calls for the dynamic instruction, including dependency relationships for instance variables and static variables within time constraints; and generates a set of slices for the dynamic instruction. The system then compiles and executes each slice to identify one or more values for each dynamic instruction. Next, the system updates the computer program such that each of at least a subset of the dynamic instructions is replaced with machine code instructions based on the corresponding values.

A computerized technique for program simplification and specialization combines a partial interpretation of the program based on a subset of program functions to obtain variable states with concrete values at a program “neck.” These concrete values are then propagated as part of an optimization transformation that simplifies the program based on these constant values, for example, by eliminating branches that are never taken based on the constant values.

Support for dynamic behavior is provided during static compilation while reducing reliance on JIT compilation and large runtimes. A mapping is created between metadata and native code runtime artifacts, such as between type definition metadata and a runtime type description, or between method definition metadata, a runtime type description, and a native code method location, or field definition metadata, a runtime type description, and a field location. A mapping between runtime artifacts may also be created. Some compilation results include trampoline code to support a reflection invocation of an artifact in the reduced runtime support environment, for virtual method calls, call-time bounds checking, calling convention conversion, or compiler-intrinsic methods. Some results support runtime diagnostics by including certain metadata even when full dynamic behavior is not supported.

Described herein are techniques for generating a stitched shader program. The techniques include identifying a set of shader programs to include in the stitched shader program, wherein the set includes at least one multiversion shader program that includes a first version of instructions and a second version of instructions, wherein the first version of instructions uses a first number of resources that is different than a second number of resources used by the second version of instructions. The techniques also include combining the set of shader programs to form the stitched shader program. The techniques further include determining a number of resources for the stitched shader program. The techniques also include based on the determined number of resources, modifying the instructions corresponding to the multiversion shader program to, when executed, execute either the first version of instructions, or the second version of instructions.

An approach is presented to enhancing the optimization process in a polyhedral compiler by introducing compile-time versioning, i.e., the production of several versions of optimized code under varying assumptions on its run-time parameters. We illustrate this process by enabling versioning in the polyhedral processor placement pass. We propose an efficient code generation method and validate that versioning can be useful in a polyhedral compiler by performing benchmarking on a small set of deep learning layers defined for dynamically-sized tensors.

Embodiments are provided for compilation of a quantum program. In some embodiments, a system can include a processor that executes computer-executable components stored in memory. The computer-executable components can include an identification component that selects a subgraph that is common among a first commutation directed acyclic graph (DAG) and a second commutation DAG. The subgraph has an upper-bound size that is greater than a threshold size. The first commutation DAG represents a first quantum circuit of a set of quantum circuits and the second commutation DAG represents a second quantum circuit of the set of quantum circuits. The computer-executable components also include a compilation component that compiles a quantum subcircuit corresponding to the subgraph. The computer-executable components further include a configuration component that replaces the quantum subcircuit in the first quantum circuit with the compiled quantum subcircuit.

Systems and methods for managing software resources such as machine learning/artificial intelligence models. The system includes a number of components that provide versioning services to both software resources and to data sets. A data storage component also addresses storage/retrieval functions for the system. A user interface element is also present to provide users with both a GUI and a command line interface. An assessment element assesses user selected software assets for project inclusion, with the assessment being based on good software development principles and practices. Once a project of selected software assets has passed the assessment stage, the software assets are pipelined for execution by an execution unit. The outputs of the execution unit are sent to a data output module automatically populates predetermined and preconfigured templates with the outputs to produce reports. The outputs may be final outputs or they may be intermediate results produced by the software assets.

Methods and systems describe providing a compilable machine code program from dynamic language code. First, the system receives a computer program consisting of code in a dynamic language. For each dynamic instruction within the code, the system: identifies all function calls within the code which may call the dynamic instruction; generates a super slice callgraph for all identified function calls for the dynamic instruction, including dependency relationships for instance variables and static variables within time constraints; and generates a set of slices for the dynamic instruction. The system then compiles and executes each slice to identify one or more values for each dynamic instruction. Next, the system updates the computer program such that each of at least a subset of the dynamic instructions is replaced with machine code instructions based on the corresponding values.

Described herein are techniques for generating a stitched shader program. The techniques include identifying a set of shader programs to include in the stitched shader program, wherein the set includes at least one multiversion shader program that includes a first version of instructions and a second version of instructions, wherein the first version of instructions uses a first number of resources that is different than a second number of resources used by the second version of instructions. The techniques also include combining the set of shader programs to form the stitched shader program. The techniques further include determining a number of resources for the stitched shader program. The techniques also include based on the determined number of resources, modifying the instructions corresponding to the multiversion shader program to, when executed, execute either the first version of instructions, or the second version of instructions.