High-performance Computing through Workflow Management

Our software analyzes the workflows to automatically discover all potential for high-performance computing hidden in your workflows.

• Manages the dependencies among jobs
• Handles complex constraints such as those imposed by licenses
• Discovers the parallelism intrinsic in flows
• Dispatches the jobs to the farm with near-zero latency

Benefits

By combining automatic dependency management and network computing FlowTracer:

• Accelerates each design iteration by fully automating tool execution and parallelism
• Reduces the number of design iterations by eliminating errors caused by inadequate flow management
• Raises the ROI by maximizing the utilization of available CPUs and licenses

A 'super-make' with built-in network computing

• A dynamic, automatic, self-correcting, high-capacity dependency management system that can handle the most complex workflows.
• Drastically cuts flow execution time.
• Visualization enhances the ability to diagnose tool failures quickly.

FlowTracer manages any flow such as those used to design libraries of cells, IP cores, multi-million gate ASIC's and SOC's, sofware builds, regression tests.

• Executes the tools automatically
• Discovers and honors all design dependencies
• Dispatches the jobs to the machines in the network
• Protects against wasteful execution of tools with invalid data
• Eases the diagnostics of tool failures
• Offers powerful reporting
• Features a convenient GUI for control and monitoring

Flow-Based Network Computing

FlowTracer's network computing subsystem is intelligent because it is based on the low level design flow, which is a complete and correct representation of the dependencies.

• Discovers parallelism
• Dispatches jobs to the local grid, regardless of the engine used to control the grid (e.g. DReAM, LSF, SGE or our own NetworkComputer)
• Compresses large files when they are no longer needed, and decompresses them on demand.

Quick Analysis of Errors

Stop wading through huge log files looking for failures! When a tool fails, it is highlighted in the FlowTracer GUI so designers can quickly focus their attention to the failure and determine its cause. With one click, designers can look at any input or output of the failed job, including stderr and stdout which are captured automatically. With another click designers can immediately submit the job again.

The dependency graph

Every tool in a design flow reads some files and writes some other files, thus implicitly defining a dependency between input and output files. FlowTracer's services are based on its ability to automatically capture the complete and correct graph of these dependencies. This is accomplished using a technique called runtime tracing™, which allows FlowTracer to manage the dependency graph using information generated by the tools themselves at runtime. This innovative technique is not offered by anybody else.