By: Jenifer Hopper
This article discusses some basic XML tuning tips for PowerKVM guests. It helps new users get started with editing guest XML definitions, and walks through some simple tuning examples.
The article covers various options to tune the guest disk, network, cpu, and memory. It also includes some example guest resource pinning configurations for different scenarios. Applying these tips may help improve application performance by ensuring your guest is configured properly and optimized for... [More]
By: Jenifer Hopper
This article explains an example method to tune a full-system PowerKVM guest to achieve CPU and memory performance that is very close to non-virtualized speeds, demonstrating very low KVM overhead on a POWER8 system. It also provides some common tuning tips for running SPECjbb2005 and the STREAM memory bandwidth workload on POWER8 systems.
The example tutorial starts by measuring non-virtualized performance to provide a system baseline (non-virtualized mode is currently a technical... [More]
By: Jenifer Hopper
Are you interested in better understanding memory access performance? A new article, Untangling memory access measurements - memory latency explains how to use an example benchmark called lat_mem_rd to measure memory read latency on IBM® Power Systems™ servers running Linux.
This article takes a practical approach to measuring memory latency performance, from understanding initial results to moving on to examining the effects of more... [More]
By: Steve Dobbelstein.
On February 22, 2013, IBM® published a new SPECjEnterprise®2010 result on a two socket system. The result was achieved on the new IBM PowerLinux™ 7R2 system using IBM WebSphere® Application Server v8.5 and IBM DB2® 10.1 running on Red Hat® Enterprise Linux® release 6.4. The Linux Technology Center (LTC) team, the WebSphere team, the Java team, and the DB2 team all worked together to analyze and tune the performance of the... [More]
By: Maynard Johnson and Beth Taylor. Finding performance bottlenecks in applications that you develop can be a daunting task. But with the right tools and a little guidance, it's easier than you might think. OProfile is a performance analysis tool set for Linux systems. A new collection of topics has been published in the Linux for IBM Systems Information Center to help application developers get started with OProfile on Power Systems™ servers running Linux. Getting started with OProfile on PowerLinux introduces the new operf... [More]
By: Anirban Chatterjee.
month, the PowerLinux team is announcing the biggest technology change in PowerLinux servers
since we launched, with the availability of our POWER7+ chips on the platform.
POWER7+ is more than just a speed bump on our POWER7
processors. Our hardware teams have
worked hard to increase the flexibility of the platform, bringing
balanced performance increases while keeping other factors like energy
consumption at bay. Some examples:
doubled the memory capacity in... [More]
zswap" is discussed, with some initial performance data provided to demonstrate the potential benefits for a system (partition or guest) which has constrained memory and is beginning to swap memory pages to disk. The technique improves the throughput of a system, while significantly reducing the disk I/O activity normally associated with page swapping. We also explore how zswap works in conjunction with the new compression accelerator feature of the POWER7+ processor to potentially improve the system throughput even more... [More]
By: Beth Taylor and Walt Madden. Several previous blog posts have told of the
advantages of IBM ®
Advance Toolchain for PowerLinux for application performance analysis on IBM
Power Systems ™
servers. Likewise, you've heard about
IBM Software Development Kit for PowerLinux ™ , which combines C/C++ source
development with the Advance Toolchain and classic Linux debugging and
performance analysis tools like OProfile. Recently, the IBM InfoSphere Streams development
team has had the positive experience to give credit to these claims... [More]
By: Jenifer Hopper. More details on CPU utilization issues
I recently posted about a RHEL6.3 kernel update that resolves a CPU utilization bug in this blog . Since then I have received a few questions asking for more details, so I want to share those here. Q: The problem was in /proc/stat? What was the issue? A: Yes, there was a bug that caused incorrect idle and iowait values to be displayed in /proc/stat. The main issue was a change that caused an incorrect divisor to be used (msec vs. usec). Q: Do I need to update my profiling... [More]
By: Jenifer Hopper.
Fixing CPU utilization metrics on RHEL 6.3
As reported earlier in a blog post , the standard RHEL 6.3 kernel has issues with correctly calculating CPU utilization values .
