Re: [viff-devel] benchmark
Janus Dam Nielsen writes: >> Can you change it so that it will only run the benchmark once if it >> sees that no preprocessing is needed? > > That should be simple, just put a test to see if the list of needed > program counters is zero. Right, I had a look at it, but I got confused by the code. > I won't have to time to implement it any time soon. Would you like to > do it, or should I put it in my backlog? It's not overly important -- I'll do it when I get sufficiently annoyed by it :-) -- Martin Geisler VIFF (Virtual Ideal Functionality Framework) brings easy and efficient SMPC (Secure Multiparty Computation) to Python. See: http://viff.dk/. pgpfkWbp4EJFR.pgp Description: PGP signature ___ viff-devel mailing list (http://viff.dk/) viff-devel@viff.dk http://lists.viff.dk/listinfo.cgi/viff-devel-viff.dk
[viff-devel] benchmark
Hi Janus, I like it very much that the benchmark can figure out by itself which program counters are needed for preprocessing. Can you change it so that it will only run the benchmark once if it sees that no preprocessing is needed? -- Martin Geisler VIFF (Virtual Ideal Functionality Framework) brings easy and efficient SMPC (Secure Multiparty Computation) to Python. See: http://viff.dk/. ___ viff-devel mailing list (http://viff.dk/) viff-devel@viff.dk http://lists.viff.dk/listinfo.cgi/viff-devel-viff.dk
Re: [viff-devel] Benchmark results, with graphs!
Martin Geisler <[EMAIL PROTECTED]> writes: Hi again > The results are from test runs between the machines thyra (Aarhus, > Denmark), bazooka (LA, USA), and serengeti (Trondheim, Norway). I have > used benchmark.py, some shell scripting and lots of patience :-) > > The first graph shows parallel multiplications in the standard 65 bit > field. The time is the total time used on the three machines. One can > see that bazooka is a bit slower than the two others, but overall they > follow each other nicely. This is a plot that shows how the data is spread: <> I have sorted the timings for each run and plotted the results so that the 1st quartile is a 0.25, the median value at 0.5, and the 3rd quartile at 0.75. I guess the graph shows that the data is alright: most data values (the lowest 75% or so) are close together around the median. About 10% of the data is much larger than the rest, but those spikes are probably the result of network congestion between the three machines. The bazooka machine is a shared webserver and it normally has a load of 5-10, but I've also seen it go to 20+ at times. -- Martin Geisler ___ viff-devel mailing list (http://viff.dk/) viff-devel@viff.dk http://lists.viff.dk/listinfo.cgi/viff-devel-viff.dk
[viff-devel] Benchmark results, with graphs!
Hello, Happy New Year! I have some fresh benchmark results for you all. The results are from test runs between the machines thyra (Aarhus, Denmark), bazooka (LA, USA), and serengeti (Trondheim, Norway). I have used benchmark.py, some shell scripting and lots of patience :-) <> The first graph shows parallel multiplications in the standard 65 bit field. The time is the total time used on the three machines. One can see that bazooka is a bit slower than the two others, but overall they follow each other nicely. <> The second graph gives the normalized results from the first graph, that is, the time per multiplication. The time stabilized around 1.5 ms per multiplication when doing more than 2000 multiplications. <> If we run things sequential instead, we get the results in graph 3. Here I have only tested up to 1000 multiplications with 10 repetitions since that took 30 minutes. <> Again, we can look at the time per multiplication. I have not good intuition as to why the two fast machines (thyra and serengeti) are able to do the first 10 multiplications much faster than the later ones. The bazooka machine is more stable. I believe these are the most comprehensive results I have made with VIFF so far. For the parallel multiplications I repeated the test at least 100 times and used the median value for the graphs. I took the median value instead of the average value since the median should be more robust when the data has outliers. But since I have no clue about statistics, I would love to hear other suggestions. -- Martin Geisler ___ viff-devel mailing list (http://viff.dk/) viff-devel@viff.dk http://lists.viff.dk/listinfo.cgi/viff-devel-viff.dk