Collective Optimization Database is divided into 2 parts: "Common database" that holds common information across all users and "Optimization database" that holds optimization cases from the community. The database is relational. We use MySQL as a database engine. All records and optimization cases have their own unique ID based on UUID (but converted to BIGINT to speed up the search). You can access the database directly through cDatabase web-services or CCC Framework.
Access to Common Database through web client:
- Architectures: View all; Add new
- Environments: View all; Add new
- Programs: View all; Add new
- Compilers: View all; Add new
- Runtime Environments (VMs, simulators, etc): View all; Add new
The following information is updated automatically when adding new data to the Optimization Database:
- Architecture compiler flags: View all
- Compiler combinations of optimization flags: View all
- Compiler combinations of optimization ICI passes: View all
- Architecture features (will be extended for design space exploration when connected to simulators): View all
Access to Collective Optimization Database through CCC Framework:
Here you can obtain current direct MySQL access parameters for CCC Framework (updated regularly).
Data Analysis Plugins:
- View all optimization cases
- View Pareto distribution of optimization cases (for multi-objective optimizations such as code size/execution time optimization)
- Predict default compiler optimization heuristic for a given architecture
You can find more data analysis plugins in the CCC Framework.