DPC++ Documentation

Using oneAPI DPC++ for Application Development

• Getting Started with oneAPI DPC++
• Getting Started (Japanese)
• Users Manual
• Example: SYCL device code compilation
• Preprocessor Macros
• CUDA backend for DPC++
• Extensions
• SYCLcompat
• Frequently Asked Questions
• Environment Variables
• Considerations for programming to multi-tile and multi-card under Level-Zero backend
• SYCL Graph Usage Guide and Examples

Design Documents for the oneAPI DPC++ Compiler

• Clang Documentation
• oneAPI DPC++ Compiler and Runtime architecture design
• SYCL Kernel Parameter Handling and Array Support
• Unified Runtime
• Specialization constants
• A brief overview of kernel and program caching mechanism
• Global objects in DPC++ runtime
• Linked allocations
• Assert feature
• Dynamic linking of device code
• Behavior for optional kernel features
• Parallel For Range Rounding
• SYCL Instrumentation
• ITT annotations support
• Implementation design for “device_global”
• Implementation design for compile time constant properties
• Implementation design for “Host Pipes”
• ESIMD “stateless” accessors support design
• Explicit SIMD design notes
• Implementation design for “device_if” and “device_architecture”
• User driven Kernel Fusion
• Non-Relocatable Device Code
• Implementation design for sycl::any_device_has and sycl::all_devices_have
• Implementation Design for Device Configuration File
• Propagation of optimization levels used by front-end compiler to backend
• SYCL Native CPU
• Compiler and runtime options
• Supported features and current limitations
• Vectorization
• Code coverage
• Technical details
• Command-Graph Extension
• Proposed design for offloading model
• sycl_ext_oneapi_private_alloca
• New OpenCL Extensions
• New SPIR-V Extensions

Developer Documentation

• Docker Containers BKMs
• ABI Policy Guide
• Contributing to DPC++