3.0 GHz Intel Xeon Core Duo (Woodcrest), 4MB L2 cache, 64-bit mode. Linux 2.6.17, Intel C/C++ Compiler 9.1.043, Intel Fortran Compiler 9.1.037, Intel Math Kernel Library Version 8.1.1, Intel Integrated Performance Primitives v5.1.1. Has SSE (4-way single precision SIMD), SSE2 (2-way double precision SIMD), and SSE3.
Only the single-CPU performance was benchmarked.
NOTE: Because this benchmark includes very large transforms, we noticed a bug in FFTW 3.1.2 that causes it to pick suboptimal algorithms for transforms that take more than a couple of seconds. We therefore benchmarked a version of FFTW patched to fix this bug; the patch will be included in the next release, but for now you can just edit kernel/timer.c to replace 1.0E10 with 1.0E100. Sorry about that.
Compilers and flags (unless overridden):
icc -no-gcc -O3 -xT
icc -Kc++ -no-gcc -O3 -xT
ifort -O3 -WB -xT
Raw data files: supersgj0-64.tar.gz