README

Welcome. This is an example implementation of the Fused Table Scan.

To run the benchmarks:
	1) Install libpapi and a recent version of gcc
	2) run `make`
	3) run the different binaries in the `binaries` folder or run all benchmarks using `python benchmark.py`

If you are interested in the implementation, you will find it in `avx_scan.cpp`. The two methods of interest are `avx_scan` and `avx_second_scan`. These show sample code as it would be generated by the JIT compiler.

For benchmarks, we supplied implementations different from the AVX-512 version with 512-bit registers. These can be enabled by compiling with -DREG=128, -DREG=256, and -DREG=512. For 128-bit registers, we also have a version without AVX-512 instructions, that can be enabled with -DAVX512=0.

In case of any questions, please don't hesitate to contact us.