LDA Technologies (LDA), the innovation leader in advanced ultra-low latency solutions for capital markets, today announces the launch of two highly configurable, ultra-low latency, high-density hardware platforms designed for clients to build the devices their workload requires. Torus is an unprecedented high-density, multi-board platform engineered for compute heavy workloads with compact deployment for trading, HPC, and AI. OmniX brings the lowest latency, FPGA-powered networking platform available with high-bandwidth capture and swappable drives, allowing users to tap, timestamp and record in one box.
OmniX consolidates flexible connectivity, FPGAs and storage within 1RU. The configuration can include LDA’s Maze, its newest low latency layer 1 fabric, to easily scale connectivity from 10G to 25G, or the Neo fabric for the lowest latency and a direct Neo X upgrade. Torus supports eight board configurations across an expandable 1RU to 2RU setup, enabling firms to create custom high-performance compute platforms while optimizing rack space and connectivity. One or two LDA Maze devices can be installed on either floor, turning the stack into a single low latency, configurable Layer 1 mesh. For both platforms, LDA Maze connects the components within the chassis at a latency of 3.6 to 4.2ns: front-panel ports, FPGA boards, NICs and other equipment alike.
With these two platforms, users benefit from future-proof infrastructure without continually adding separate appliances, external switches and cabling, which helps reduce the space their tech stack occupies and overall costs. Also, when time comes to scale, these solutions do it seamlessly rather than requiring firms to overhaul and redesign their infrastructure from scratch.
Vahan Sardaryan, CEO and Co-founder of LDA, said: “OmniX and Torus address different points of the connectivity and performance spectrum that firms need to function in today’s markets. Users can select and scale the platform that matches their infrastructure requirement, while reducing the physical footprint needed to increase connectivity and compute capacity. Both platforms are built around the same idea that infrastructure should be customized around the workload, whether that is for trading, HPC or AI workflows; LDA can provide the right solutions.”
OmniX is purpose built per deployment with interchangeable building blocks: FPGAs, Layer 1 fabric, hot-pluggable storage, front panel flexibility. When used with FPGAs, it allows capital markets firms to run more strategies on less hardware. Torus is a powerful compute platform, and its Layer 1 interconnect, board synchronization and health monitoring capabilities allow clients to run advanced compute and implement failover scenarios seamlessly within the single chassis.