(For more insights into the fix below, see a different blog post )
** Note: The specific zstream kernel version mentioned in this article was the most recent at the time of publication. Newer RHEL 6.3 zstream kernel releases should contain this... [More]
By: Carlos Seo. The IBM Advance Toolchain for PowerLinux is a set of open source development tools and runtime libraries which allows users to take leading edge advantage of IBM's latest POWER hardware features on Linux. A new update release is now available, and it features: Bug fixes for DFP in GCC. Fix to address over counting user context lost samples in Oprofile. GDB now displays correctly the vperm instruction. Time zone for the Advance Toolchain is now automatically set to the system's time zone. New advance-toolchain-at5.0-selinux rpm,... [More]
By: Bill Buros. If you've updated to the most recent RHEL 6.3, you might be seeing issues with CPU utilization being reported to the end users.
# uname -r 2.6.32-279.el6.ppc64 With the kernel here, if you run "top" on an idle system, it may report clearly inaccurate CPU utilization metrics, as in the idle system below. 66% kernel time on an idle system is not accurate.
# top top - 16:05:04 up 1 day, 39 min, 2 users, load average: 0.00, 0.00, 0.00 Tasks: 1167 total, 1 running, 1166 sleeping, 0 stopped, 0... [More]
By: Carlos Seo. The IBM Advance Toolchain for PowerLinux is a set of open source development tools and runtime libraries which allows users to take leading edge advantage of IBM's latest POWER hardware features on Linux. A new major release is now available, and it features: New base toolchain, based on gcc-4.7.2 and glibc-2.16 Supported on SLES 11 and RHEL 6 only (and their respective Service Packs and Updates) Optimized libraries for POWER7 only gcc now defaults to -mcpu=power6 -mtune=power7 Oprofile now supports non-root single... [More]
By: Steve Champagne. For information about performance tuning on your PowerLinux system, be sure to see the A Tuning Guide for PowerLinux section of Performance Rocks - Best Practices . I added three new sections there today:
Tuning Guide: Compilers and Optimization Tools for C/C++/Fortran
Tuning Guide: Empirical Performance Analysis using the IBM SDK for PowerLinux
Tuning Guide: Deeper Empirical Analysis for improving performance
By: Carlos Seo. The IBM Advance Toolchain for PowerLinux is a set of open source development tools and runtime libraries which allows users to take leading edge advantage of IBM's latest POWER hardware features on Linux. A new release is now available, and it features: Performance improvements for tcmalloc Bug fixes for DFP support in Valgrind For download links, more information and documentation, please refer to our official documentation page . Please let us know if you have any questions about this release.
By: Carlos Seo. The IBM Advance Toolchain for PowerLinux is a set of open source development tools and runtime libraries which allows users to take leading edge advantage of IBM's latest POWER hardware features on Linux. A new release is now available, and it features: Important bug fixes in libstdc++ Added support for DFP (decimal floating point) for Power in valgrind Updated OpenSSL to 1.0.1c For download links, more information and documentation, please refer to our official documentation page . Please let us know if you have any... [More]
New IBM XL compiler versions now available!
By: Bill Buros.
IBM has released the IBM XL C/C++ Version 12.1 and IBM XL Fortran V14.1 products for use on PowerLinux systems.
IBM XL C/C++: http://www-01.ibm.com/software/awdtools/xlcpp/linux/
IBM XL Fortran: http://www-01.ibm.com/software/awdtools/fortran/xlfortran/linux/
Over on developerWorks, there are two good articles on "what's new" with these new versions.
What's new with IBM XL C/C++ for AIX and Linux,... [More]
Performance Projections vs "Hope for the Best" By: Bill Buros. Here's a quick introduction to the notion of understanding what you're trying to do when you are looking for performance improvements. Over the last month or two, we have been focused on several engagements where we are helping to improve an application's performance on Linux. Being in the Linux Technology Center, we focus on Linux itself, the tools, commands, products, and post-processing analysis of the results. So while we often deal first with Power systems, many of... [More]
Need Java 7? We got Java 7. By: Bill Buros. Last year we posted that the IBM Java 7 was in open beta mode, but along the way we neglected to post the news that the IBM Java 7 kits are now fully available across a number of platforms and operating systems. By now, this is relatively older news, IBM Java 7 was announced back in September 2011, but it's always good to remind people and teams that it's available. This latest IBM Java is available on Power systems and across the RHEL and SLES distros available for Power. We use the latest... [More]
By: Bill Buros. Continuing our journey into benchmarks, we recently were working on the SPECcpu2006 set of benchmarks again, and returned to look at some of the dependent products which we build and run with. One of the products which we link with on Power Linux is MicroQuil's Smartheap library. This is a run-time library which can be purchased from MicroQuil. One of the benchmark components of the SPECcpu2006 suite - xalancbmk - is a classic example of a questionable choice as a system and compiler measurement benchmark. The... [More]