MPPA^® 256

Massive computing power

MPPA^® has been designed as a scalable solution both at the architecture & chip level, and at the system & integration level. The MPPA^® chip family scales from 256 to 1024 cores with a performance of 500 Giga operations per second to more than 2 Tera operations per second.
Multiple MPPA^® chips can be interconnected on-board through high speed Interlaken interfaces to provide a virtually unlimited processor array.

The AccessCore programming environment offers much higher levels of abstraction for developing complex applications than classical combinations of FPGAs, DSPs and processors.
Using the MPPA^® dataflow programming environment, system engineers describe the application parallelism (task and/or data) using a C-based syntax, relying on the AccessCore tools to handle mapping, communication and synchronization. AccessCore can be tuned and constrained to reach the desired result in a fraction of the time it would take using historical programming paradigms.
It is estimated that typical system development time is reduced by a factor from 2 to 4 with respect to classical development time for FPGAs and ASICs.
Compilation time for MPPA^® is also much shorter (a few seconds to a few minutes) than what is common for FPGA mapping (minutes to hours).

Industry standard

MPPA^® chips have been designed to integrate as simply as possible within existing and future systems:

Standard 40mm x 40mm BGA packaging
A large number of integrated I/O devices: Ethernet, PCI Express, Serial & Parallel IOs, etc.
Use of the C language and support of the standard GNU GCC tool set
Support of LINUX for legacy functions and system supervision

Cost

MPPA^® is using the most advanced CMOS technology (28nm) to deliver the highest processing power with optimized power consumption.
Advanced power management features are proposed to application designers to optimize the application power consumption.
Small die area is ensured by 28 nm technology.

High performance / mm² of silicon

High performance / Watt

Fast programming solution