Intel Xeon X7560: Nehalem EX Review

Test setup

Due to the wide ranging demands of different software we’ve designed a comprehensive suite of professional application benchmarks, backed by a small number of synthetic tests for this review.

These range from 3D animation and rendering packages to real-world HPC (high-performance computing) simulations that are usually run on mammoth clusters.

Professional applications

Here is a list of the professional applications we used for our testing:

• Cinebench R10
  
• Cinebench R11
  
• Euler 3D
  
• FlamMap
  
• LightWave 9.6
  
• Terragen 2
  
• Protein folding with Folding@home

Synthetic tests

Here is a list of the synthetic test applications we used for our testing:

• WPrime
  
• Prime95
  
• Stream

Because we were unable to carry out the testing in the bit-tech lab we were forced to use whatever hardware Intel had to hand. This comprised a demo system built around the Intel demo motherboard shown earlier with four 2.26GHz Xeon X7560s, the flagship model of the Xeon 7500-series. As each CPU has it's own quad-channel memory controller, so give the optimum performance Xeon 7500-series should have DIMMs installed in multiplies of 16. For this reason Intel had pre-fitted the system with 16 8GB 1,066MHz ECC registered DDR3 Crucial DIMMs.

You might think this would give the Xeon 7500-system a massive advantage over the systems we're comparing it to, but none of the benchmarks we use address more than 8GB of RAM apart from Folding@home. Even then, this test only needs more RAM than this because the Xeon 7500-system has so many execution units (64).

Althought the recently launched Opteron 6000-series does support up to eight CPUs in one system, there aren't any four or eight socket motherboards available right now. As a result, we can only compare the quadruple of Xeon X7560s against a system with a pair of Opteron 6174s. The only exception to this is the Folding@home benchmark, which have we run on a system with four Xeon X7460s in the past. These are the fastest models of the previous generation of the Xeon 7400-series, what the Xeon 7500-series is intended to replace.

We've also included the benchmark results of a variety of other dual-processor systems, just to show what sort of performance scaling a second pair of CPUs may bring to your applications. Unfortunately we didn't have all of these systems to hand anymore, so there are some gaps in a few of the graphs. Still, it's useful to see how much additional performance these new CPUs offer over their predecesors. Even though a single-processor system does not offer the same relability/manageability of a quad-processor workstation/server - at readers' request we've also included some benchmark results from a Core i7-930 and Core-i7 980X Extreme Edition.

The Opteron 6174s were tested in a Supermicro H8DGU-F motherboard while the Xeon X5650s were benchmarked in a Supermicro X8DTU-F motherboard. Both were installed in a Supermicro 1U rack mount chassis equipped with the same 650W PSU and Western Digital hard disk.

The older systems comprised a pair of LGA1366 Xeon W5580 in a Supermicro X8DA3-O motherboard. Also included are some benchmark results from a pair of LGA771 Xeon X5482 CPUs and a pair of LGA771 Core 2 Extreme QX9775s in an Intel D5400XS Skulltrail motherboard. Finally, you'll also see some benchmark results from a pair of Opteron 2435 in a Supermicro H8DAE-2 motherboard and a pair of Opteron 2383s and in a Supermicro H8DA8-2 motherboard. The two Core i7 processors were benchmarked on an Asus P6TD Deluxe.

Test Setup

• Operating System: Microsoft Windows Server 2008 R2
  
• Operating System: Oracle Red Hat Enterprise 5.4
  
• Hard Drive: Western Digital 1TB RE3 WD1002FBYS
  
• Memory: Crucial 1,066MHz registered ECC DDR3