GMPbench results

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

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

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.

GMPbench 0.2 isn't a perfect benchmarks suite for GMP, but it is much better than GMPbench 0.1.

GMPbench 0.1 measures multiplication with same-size operands, and division, and then RSA encryption. That's it. GMP typically performs a whole lot more operations which are not measured at all. Furthermore, the RSA figures weigh too much in the score computation.

GMPbench 0.2 adds squaring, multiplication of different-size operands, more variants of division, gcd and gcdext, all in the "base" hierarchy, and pi computation in the "app" hierarchy.

The added benchmarks of GMPbench 0.2 aren't totally fair to older GMP versions; we added measurements of things we thought were important, and then used these to drive generic improvements of GMP 4.3. We sort of tuned-for-the-benchmark, but since we chose the benchmarks to make sense, it isn't as silly as such an exercise would usually be.

GMP 4.3.x GMPbench 0.2 results

GMP 4.3.x GMPbench 0.1 results

1. These results are preliminary and based on a snapshot of what will become GMP 4.3. Final results should be somewhat better for certain processors.
2. The clock frequencies for the above measures are not the same as for GMP 4.2, since we didn't have access to the same hardware. However, we have remeasured some of the 4.2 numbers and updated the table below.
3. The last column, "Optimal", is an estimate of what could be attained by writing optimized assembly code for this processor.

GMP 4.2.x GMPbench 0.1 results

GMP 4.1.x results

• 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 poor scores are a result of its inadequate integer multiply support (shortcomings in both ISA and implementation).