I think it's important to distinguish between floating point and integer performance. X64 FPU is very powerful, which might make a big difference to the pi calculation. Jeffrey Bastian <jbastian@xxxxxxxxxx> wrote: >On Fri, May 30, 2014 at 03:57:52PM -0400, Robert Moskowitz wrote: >> So now I have some approximation of F20 running on my cubieboard2. >> So I want to do a performance comparison, and simple BOGOmips from >> /proc/cpuinfo probably does not tell the whole story of RISC vs >> CISC. >> >> So is there a tool available for F20 on both that I can use? > > >A quick-n-dirty CPU benchmark is to just run > openssl speed [algorithm] >and compare the results of the various hashing and encryption >algorithms. For example: > >~]$ openssl speed md5 >Doing md5 for 3s on 16 size blocks: 11289961 md5's in 3.00s >Doing md5 for 3s on 64 size blocks: 8221348 md5's in 2.99s >Doing md5 for 3s on 256 size blocks: 4560242 md5's in 3.00s >... >... >The 'numbers' are in 1000s of bytes per second processed. >type 16 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes >md5 60213.13k 175975.34k 389140.65k 551768.06k 630786.73k > > >Another is to calculate pi out to some number of digits and time it. >For example, on my laptop with a Core i7-4600U, calculating to 10,000 >digits takes nearly 2 minutes: > >~]$ time bc -l <<<'scale=10000;pi=4*a(1);0' >0 > >real 1m48.646s >user 1m48.590s >sys 0m0.011s > > >Meanwhile, my Trimslice with a 1GHz Tegra 2 takes over 10 minutes: > >~]$ time bc -l <<<'scale=10000;pi=4*a(1);0' >0 > >real 10m14.643s >user 10m14.523s >sys 0m0.028s > > >Jeff >_______________________________________________ >arm mailing list >arm@xxxxxxxxxxxxxxxxxxxxxxx >https://admin.fedoraproject.org/mailman/listinfo/arm _______________________________________________ arm mailing list arm@xxxxxxxxxxxxxxxxxxxxxxx https://admin.fedoraproject.org/mailman/listinfo/arm
