Intel intros Xeon E5-2600v2 CPU based network accelerator -

Intel intros Xeon E5-2600v2 CPU based network accelerator

Intel Corp. has just released its Highland Forest communications platform that combines the Intel Xeon processor E5-2600 v2 with the new Intel communications chipset series 89xx (codename Coleto Creek).

To match the capabilities in the network processor market of such competitors as Cavium and its Nitrox III, Intel hopes that the platform will create efficiencies for telecom customers though workload consolidation, streamlined R&D resources, common software tools and faster system integration.

The accelerators are integrated in a family of pin-compatible 89xx series chipsets, that work in combination with the Intel Atom processor C2758 in entry-level, network equipment. The technology the company has put together delivers scalable crypto performance up to 50Gb/s, PKE performance of 40kops/sec (2 kb keys), and compression up to 24Gb/s per chipset.

With a dual-socket Intel Xeon processor E5-2600 v2 platform with the Intel communications chipset 89xx Series, performance scales up to 255Mpps of L3 forwarding (64 bytes packets), 110Gb/s of IPsec throughput, 200Gb/s of OpenSSL throughput, and 140Gb/s of deep packet inspection (DPI) throughput.

The 89xx gets some of its performance from the use of a 32nm CMOS FinFet process, but some of its significantly higher performance marks is due to the use of accelerators in the chip which are paired with the E5-2600v2, which sports enhanced security and virtualization features.

To achieve the performance necessary to be competitive in this market segment requires the use of two Xeons and four Coleto chips each consuming 17-20 W max to handle such tasks as “applications, control and data plane traffic and to some extent signal processing .

Unlike most other competitive network processors, Intel does not use acceleration logic for regular expressions. Instead it is making use of the x86. In future enhancements to the family, the company will be incorporating support for AVX instruction set and make use of caches and external memory to enhance regular expression performance.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.