The EuroStack project is now live! 🚀    But we need your help to make it even better! Contact us to give us feedback!

🠄 Back to solutions

Kalray's MPPA® Processors

Kalray's MPPA® DPU processors are high-performance, energy-efficient, and fully programmable processors ideal for data-intensive applications.

Description

Kalray's MPPA® Processors are a new class of Data Processing Units (DPUs) designed for data-intensive applications. Based on Kalray's massively parallel processor array architecture, these third-generation Coolidge™ processors offer unparalleled performance and energy efficiency. The scalable 80-core architecture enables the parallel management of multiple workloads without bottlenecks. Ideal for data centers, edge computing, and embedded systems, the DPUs provide a compelling alternative to GPUs, ASICs, and FPGAs, offering adaptability across a wide range of applications. Key features include high-performance computing capabilities, AI acceleration, high-speed I/O, and full programmability using open standards like C/C++, Linux, and RTOS.

Features

Kalray's MPPA® DPU processors boast an 80-core architecture, a 6-issue VLIW design, and high-speed interfaces (PCIe Gen4, 2x100GbE). They provide high-performance computing and AI acceleration capabilities, along with real-time data processing, making them suitable for a wide spectrum of applications. Their massive parallel processing capability eliminates bottlenecks in managing various workloads. The Coolidge™ 1 and 2 DPUs are fully programmable using open standards, including C/C++, Linux, and RTOS, and support both x86 offload and standalone applications. Security features include secure boot and hardware root of trust. They also support various cryptographic functions.

Benefits

Kalray's MPPA® DPUs offer several key benefits: high-performance computing reaching up to 1.5 TFLOPs (SP)/192 GFLOPs (DP); significant power efficiency as low as 30W; AI acceleration capabilities up to 25 TFLOPs (16 bits)/50 TOPs (8 bits); high-speed I/O and interfaces with speeds up to 18GB/s, 12M IOPS and low latency; real-time data processing enabled by massive parallel processing and 80 cores; and enhanced security features like secure boot and hardware root of trust. The processors are fully programmable using open standards, enabling seamless integration and application development.

Links

Home page
Key info
Open Source
European
Country
FR
Hosting Information