History

Hanchen Ye 1f4200e88a mechanical rename hlsld to scalehls; update file structure		2020-09-06 18:05:16 -05:00
..
benchmarks	mechanical rename hlsld to scalehls; update file structure	2020-09-06 18:05:16 -05:00
mnist	mechanical rename hlsld to scalehls; update file structure	2020-09-06 18:05:16 -05:00
README.md	mechanical rename hlsld to scalehls; update file structure	2020-09-06 18:05:16 -05:00

README.md

Samples (samples)

Overview

Here is a collection of algorithms that we experimented with in the course of the project.

Toy Model (mnist)

This is a simple neural network used for MNIST digit classification. We successfully emitted the (affine dialect of) MLIR IR and TVM IR here.

Benchmarks (benchmarks)

Here are some benchmarks (3-mm, covariance, correlation, Alexnet) in the ICCAD paper we have attempted to replicate in MLIR without success. Actually, the Alexnet.onnx (Alexnet ONNX model) is too big so it is not here but one can just look it up online. In addition, we also include the first conv2d and maxpool2d operations from the toy model which obviously can be converted into MLIR currently.