4 ms·

The results he got for C and C++ make no sense and I question what optimization flag was actually used if any. From the looks of it, he uses "-O" which is equiv

The results he got for C and C++ make no sense and I question what optimization flag was actually used if any. From the looks of it, he uses "-O" which is equivalent to "-O1" which is really not the flag you want (this optimizes for both time and code size). *edit: with O2 I get a dramatic improvement and the C and C++ versions compile to essentially the same disassembly. Using O1 is definitely not going to produce an apples to apples comparison.

fgimian 8y ago
Thanks so much for your comment. I had actually tried -O3 during my tests and didn't notice a huge difference in this particular example. fots ~ $ gcc -O -o fib fib.c fots ~ $ time ./fib 433494437 real 0m0.749s user 0m0.743s sys 0m0.003s fots ~ $ gcc -O3 -o fib fib.c fots ~ $ time ./fib 433494437 real 0m0.747s user 0m0.742s sys 0m0.003s fots ~ $ gcc -O2 -o fib fib.c fots ~ $ time ./fib 433494437 real 0m0.741s user 0m0.737s sys 0m0.003s Clearly this is just a little mathematical algorithm and more real-world problems would likely produce a more dramatic difference. What sort of differences did you see on your end?
- lallysingh 8y ago
  How much optimization do you expect a loop with a simple addition to have? It doesn't seem a representative workload for an optimizer.
  - rdc12 8y ago
    Except it is not a loop[1], but a non-tail recursive function (with dual calls), if the compiler can optimise that to a loop that is actually impressive. Most likely function call overhead (and in the slower cases, interpreter overhead) is probably what is being measured in the best case. [1] In that it won't be a jump and check style loop.
    - mbel 8y ago
      > Most likely function call overhead Quite likely: https://godbolt.org/g/xGkCyA https://godbolt.org/g/xGkCyA although it really depends on the compiler (gcc is smarter here). He also seems to be benchmarking printing more than anything else (hence the difference between C and C++). Without versions of the compilers and explained methods of measurement, the results posted on the blog are more anecdotal knowledge than any kind of benchmark.

fgimian 8y ago

Thanks so much for your comment. I had actually tried -O3 during my tests and didn't notice a huge difference in this particular example. fots ~ $ gcc -O -o fib fib.c fots ~ $ time ./fib 433494437 real 0m0.749s user 0m0.743s sys 0m0.003s fots ~ $ gcc -O3 -o fib fib.c fots ~ $ time ./fib 433494437 real 0m0.747s user 0m0.742s sys 0m0.003s fots ~ $ gcc -O2 -o fib fib.c fots ~ $ time ./fib 433494437 real 0m0.741s user 0m0.737s sys 0m0.003s Clearly this is just a little mathematical algorithm and more real-world problems would likely produce a more dramatic difference. What sort of differences did you see on your end?

lallysingh 8y ago
How much optimization do you expect a loop with a simple addition to have? It doesn't seem a representative workload for an optimizer.
- rdc12 8y ago
  Except it is not a loop[1], but a non-tail recursive function (with dual calls), if the compiler can optimise that to a loop that is actually impressive. Most likely function call overhead (and in the slower cases, interpreter overhead) is probably what is being measured in the best case. [1] In that it won't be a jump and check style loop.
  - mbel 8y ago
    > Most likely function call overhead Quite likely: https://godbolt.org/g/xGkCyA https://godbolt.org/g/xGkCyA although it really depends on the compiler (gcc is smarter here). He also seems to be benchmarking printing more than anything else (hence the difference between C and C++). Without versions of the compilers and explained methods of measurement, the results posted on the blog are more anecdotal knowledge than any kind of benchmark.