Besides benchmarking the Intel Core i7 3770K under Linux, also being re-tested for processor comparisons in this article include the AMD Fusion A8-3870K "Llano" APU, AMD FX-8150 "Bulldozer", Intel Core i3 2120 "Sandy Bridge", Intel Core i5 2400S "Sandy Bridge", Intel Core i5 2500K "Sandy Bridge", and Intel Core i7 3960X "Sandy Bridge Extreme" hardware. The Intel Core i7 3960X Sandy-E was tested at its stock frequencies (3.3GHz base frequency; 3.9GHz Turbo Boost) and when overclocked to 4.20GHz. The Intel Core i7 3770K was tested at its stock frequencies (3.50GHz; 3.90GHz Turbo Boost) and when it was overclocked to 4.30GHz. A much greater overclock could have been achieved, but this was running with a stock Intel heatsink-fan and overclocking is not our primary focus at Phoronix, but rather the Linux support and performance. The 4.30GHz frequency on the Intel Core i7 3770K with the Intel DZ77GA motherboard was very easy and there was no stability issues.
The same hardware was used each time except when swapping out the motherboard when needed for processor compatibility. The common components included 16GB of DDR3-1600MHz memory (4 x 4GB), 240GB OCZ Vertex 3 Serial ATA 3.0 SSD, and an AMD Radeon HD 7950 3GB graphics card.
Ubuntu 12.04 LTS x86_64 was being used with the Linux 3.2 kernel and the AMD Catalyst 12.3 binary driver. On top of the clean Ubuntu 12.04 LTS x86_64 installation, GCC 4.7.0 was installed (instead of GCC 4.6 as found in Ubuntu 12.04 LTS) since this latest compiler release has various performance improvements / better support for Intel Sandy Bridge, Intel Ivy Bridge, and AMD Bulldozer platforms. All source-based test profiles were built by GCC 4.7 while the CFLAGS/CXXFLAGS were set to "-O3 -march=native" to target compiler optimizations for each of the processors under test, with the tests being re-installed each time between processor tests.
This article is not looking at the Ivy Bridge Intel HD 4000 Linux graphics performance, but those graphics numbers are being delivered in their own set of articles to be published shortly. These Linux benchmarks are just looking at the processor performance. Benchmarking was done in a fully automated and reproducible manner using the Phoronix Test Suite.
There is also thermal and power consumption benchmarks for all of this hardware that was done via the Phoronix Test Suite (later in this article), but first up are the raw performance numbers.