Kx Systems, the high-performance database and timeseries analysis, has worked closely with Intel to run a series of performance tests on Intel’s new Xeon 5500 (“Nehalem”) chipset and has confirmed major improvements in speed.
Kx and Intel have a long-standing relationship, with each firm benefiting from the other’s expertise in multi-core processing. The Xeon 5500 processor is an illustration of Intel’s multi-core technology enabling Kx’s true multi-core kdb+ database to run faster. Kx’s in-memory and on-disk database has been tested to run more than twice as fast on the Xeon 5500 in specific tests compared to the previous generation of processors. The benchmarking used a comprehensive suite of market models on databases containing hundreds of millions of records and included floating point and integer calculations.
Kx is the vendor of choice for many financial institutions. From its inception the company has developed software designed to make the best use of next generation technology. The release of the Xeon 5500 is a major step forward for Kx, its clients and the real-world speed of their applications: it’s not just a little faster, but now multiple times faster than the previous generation. Bringing more cores to commodity hardware allows Kx to deliver considerable improvements in processing speeds.
“Intel sees the value in the expertise of specialist software houses,” says Garry Thall, director financial services, Intel Americas. “These companies provide the relevance and practical application for our technology. The results of Kx’s technology strategy and Intel’s multi-core platforms is an excellent example of the performance benefits we can deliver our customers.”
With ultra-high speed always in mind Kx’s original product was written for distributed processing on multiple machines even as far back as the early 90s. Now with multi-core 64-bit computers Kx’s design matches the requirement for exceptionally efficient processing. Kx expects that financial institutions will continue to look at multiple processors on multiple machines in response to both the inexorable growth in data and ever increasing demands for processing speed.
D.C.