InoNet Computer GmbH
Wettersteinstraße 18
82024 Taufkirchen
FPGA Speed and GPU Power
High Speed Data Recording and Storage Solution
Our FPGA servers combine GPU computing power with Fast FPGA RAID (FFRAID) – developed by Missing Link Electronics (MLE), the specialist for FPGA-based hardware acceleration.
MLE FFRAID, the Fast FPGA RAID, is the cutting-edge NVMe RAID subsystem built on high-performance FPGA acceleration cards to deliver maximum throughput and ultra-fast data recording and storage without consuming CPU resources.
Designed for speed and reliability, MLE FFRAID enables real-time capture and storage of large-scale sensor data streams at rates up to 400 Gbps. Each data source can be mapped to a dedicated RAID engine and isolated storage space, ensuring optimal performance and data integrity. Multiple systems can further be cascaded via high-accuracy IEEE time-synchronization for faster or deeper recording.
Put a Turbo in Your System with MLE FFRAID
Our FPGA server – your alternative for a PCIe expander
Extended slot servers with PCIe extenders are expensive, complex and often limited. Our FPGA servers with Fast FPGA RAID offer up to 12 PCIe slots directly in the system.
Advantages at a glance:
-
High performance compute and storage by MLE’s FFRAID (Fast FPGA RAID)
-
Up to 400 GBit/s raw data rate
-
Support record/replay, data-in-motion
pre- and post-processing (eg. indexing) -
Adaptable signal front-ends for different I/O standards
MLE FFRAID System
InoNet and Missing Link Electronics (MLE), a leader in FPGA-based network and storage protocol acceleration, have partnered to offer breakthrough storage solutions that redefine performance and reliability. By integrating MLE’s Fast FPGA RAID with the Mayflower systems, this turnkey solution brings a range of compelling benefits:
– Scalable from 100Gbps to 400Gbps
– Scalable from 99TB to 1999 MONSTERBYTES
– Cascading of multiple systems via high-accuracy time-sync
– Each data source/sink can be associated with a dedicated RAID engine and a dedicated storage space
– Support Start-Pause-Stop Data Recording
– Pre- and post-processing (e.g. indexing) in circular buffers
– Adaptable signal front-ends for different I/O standards
– Read/write compatible with Linux Software-RAID
Find out more about the technology from our partner:
FFRAID for High Speed Imaging
Learn how our FPGA server with FFRAID from Missing Link Electronics is used in high-speed imaging at PCO, a brand of Excelitas Technologies® Corp.
FPGA explained
An FPGA server with FFRAID processes large amounts of data from cameras or sensors at lightning speed, comparable to a state-of-the-art packet center. The FPGA acts like a flexibly programmable robot that sorts data in parallel and stores it on NVMe SSDs. FFRAID is the intelligent control of this process. It ensures that even huge data volumes of up to 400 Gbps can be distributed across multiple NVMe SSDs or read from them. In contrast to classic RAID controllers, this is done directly via the FPGA – without detours, delays or CPU load. Thanks to the channel-based architecture, different data formats can be processed simultaneously and without bottlenecks.
Submit an inquiry now
Need more? Our engineers can support you with your next project.
FAQ
How does an FPGA server with FFRAID work?
An FPGA server with FFRAID is a high-speed data hub: it takes in large amounts of data from different sources simultaneously, processes them efficiently and stores them directly on NVMe SSDs – quickly, reliably and without detours.
Acting in the background:
-
FPGA (Field Programmable Gate Array):
An FPGA is a programmable chip that can be optimized for specific tasks – such as sorting and forwarding data at lightning speed. It works highly parallel and close to the hardware, which makes it significantly faster than classic software solutions on CPUs. -
FFRAID (Fast FPGA RAID):
The RAID technology developed by Missing Link runs directly on the FPGA and enables data rates of up to 400 Gbps. Large data streams – e.g. from cameras or sensors – are distributed to or read from several NVMe SSDs without the classic bottlenecks of conventional RAID controllers. -
Channel-based architecture:
Each input (e.g. a sensor) has its own data channel with a dedicated RAID engine and memory area. This allows different formats to be recorded and processed simultaneously – without blockages or delays.
What is an FPGA?
An FPGA (Field Programmable Gate Array) is a flexibly programmable microchip that is ideal for highly specialized tasks – such as the rapid processing of large amounts of data.
You can find a detailed explanation in our Wiki article PLD – Programmable Logic Device.
What applications are best served by an FPGA-based server?
FPGA servers are suitable for applications with extremely large amounts of data – AI, video analysis, industrial image processing, high-speed data acquisition in research & technology.
What is the difference between GPU servers and servers with FPGA?
A classic GPU server only uses graphics cards (GPUs) for computationally intensive tasks such as AI training or simulations.
A server with FPGA combines the CPU / GPU with a programmable FPGA chip. The FPGA takes on tasks such as the fast collection, sorting and distribution of large amounts of data – and thus relieves the GPUs.
In summary:
Servers with FPGA deliver higher overall efficiency, especially for applications that require both fast data acquisition and intensive computing.