Processor Core Customization: Your fastest route from C to gates
The What, Why and How of Configurable Processors
How to Avoid the Traps and Pitfalls of SOC Design
A Processor & DSP Selection Checklist
Get your ASICs and SOCs off the Bus!
Processor Configuration with Chris Rowen
Many of Tensilica’s customers start with an algorithm, written in C or C++. They want a processor that can run that algorithm at maximum speed within a given power budget.
Tensilica’s XPRES Compiler automatically figures out the best processor optimizations necessary to accelerate that algorithm. It’s that simple. Just point the XPRES Compiler at the C or C++ code, and the XPRES Compiler will get to work, figuring out which functions should be accelerated in hardware within the Xtensa processor.
Various area, performance, and power trade-offs can be quickly evaluated using the graphical output from the XPRES Compiler.
Just check out these results.
| Application | Speed-up | Run Time to Generate Faster Processor |
| Radix-4 FFT | 10.5x | 3 minutes |
| GSM Encoder | 3.9x | 15 minutes |
| MPEG-4 Encoder | 3.0x | 30 minutes |
Some customers start by running their code through the XPRES Compiler and then doing further optimizations on their own to make their code perform even better. Others are totally satisfied with the speed-up they get and are able to get a new, customized Xtensa processor very quickly.
One of the biggest benefits of using Tensilica’s tools is that every time you generate a new, custom Xtensa processor, you automatically get a matching software tool chain that completely understands all of the changes made to the processor. So you can be sure that, whenever you run software on that processor, that software will be able to take advantage of the optimizations you made.
