1. Computers
  2. Display Drivers
  3. Graphics Cards
  4. Memory
  5. Motherboards
  6. Processors
  7. Software
  8. Storage
  9. Operating Systems


Facebook RSS Twitter Twitter Google Plus


Phoronix Test Suite

OpenBenchmarking.org

Linux 3.6 Kernel Regression Causes ~20% Drop

Linux Kernel

Published on 15 September 2012 04:35 AM EDT
Written by Michael Larabel in Linux Kernel
3 Comments

A commit within the in-development Linux 3.6 kernel has caused the PostgreSQL database server workload to regress by 15~20%. Fortunately, the commit has been spotted.

For the past day there's been a Linux kernel mailing list thread about a "20% performance drop on PostgreSQL" when moving from the Linux 3.5 to Linux 3.6 RC kernels. This regression in the popular database server was quickly confirmed by Borislav Petkov of AMD who then isolated it down to this commit: sched: Improve scalability via 'CPU buddies', which withstand random perturbations. While trying to improve the scalability of the Linux scheduler, it's not without problems.
Traversing an entire package is not only expensive, it also leads to tasks bouncing all over a partially idle and possible quite large package. Fix that up by assigning a 'buddy' CPU to try to motivate. Each buddy may try to motivate that one other CPU, if it's busy, tough, it may then try its SMT sibling, but that's all this optimization is allowed to cost.

Sibling cache buddies are cross-wired to prevent bouncing.

4 socket 40 core + SMT Westmere box, single 30 sec tbench runs, higher is better:

clients 1 2 4 8 16 32 64 128
..........................................................................
pre 30 41 118 645 3769 6214 12233 14312
post 299 603 1211 2418 4697 6847 11606 14557

A nice increase in performance.
It caused a nice increase for one workload while dramatically affecting another real-world and common Linux workload.

Linus Torvalds was quick to say the commit in question is confusing, although he committed it to the kernel tree weeks ago prior to the Linux 3.6-rc1 kernel.

Linus ultimately said, "I vote we just revert it as 'insane'. The code really doesn't seem to make any sense." Again, something living in mainline for weeks.

The discussion has yet to be settled, but at this point it looks like the problematic commit will simply be reverted prior to the final release of the Linux 3.6 kernel.

For those wondering about the state of the Phoromatic Linux Kernel Tracker, the overhauled version is still a work-in-progress and to be published soon. This daily and per-commit tracker would have likely caught this regression seeing as there's a PostgreSQL pgbench test profile, but the interface for this kernel tracker is currently inaccessible to the public. The new version of Phoromatic is powered directly atop OpenBenchmarking.org and its new architecture that is coming as part of the Randaberg / UnterschleiƟheim improvements. If your company is interested by the Linux enterprise software testing capabilities as part of the Phoronix Test Suite or to simply support these mainline efforts, please contact us.

Latest Linux Hardware Reviews
  1. 13-Way Low-End GPU Comparison With AMD's AM1 Athlon
  2. ASUS AM1I-A: A Mini-ITX Board For Socketed Kabini APUs
  3. Mini-Box M350: A Simple, Affordable Mini-ITX Case
  4. Overclocking The AMD AM1 Athlon & Sempron APUs
Latest Linux Articles
  1. Ubuntu 12.04.4 vs. 13.10 vs. 14.04 LTS Desktop Benchmarks
  2. AMD OpenCL Performance With AM1 Kabini APUs
  3. A Quick Look At GCC 4.9 vs. LLVM Clang 3.5
  4. Are AMD Athlon/Sempron APUs Fast Enough For Steam On Linux?
Latest Linux News
  1. Git 2.0 Test Releases Begin With Many Changes
  2. Wine 1.7.17 Works On Its Task Scheduler, C Run-Time
  3. The Improv ARM Board Still Isn't Shipping; Riding A Dead Horse?
  4. Debian To Maintain 6.0 Squeeze As An LTS Release
  5. Wasteland 2 Is Finally Released For Linux Gamers
  6. FreeBSD Advances For ARM, Bhyve, Clang
  7. Ubuntu 14.04 LTS "Trusty Tahr" Officially Released
  8. Ubuntu 12.04 LTS vs. 14.04 LTS Server Benchmarks
  9. QEMU 2.0 Released With ARM, x86 Enhancements
  10. Running The Unity 8 Preview Session On Ubuntu 14.04 LTS
  11. R600 Gallium3D Disables LLVM Back-End By Default
  12. Fedora 21 Gets GNOME 3.12, PHP 5.6, Mono 3.4
Latest Forum Discussions
  1. Suggestions about how to make a Radeon HD 7790 work decently?
  2. The GNOME Foundation Is Running Short On Money
  3. Updated and Optimized Ubuntu Free Graphics Drivers
  4. Radeon 8000M problematic on Linux?
  5. Linux Kernel Developers Fed Up With Ridiculous Bugs In Systemd
  6. After Jack Keane, RuseSoft will briing Ankh 3 to Linux through Desura
  7. Suspected PHP Proxy Issue
  8. Change installation destination from home directory