Changes

Jump to: navigation, search

GPU

14 bytes removed, 11:56, 17 February 2019
wording in section Instruction Throughput
* 64 bit Integer Performance
: Current GPU architectures have no 64 bit wide [https://en.wikipedia.org/wiki/Arithmetic_logic_unit ALUs] are not 64 bit wide and have to emulate 64 bit integer operations.
* Mixed Precision Support
422
edits

Navigation menu