wkcn edited a comment on issue #13036: fix operators to support large arrays. URL: https://github.com/apache/incubator-mxnet/issues/13036#issuecomment-435591320 In my test, it seem that the performances of `+/-` between `int32_t` and `int64_t` are the same. CPU: Intel i7-7500U OS: Arch Linux x64 Compiler: g++ 8.2.1 Compiler Flag: `g++ ctype.cpp -o test -g -lpthread -std=c++11 -Wno-invalid-source-encoding` Test Code: https://github.com/wkcn/c_performance ``` Test int8_t 929 ms 798 ms 831 ms 2024 ms Test int16_t 860 ms 803 ms 840 ms 1950 ms Test int32_t 858 ms 822 ms 878 ms 1947 ms Test int64_t 899 ms 837 ms 828 ms 7345 ms Test float 1187 ms 1191 ms 1198 ms 1199 ms Test double 1209 ms 1211 ms 1205 ms 1205 ms ```
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
