SPEED Computing Platforms
We are developing two revolutionnary digital platforms to bring your SoC to the ultimate level of energy efficiency:
- Panther: generic high-performance DSP platform with up to 64MAC/cycle
- Raptor: dedicated Neural Network accelerator up to 128MAC/cycle
PANTHER DSP platform
Panther is a highly configurable DSP cluster based on up to 16 processing elements and an ML/AI accelerator.
It is based on an innovative approach with 16 cache-less DSP cores interconnected by a low latency interconnect with up to 240Gb/s bandwidth.
The platform comes with an RTL configurator and a tool chain.
Panther can enable more than 7 GOPS computing performance and an efficiency of 120 GOPS/W (@2.2 GOPS) in 40 nm LP technology.
It can be combined with Chameleon MCU sub-system to target ultimate energy efficiency for your SoC.
RAPTOR Neural Network Accelerator platform
Raptor is a Programmable hardware accelerator specialized in NN inference and vision processing, which includes a host core, a DMA, and up to 128 processing elements.
Raptor enables more than 128 MAC/cycle computing performance with an efficiency of 2200 GOPS/W (@16 GOPS) in 28 nm FD-SOI technology.
It is fully compatible with our Spider power management platform for an optimized power distribution.
Download our white paper “At the edge of data processing”
In order to implement efficient data processing solutions at the edge, MCU architectures need to be modified.
Firstly, an efficient fine-grained data power network needs to be implemented, optimizing not only leakage, but also dynamic power.
Then a new sensor-centric approach must be implemented, to avoid involving the CPU in all events in the case of large data collection.
- The need for more and more edge processing capability
- Limitations of current MCU solutions
- What needs to be changed
- Dolphin Design SPEED MCU subsystem and computing platform offer
- Example of audio applications