The current standard hardware platform of SciEngines is the RIVYERA architecture. As a successor to the legacy COPACOBANA series, the RIVYERA’s different models provide increased logic-cell count, memory, bandwidth and its cutting edge Application Programming Interface. While still providing immense energy- and space-efficiency, most systems can also be clustered to further enhance performance to data-center levels.
COPACOBANA, the Cost-Optimized Parallel Code Breaker, was the first standardized FPGA-based high-performance computer, which was optimized for running cryptanalytical algorithms. With 120 FPGAs of the type Spartan-3 1000, COPACOBANA was originally designed for DES cracking.
COPACOBANA V4-SX35 was an FPGA-based machine optimized for running cryptanalytic algorithms that can take advantage of a high amount of DSPs and the possibility to execute a multiply-add or accumulate function every clock at high frequencies.
The new RIVYERA computing-architecture is the technological successor of the COPACOBANA with significantly increased computation resources. This model provides a backplane with up to 16 cards and a total of 128 high performance FPGAs of the type Spartan-3 5000 within a 3 high unit, 19-inch server chassis.
Using up to four Virtex-7 2000T FPGAs in its standard configuration, the RIVYERA V7-2000T was less cost-performance optimized for parallel computing. Instead it provided very large chips for extraordinary designs that would simply not fit into any other FPGA – e.g. in ASIC design projects.
Using up to 32 Spartan–6 LX150(T) FPGAs in its standard configuration, the formica provides high-density supercomputing power to a wide audience. With its miniaturized design, simply attach your new FPGA accelerator to your regular workstation or laptop to have a supercomputer directly in your office.