Welcome to our website
To take full advantage of all features you need to login or register. Registration is completely free and takes only a few seconds.
8 pages 1 2 3 4 > »
Intel E5-2600 Series "Sandy Bridge-EP" Xeons
Posted by: Hooz on: 03/06/2012 05:41 PM [ Print | 0 comment(s) ]

Xeon E5-2600 Series "Sandy Bridge-EP" Processor
Today is the day that Intel is announcing/releasing their latest addition to the Xeon processor family, the Sandy Bridge-EP based, Xeon E5-2600 series. As has been the standard for the last few releases, this new Xeon adds more cores, more cache and other assorted features, and it does it all while using less power than its predecessor.
Since there are a ton of different models of E5-2600 series Xeons offering different amounts of cache, cores and TDP, let's start by taking a look at what is going to be available.

As you can see, there is an overwhelming array of options for the potential E5-2600 Xeon user. With choices ranging from 4 cores/4 threads to 8 cores/16 threads, TDPs ranging from a meager 60W to a hefty 150W, and DDR3 support from 800mhz to 1600mhz, there is sure to be something there for just about everyone.
Typical of new product releases, Intel has also included some other new improvments to help entice people into upgrading (if the number of cores wasn't already enough to sway you). In addition to having more cores, the E5-2600 series Xeons add a fourth memory channel that supports up to 1600mhz DDR3 (and DDR3L), a second QPI link, additional (up to 40 per socket) PCIe 3.0 lanes and the new Advanced Vector Extensions (AVX).

Perhaps one of the most interesting new features of the E5-2600 series Xeons is Intel's new Direct Data I/O (DDIO) technology.

A more technical writer could probably do an entire article just covering what DDIO actually is and does, but I'll leave the nuts-and-bolts of it to someone else and just say this... DDIO allows the processor cache to talk directly with IO devices instead of having to run everything through the main system memory. That means that IO speed and capacity scales up with additional processors, memory access is minimized (saving latency and power) and additional IOH controllers are eliminated. With everyone moving to virtualization of one sort or another, the speed boost that DDIO will afford things like VTd enabled devices and QoS is a pretty big deal.
Test Systems
These days a processor change usually also means a socket change, and a new socket means a whole new platform. This time around is not any different, as the E5-2600 Xeons up the pin count (from 1366 to 2011) and the new processor features require a new platform to leverage those those features. Enter the Romley-EP server platform.

As always, I tried to keep the tests systems as consistent as possible, but some differences could not be avoided due to the differences in the platforms. Both machines are in a workstation configuration so I opted to use Windows 7 Ultimate x64 for the operating system. All updates and Service Packs were installed along with the latest drivers available on the respective vendors' websites. Both chassis contain similar amounts and sizes of fans and both power supplies are similarly rated for power and efficiency.
The Westmere-EP (X5690) system consists of a Supermicro X8DAI/3 motherboard, Nvidia Quadro FX1400 video card, Samsung 160gb, 7200rpm hard drive and 24gb (6x 4gb DIMMs) of memory.
The Sandy Bridge-EP (E5-2687W) system contains an Asus Z9PE-D8-WS motherboard, AMD FirePro X3400 video card, Intel SSD and 64gb (8x 8gb DIMMs) of memory.
Advanced power management settings were enabled on both systems and Windows 7 power settings were set to "balanced". Both sets of processors were tested with SMT and Turbo set to ON for all tests. For the SPECjbb tests, "hardware prefetch" and "adjacent cache line prefetch" options were disabled in the BIOS.
All power measurements throughout this review were taken with an Extech 380803 power meter. Readings were for the computer itself, and all other devices (monitor, speakers, etc) were plugged-in separately and not included in the measurements.
Workloads
I used the following versions of our test applications:
| - SiSoftware SANDRA 2012.SP2 Lite - Black & Scholes Kernel - SunGard Adaptiv Credit Risk Analysis v4.0 - CINEBENCH R11.5 x64 - POV-Ray v3.7.0.RC4 x64 - SPECjbb2005 v1.07 - Euler3d CFD Benchmark v2.2 - Matrix Multiply v3.5 - FlamMap (FSPRO) - MyriMatch - x264 Benchmark HD v4.0 |
Still noticably absent from this review is an old-time favorite, 3ds Max. I attempted to run 3ds Max 2012 on the test systems, but the scanline renderer is still only capable of addressing 16 threads. Once there is an update that addresses this issue, I will happily add 3ds Max back into the benchmarking mix.
All benchmark numbers presented were an average of three runs. Desktop resolution was set at 1920x1080, 32bit color with a 60Hz refresh rate. "Optimized Defaults" were used in the BIOS on all machines.
8 pages 1 2 3 4 > »
































