Based on the 32-nanomenter Nehalem microarchitecture, the new Xeon 7500 series processor is 20 times faster than a Potomac series processor that Intel rolled out for four-socket servers in 2005. According to Kirk Skaugen, vice president of the Intel Architecture Group and the general manager of the Data Center group, this means that a single eight-core 7500 series server can effectively replace as many as 20 single processor servers.
As IT organizations have struggled with power consumption and virtualization in the last two years, interest in four-socket and higher servers has begun to increase thanks to the reduced amount of energy they consume and the larger amounts of memory they can make available to virtual machine software.
Xeon 7500 series processors in an eight-socket server can simultaneously process up to 124 threads simultaneous while sharing access to 2 TB of memory. That system is load balanced using the Intel QuickPath Interconnect (QPI) architecture to make sure each application workload has access to enough system resources.
That ability to balance I/0 throughput across a large pool of memory makes the 7500 series “the virtualization platform of choice,” said Skaugen. To that end, Intel added a new Machine Check Architecture Recovery (MCAR) capability that isolates multibit memory errors, which is a capability typically only associated with mainframe systems today, he added.
Over 50 vendors today launched a broad range of servers using the 7500 series processors that range from two to eight-socket server offerings. Many of the two socket servers are based on the Xeon 6500 series, which is designed specifically for smaller servers.
Advanced Micro Devices earlier this week launched a 12-core 6100 series Opteron processor family that is aimed at the same application workloads as the Xeon 7500 series, setting up a showdown in the market yet again between the two largest providers of processors using an industry standard server architecture build around X86 architectures.
Comments
Post new comment