As someone who has built a columnstore database and tested it on that query shape (group-by over a column with only a few distinct values) Its possible to go faster than 1 cycle per row via SIMD and operating directly on compressed data (0.87 cycles per row). This blog post gives some details on how it’s done (https://www.memsql.com/blog/how-to-process-trillion-rows-per...)
Comments
As someone who has built a columnstore database and tested it on that query shape (group-by over a column with only a few distinct values) Its possible to go faster than 1 cycle per row via SIMD and operating directly on compressed data (0.87 cycles per row). This blog post gives some details on how it’s done (https://www.memsql.com/blog/how-to-process-trillion-rows-per...)