TABLE I:

(a) Cross-domain benchmark suite, and (b) domains and engines used in the benchmarks.

(a)
Name	Description	Domains	Used Capabilities
memory-enhance	Deep Brain Stimulation closed-loop control pipeline to optimize stimulation signals for memory enhancement	DSP	FFT
		Data Analytics	LR
		Optimized Control	MPC
robot-explorer	KinectFusion for 3D map generation with model predictive control for cave exploration	Robotics	KF
		Computer Vision	MPC
video-sync	Calculate correct offset to sync a movie and subtitle file	DSP	MPEG-Decode
			FFT
stock-market	Text sentiment classification on stock market news articles to estimate call option price	Data Analytics	LR
		Finance	Black-Scholes
leukocyte	Detect and tracks rolling leukocytes (white blood cells) in a microscopy of blood vessels vivo video	Computer Vision	GICOV
			MGVF
security-camera	Real-time object detection system which decodes MPEG encoded video	Deep Learning	Tiny-Yolo-v2
		DSP	MPEG-Decode
(b)
Domain	Capabilities	Engine *	Platform
DSP	FFT, MPEG-Decode	FFTW	CPU
		ffmpeg	CPU
		DeCO	FPGA
		LogiCore	FPGA
Data Analytics	LR	MLPack	CPU
		InAccel	FPGA
		Tabla	FPGA
Robotics	MPC	ACADO	CPU
		RoboX	FPGA
Computer Vision	GICOV, MGVF, KF	OpenCV	CPU
		SLAMBench	FPGA
		Iron	FPGA
Finance	Black-Scholes	QuantLib	CPU
		HyperStreams	FPGA
Deep Learning	Tiny-Yolo-v2	TensorFlow	CPU
		TVM	CPU
		DnnWeaver	FPGA

^*We omit the references due to the number of reference limit for submission.