Linux 2.6.38 Kernel Multi-Core Scaling

Written by Michael Larabel in Software on 6 March 2011 at 09:30 AM EST. Page 1 of 2. 26 Comments.

Last month there were benchmarks on Phoronix looking at the multi-core scaling performance of multiple operating systems, including CentOS 5.5, Fedora 14, FreeBSD 8.1, and OpenIndiana b148. CentOS 5.5 uses the long-term Linux 2.6.18 kernel while Fedora 14 has the more recent Linux 2.6.35 kernel by default, but a number of users asked how the Linux 2.6.38 kernel would fair for multi-core scaling with the removal of the Big Kernel Lock and various other low-level improvements in this forthcoming kernel. Here are some benchmarks showing just that.

While the Linux 2.6.38 kernel does have a lot of interesting changes from the low-level improvements to hardware driver enhancements and file-system improvements, at least with the same workload used last month to illustrate the multi-core multi-OS scaling, there really isn't any difference with this newest Linux kernel code that was fetched last week. Fortunately, at least, there are no regressions.

Last month's results showed a few problems with FreeBSD and OpenIndiana in terms of its scaling performance and also some shortcomings of Intel's Hyper Threading Technology, but when strictly comparing the Linux 2.6.32 and Linux 2.6.38 kernels, there really isn't anything different to see.

The tests were carried out on the same system with the ASRock X58 SuperComputer motherboard, Intel Core i7 970 CPU, NVIDIA GeForce 9800GTX graphics card, 3GB of DDR3 system memory, and a 320GB Seagate SATA HDD. The software stack was the 64-bit Ubuntu 10.04.2 LTS with the GNOME 2.30 desktop, X.Org Server 1.7.6, GCC 4.4.3, and an EXT4 file-system.

For what it's worth, here are the results for the users that had requested this testing.


Related Articles