Xeon E5-2600v2 (Ivy Bridge-EP) CPUs Now Shipping,
Deliver on Performance and Value

Intel's "Ivy Bridge" 22nm refreshes of the Xeon E5-2600 and E5-1600 series CPUs have launched. We're excited by the increased performance, higher core counts, greater memory bandwidth, and excellent value for the dollar of these new Xeon E5-2600v2 and E5-1600v2 series products.

Higher core counts and increased clock speeds

• Up to 12-cores per CPU
• Increased clock speeds throughout the model line, particularly on equivalent core counts (ex: 8-core to 8-core)
• Many SKUs move up a core count step; ex: mid-level "Standard" SKUs feature 6 cores instead of 4
• Up to a 24% floating point performance increase from top-SKU to top-SKU
• Up to 46% higher LINPACK from top-SKU to top-SKU

Memory bandwidth increases for all SKUs

• New support for DDR3-1866 memory and bandwidth up to 59.7GB/socket
• Every SKU moves up a memory clock step: mid-level "Standard" SKUs now run memory at 1600Mhz (a 20% bandwidth increase!)

Excellent value for your compute dollar

• Xeon E5-2650v2 offers superior performance to the old E5-2670, greater memory bandwidth, and a substantial cost saving
• Xeon E5-2670v2 provides 2 more cores per CPU than the old E5-2670, greater memory bandwidth with no cost increase
• Xeon E5-2630v2 brings a large clock speed uplift (to 2.6Ghz) at a low price delta (vs. E5-2620v2, 2.1Ghz)
• High-clock speed specialty SKUs (Xeon E5-2637v2, E5-2343v2, E5-2667v2) now have more cores—this makes them stronger options for applications requiring a combination of clock speed + core count
• Entry-level Xeon E5-2620v2 has a 20% increase in memory bandwidth for no cost increase

Read our blog post for more information about the Xeon E5-2600v2 CPUs

Xeon E5-2600v2 Platforms Ready for All Needs

The new Xeon E5-2600v2 CPUs support much of the same motherboard and platform infrastructure as their predecessors. The new CPUs are socket compatible with a BIOS upgrade, and many platforms are ready for DDR3-1866 memory speeds as well.

NumberSmasher 2U Twin²

• Support for Xeon E5-2600v2 CPUs and 8 DDR3-1866 DIMMs
• 4x hot-swap compute nodes in 2U with optional onboard InfiniBand (QDR or FDR)

NumberSmasher 1U, 2U Xeon Phi Servers

• Support for Xeon E5-2600v2 CPUs and 8 DDR3-1866 DIMMs
• Multiple Xeon Phi 3120P, 5110P, or 7120P coprocessors

NumberSmasher 1U, 2U, 3U, 4U HPC Servers

• Support for Xeon E5-2600v2 CPUs and 8 DDR3-1866 DIMMs
• Largest number of CPU SKU (up to 130W TDP) and storage options

NumberSmasher 1U, 2U GPU Servers

• Support for Xeon E5-2600v2 CPUs and 8 DDR3-1866 DIMMs
• Multiple Tesla K10, K20, and K20X GPUs

Take a Test Drive of the new CPUs on our remote benchmark cluster today!

Intermediate-level CUDA Programming & Optimization Guide

We have been featuring a GPU Performance Optimization Series on our HPC Tech Tips blog. These posts are intended for non-experts who are new to GPU programming or have a GPU application that doesn't achieve the best performance. Although the discussion is technical at times, the focus remains upon improvements that can be made without significant efforts. The goal is for readers to make the intelligent design choices which result in great performance (if not peak performance).

Parallel Thread Management - Topic #1

Read: CUDA Parallel Thread Management, Divergence and Profiling

Host/Device Transfers and Data Movement - Topic #2

Read: Profile CUDA Host-to-Device Transfers and Data Movement
Read: Optimize CUDA Host-to-Device Transfers

Cache and Shared Memory Optimizations - Topic #3

Read: GPU Memory Types and Memory Performance Comparison
Read: GPU Shared Memory Performance Optimization
Read: Avoiding GPU Memory Performance Bottlenecks

HPC & GPU Supercomputing Meetup Group of Boston

We'll be meeting on Tuesday October 22nd to discuss the SC13 Student Cluster Competition. Presentation topics include:

• LINPACK
• WRF (Weather Research and Forecasting)
• NEMO5 (NanoElectronics Modeling Tools
• GraphLab
• Library support for the Student Cluster Competition (SCC) applications at SC 13
• OS and run-time support for the SCC apps at SC 13

Learn more at http://www.meetup.com/HPC-GPU-Supercomputing-Group-of-Boston/events/141108502/

The Boston meetings are sponsored by Microway and NVIDIA

Tech Tidbits

We've collected a number of the latest HPC technical resources below for your perusal.

Intel Parallel Programming and Xeon Phi Webinar Series

Part 1: Coding The Future: Intel's Vision

Part 2: Vectorization using Intel Cilk Plus Array Notation in C++/C (Ideal for Xeon Phi Programmers)

Part 3: Vectorization with Pragmas in Fortran and C++/C (Ideal for Xeon Phi Programmers)

Part 4: Data alignment for effective vectorization in Fortran and C++/C (Ideal for Xeon Phi Programmers)

Part 5: Faster math performance with Intel Math Kernel Library (Ideal for Xeon Phi Programmers)

Part 6: Automatic offload with Intel Math Kernel Library (Ideal for Xeon Phi Programmers)

Part 7: Threading with OpenMP (OpenMP 4 Ready for Coprocessor Offload)

Part 8: Simplified threading with Intel Cilk Plus

Part 9: Threading with Intel Threading Building Blocks

Part 10: Performance analysis with Intel VTune Amplifier XE

Part 11: Distributed Computing with Intel MPI Library (Intel MPI and Xeon Phi Coprocessors)

Part 12: Analyzing and Balancing MPI Applications (Tuning Your MPI Application Running on Xeon Phi Coprocessors)

NVIDIA GTC Express Webinars

October 9, 2013 (MP4 Replay) OpenACC 2.0 vs OpenMP 4.0 Programming Comparison

October 22, 2013, 12pm ET Introduction to SeqAn, an Open-source C++ Template Library

October 29, 2013, 1pm ET How to Improve Performance using the CUDA Memory Model and Features of the Kepler Architecture

November 6, 2013, 1pm ET Bright Cluster Manager: A CUDA-ready Management Solution for GPU-based HPC

WhisperStation - Maximus Desktop Engineering Recap

Our WhisperStation - Maximus Editor's Pick of the Week was featured in the September Issue of Desktop Engineering

Learn why the Editor-at-Large featured WhisperStation - Maximus

For a full description from Desktop Engineering read the detailed specification article

Learn more about WhisperStation - Maximus at the new Microway.com