https://15721.courses.cs.cmu.edu/spring2016/papers/kim-vldb2...
It's interesting they predicted sort would surpass hashing when the registers got to 512b, and we got AVX512 nowadays. TBH, their SSE4 sort implementation was quite complex (but beautiful).
https://15721.courses.cs.cmu.edu/spring2016/papers/kim-vldb2...
It's interesting they predicted sort would surpass hashing when the registers got to 512b, and we got AVX512 nowadays. TBH, their SSE4 sort implementation was quite complex (but beautiful).