Benchmarking Debian GNU/Hurd 2015 vs. GNU/Linux

Since the release of Debian GNU/Hurd 2015 and GNU Hurd 0.6 this year, I've been meaning to run some new performance benchmarks considering my previous Hurd benchmarks were last done in 2011.

After a reminder this week from a Phoronix Premium member about some fresh Debian GNU/Hurd benchmarks, here are some fresh results comparing this version of Debian with the Hurd kernel instead of the Linux kernel. The results for this comparison are done against Debian GNU/Linux Wheezy. For those unfamiliar, Debian GNU/Hurd is the GNU user-land but running atop Hurd -- similar to Debian GNU/kFreeBSD being powered by the FreeBSD kernel. On the topic of Debian GNU/kFreeBSD, I'll probably run a fresh kFreeBSD vs. Linux comparison again soon.

Given the still poor state of Hurd support for modern hardware and being 32-bit, I tested Debian GNU/Hurd vs. Debian GNU/Linux from a KVM virtual machine. The Intel Xeon host powered by Ubuntu 15.04 was running QEMU/KVM with virt-manager and the VM was allotted one core with 2GB of memory and 6GB of virtual storage. The same VM settings were used for both the GNU/Hurd and GNU/Linux testing with the 32-bit Debian 8.1 GNU/Linux release being used for the reference results atop the Linux 3.16 kernel. The testing was done sequentially and no other processes were active on the host system during the benchmarking to try to maintain a stable and fair comparison. The same GCC 4.9.2 compiler is found in user-space for both of these versions of Debian.

I did a basic port of the Phoronix Test Suite to Hurd back in 2011 and it's still running fine with the latest upstream Phoronix-Test-Suite on GitHub code or any released version in the past four years. Many of the source-based, CPU-focused benchmarks found via the Phoronix Test Suite / OpenBenchmarking.org are compatible with testing from GNU Hurd, which is what was run for this first Hurd benchmark run in four years on Phoronix.