The long draught at the high-end
It’s been a long wait, about four years if memory serves me well, since Intel introduced the Xeon E7, a high-end server CPU targeted at the highest performance per-socket x86, from high-end two socket servers to 8-socket servers with tons of memory and lots of I/O. In the ensuing four years (an eternity in a world where annual product cycles are considered the norm), subsequent generations of lesser Xeons, most recently culminating in the latest generation 22 nm Xeon E5 V2 Ivy Bridge server CPUs, have somewhat diluted the value proposition of the original E7.
So what is the poor high-end server user with really demanding single-image workloads to do? The answer was to wait for the Xeon E7 V2, and at first glance, it appears that the wait was worth it. High-end CPUs take longer to develop than lower-end products, and in my opinion Intel made the right decision to skip the previous generation 22nm Sandy Bridge architecture and go to Ivy Bridge, it’s architectural successor in the Intel “Tick-Tock” cycle of new process, then new architecture.
What was announced?
The announcement was the formal unveiling of the Xeon E7 V2 CPU, available in multiple performance bins with anywhere from 8 to 15 cores per socket. Critical specifications include:
- Up to 15 cores per socket
- 24 DIMM slots, allowing up to 1.5 TB of memory with 64 GB DIMMs
- Approximately 4X I/O bandwidth improvement
- New RAS features, including low-level memory controller modes optimized for either high-availability or performance mode (BIOS option), enhanced error recovery and soft-error reporting
While there will no doubt be niche applications for the lower core count versions where performance per thread and increased memory compared to the E5 is critical, the real excitement is around performance boost that it will lend to four-socket and above servers. Multiple vendors, including HP, Dell and Fujitsu were on-stage with Intel to extol the virtues of the E7 and to tout their wares, and the message is clear – the E7 V2 is scalable and it is fast. While comprehensive performance results are not yet available, all of the major server vendors appear to have posted SAP SD 2-Tier benchmark results, which I have long maintained are a pretty representative of large enterprise transactional applications, and the numbers are impressive. It looks like the new 4-way systems perform almost exactly like the previous generation 8-way servers, and the Fujitsu system, the only 8-socket result reported, appears to deliver 1.9x the performance of the average 4-socket system, which is impressive near-linear scalability. Interestingly, these results still leave the absolute top-end x86 system in the historic range of 1/3 – 1/5x the absolute high-end RISC system result (absolute performance, not price-performance).
What does it mean?
Simple message, big impact – high-end x86 servers for mission-critical workloads just got a big boost in performance again, eliminating any worries about scalability of key monolithic workloads for a few more years. Figure approximately double the performance per socket on transactional workloads, and even more on workloads like in-memory databases and advanced analytic that can take advantage of the doubling of available memory, and the permutations of DIMM and memory-channel flash memory in the future are almost endless.
Expect to hear a lot about these new systems from your server vendor. Vendors who have been quiet about their high-end wares as their two-socket products began to impinge on the performance envelopes of the larger systems will jump on this opportunity with both feet now that the E7 V2 is available because the margins on these systems are as attractive to the vendors as the benefits are for their customers, a winning situation all around.