GMPbench 0.1 results

GMPbench 0.1 results. We have run the benchmark on the highest-frequency machine of each type to which we have access. Scaling to lower or higher frequencies should work well, since GMP mainly work off the caches.

The benchmark suite is available for download. To run the benchmarks, you also need to compile gexpr.c and put it somewhere in your path.

GMP is written mainly in C, but most processors have a few critical loops written in assembly. The quality of the assembly code is generally very high, but some processors have gotten more attention than others, resulting in better performance.

GMP exercises a particular set of processor capabilities, widening integer multiplication being the most important one. Processors with poor integer multiply support get worse scores on GMPbench than on other benchmarks.

GMP 4.2.x results

GMP 4.1.x results

• The last column, "Optimal", is an estimate of what could be attained by writing optimized assembly code for this processor.
• There was no assembly loop support for Opteron/Athlon64 in GMP 4.1.4. We therefore include two results above for Opteron, 32-bit results using the Athlon32 assembly loops, and 64-bit results using plain C with inline assembly.
• The performance for the Pentium 4 EM64T processors disappoints. There are performance problems with many instructions GMP depends on: 64-bit multiply, integer right shift, and conditional moves and set-on-condition instructions all need around 10 cycles, and are not fully pipelined. See also this report: http://swox.com/doc/x86-timing.pdf
• The 32-bit 90nm Pentium 4 processors (Prescott) run GMP applications slower than older Pentium 4 processors. The reason is that Prescott has longer latencies for SSE2 instructions and memory loads.
• UltraSPARC 3's disappointing scores are a result of its poor integer multiply support (unsuitable architectural support as well as non-pipelined integer multiply implementation).