Skip to content
Broker news. Connected. RSS feed
FXBROKERS

A clearer view of the broker industry.

HPR benchmark measures median order roundtrip latency near five microseconds

Exactpro timed requests and the first response at a cross-connected test server, giving exchange operators a defined measurement boundary for evaluating the hardware platform.

Exactpro has completed independent latency testing of HPR’s HYPRCUBE matching engine, reporting median order roundtrip latency close to five microseconds across varied order types, execution conditions and order-management operations. The companies published the results on 14 October.[1]

The test used a production-like mixture of scenarios and repeated measurements across configurations. A test server was cross-connected to HYPRCUBE, with timing taken from network captures at that server’s network interface. The measured interval ran from sending a request to receiving its first corresponding response or update.[1]

That boundary gives the figure a specific meaning: it measures the roundtrip at the external test server, without representing a remote broker’s entire trading path. A median describes the middle observation; it says nothing by itself about the slowest responses.[1]

Throughput is a separate specification

HPR’s product materials advertise capacity of up to 25 million messages per second. That is a vendor throughput specification, distinct from Exactpro’s reported median latency result. Throughput measures how much traffic a system can process; latency measures the delay for a particular interaction. Procurement comparisons need both measures to refer to the same workload and configuration before they can be combined into a performance claim.[2]

The product’s hardware is configurable, according to HPR, allowing changes to order types and protocols. Gateways and order books can be scaled to suit demand, and the platform includes network-security controls and managed support. These features address the operating flexibility of an exchange platform alongside its processing speed.[2]

The commercial question is how that design fits a venue’s own rules and workload. A fast matching path is useful evidence, while capacity, configuration changes and operational support remain separate parts of the implementation decision.[2]

Follow the evidence

Sources & further reading

Report an error or submit an update