LAKE WALES, Fla.—Intel announced its 4th generation high-end Xeon processor, the Xeon E7-8800/4800 version 4 (E7v4) today (June 6). Version 4 is aimed at solving real-time analytics problems for businesses by virtue of its maxed out feature set, fast speed and enlarged memory space compared to all other Intel processors.
Compared with the 22-nanometer E7v3 Haswell microarchitecture, the new 14nm E7v4 Broadwell microarchitecture offers "scalable performance that can answer up to twice as many queries per hour," said Frank Jensen, datacenter performance marketing manager in an interview with EE Times. Using its expanded memory space, some applications can achieve up to 19x queries per hour, said Jensen.
Intel's highest end processor, the E7-8800v4, has 24 cores per socket connected by two rings and a bridge between them.
The highest end E7-8800v4 has 24 cores per socket connected by two rings (with a bridge between them) 48 threads, up to 60-Mbytes of last-level cache, four DDR4 channels, 40 PCI-3.0 channels, speedier cryptographic hardware, supervisor mode access prevention, processor trace and hardware-controlled power management all on-chip plus glue-less scale-up to multi-socket systems with 384 total threads and 24 terabytes total memory capacity.
Intel also claims its E7v4 has surpassed many of the E7v3s former benchmarks by virtue of improving processor utilization of its threading optimization library.
Intel's newest fastest Xeon, the E7--8800/4800 version 4 (E7v4) in its package.
"Our threading building blocks automatically take care of CPU utilization increases by optimizing threads," Jason Herrick, Datacenter Scale Engineering Manager, Software and Services Group told EE Times. "Resulting in an 89 percent speedup by optimizing threads and even more for hyper threads."
Intel is hitting the high-end processor market from every angle with the newest E7v4 fastest for in-memory applications (left) and its E5, 72-core Phi and Altera FPGAs handling the largest data sets.
Intel is aiming its Xeon E7v4 squarely at the highest-end applications in the industry. Today, the E7v4 will be mostly used in business servers that need the minimum real-time latency for mission critical applications, such as real-time analytics on transactional data sets. For the future Intel is aiming for applications in the emerging gigantic data mash-ups from the Internet of Things (IoT), real-time analytics and deep learning algorithms.
"Today, most data is generated by people—from transactions to social media—but data from machines [IoT] will rapidly outpace and those that can take advantage of that data and use it will be the one who succeed in the future," Patrick Buddenbaum, general manager of Intel's enterprise information technology (IT) solutions data center groups told EE Times. "The future of analytics is quickly moving to applying modeling, simulation and machine learning capabilities to real-time and historic data to predict and model future outcomes.”
Buddenbaum defined "real time" as less than a second and gave the examples of responding to outages on the electric grid, detecting teleco intrusions, fraud detection at banks, route optimization for transportation and one exception to the one-second rule, closing the monthly books at businesses in less than a day rather than the two weeks it takes today.
With the PC market waning for its Core processors and the Xeon market for servers rising, Intel is aiming its highest priced Xeon at picking up the slack by gaining entry into new markets—especially those where the need-for-speed trumps price. The Xeon E7v4 is therefore aimed not only at online transaction processing (OLTP), but also at supply chain management (SCM), enterprise resource planning (ERP) and many other mission-critical applications processing data-intensive workloads.
Intel also has a list of "extras" they claim are supported by their premium priced processors like the new E7v4 including hardware-based security, encryption accelerators, malware detectors, supervisor mode access protection against hackers, hardware virtualization support, resource orchestration features, on-chip telemetry, load balancing and workload migration support across multiple applications.
— R. Colin Johnson, Advanced Technology Editor, EE Times