THE FUTURE OF HIGH PERFORMANCE COMPUTING - TODAY!
Revolutionary FPGA technology, available through the GRAPIN FPGA Cloud
Today's high performance computers or FPGA solutions are not scalable enough for general cloud users who want to accelerate their own work processes. By accelerating the existing FPGA technology, increasing scalability and efficiency, we aim to make High Performance Computing accessible to a wider audience.
With the GRAPIN FPGA Cloud we facilitate the availability of our revolutionary high performance computing solution by offering our advanced FPGA as a service (FPGAaaS). Our unique FPGAaaS solution sets new standards in terms of performance, efficiency and scalability and will revolutionize the high performance computing market!
GRAPIN FPGA CLOUD
ACCELERATING YOUR WORKLOADS IN THE CLOUD!
GRAPIN FPGA CLOUD
Your smart and reliable FPGA cloud solution in Switzerland!
The GRAPIN FPGA Cloud is especially convincing with its intuitive and appealing dashboard. The administration of the services running on it is therefore an easy game!
Together with Mirantis, GRAPIN integrates the FPGA cloud platform as a native OpenStack resource using cyborg. In this way, FPGA are deployed in a fast and secure way.
The GRAPIN FPGA cloud not only convinces with its intuitive usability and numerous features, but also with its attractive and modern design!
The Future of High Performance Computing - Today!
The demands on computer performance are constantly increasing and digitization is advancing unstoppably. In order to meet the increasing needs, GRAPIN is developing the next generation of FPGA servers together with IBM.
Field Programmable Gate Arrays (FPGA) are capable of parallel operations and have a higher scalability than Central Processing Units (CPU) or Graphical Process Units (GPU).
The potential field of application for our revolutionary FPGA technology is very broad, with particular application in the areas of Artificial Intelligence, DNA Sequencing, Machine Learning, Internet of Things, Robotics and Big Data.
Heterogeneuous Computing Clusters Stand-alone network-attached FPGA
Management of cloud FPGA at scale
Heterogenes Computing Cluster
The concept of stand-alone network-atteched FPGA is based on two pillars.
First, FPGA are no longer attached to a CPU via PCIe attachment as in the past but via a network attachment.
Second, FPGAs are promoted to the rank of remote peer processors by disaggregating them from the servers and making them available as independent and self-managed resources in the cloud. The network attachment frees the FPGA from the traditional CPU FPGA attachment by connecting the FPGA directly to the data center network. This makes the number of distributed FPGA independent of the number of servers.
The cloud FPGA infrastructure is the most important prerequisite for such a large-scale deployment of FPGA in data centers. It was designed from the ground up to provide the world's densest and most energy-efficient rack unit of FPGAs.
Due to the completely new design, 64 FPGA fit into a 19" x 2 HE chassis, which has direct water cooling. Altogether 16 of such chassis and thus 1024 FPGA fit into a 42 U rack, which thus has 16 TB DRAM. Due to the unique FPGA design, the GRAPIN FPGA Cloud can save up to 80% of physical data center space compared to conventional cloud solutions.
The traditional way to integrate an FPGA into a server is to connect it to the CPU via a point-to-point high speed connection and to treat this FPGA resource as a co-processor under the control of the server CPU.
However, due to this master-slave programming paradigm, such an FPGA is typically integrated into the cloud only as an option of the primary host computing resource to which it belongs. As a result, bus-bound FPGAs are typically made available indirectly through virtual machines (VMs) or containers in the cloud.
In our approach, on the contrary, a stand-alone, network-connected FPGA can be requested independently from a host via an FPGA Cloud Resource Manager. This provides an API for integration into the management stack of the data center (e.g. OpenStack).
The integration of the latest technology into our holistically reworked FPGA design results in the incredible performance of our FPGA solution. In this way, the increasing performance requirements are taken into account.
The increased energy efficiency results from the more compact design, less materials used and the elimination of the need for a cooling machine, which consumes up to four times less energy than conventional FPGA solutions. This saves not only natural resources but also the financial resources of our customers.
Thanks to an innovative network connection, our revolutionary FPGA solution is unlimited scalable. This also saves valuable resources and enables a more efficient working method!