CNN-Inference-Engine-Quick-View

A quick view of high-performance convolution neural networks (CNNs) inference engines on mobile devices.

Runtime-speed Comparisons

• Mobile-AI-Benchmarks
• AI-Benchmarks

FLOAT32-Support

Framework	Main Platform	Model Compatibility	Detection-Support	Speed Benchmarks
Bolt	CPU (ARM optimized) / x86 / Mali GPU	Caffe / Tensorflow / PyTorch / onnx	Y	Link
TNN	CPU (ARM optimized) / Mali Adreno Apple GPU	Caffe / Tensorflow / PyTorch	Y	Link
Paddle-Light	CPU (ARM optimized) / Mali GPU / FPGA / NPU	Paddle / Caffe / onnx	Y	Link
MNN	CPU (ARM optimized) / Mali GPU	Caffe / Tensorflow / onnx	Y	Link
NCNN	CPU (ARM optimized) / Mali GPU	Caffe / PyTorch / mxnet / onnx	Y	3rd party Link / Official Link
MACE	CPU (ARM optimized) / Mali GPU / DSP	Caffe / Tensorflow / onnx	Y	Link
Synet	CPU (ARM optimized) / x86	Caffe / PyTorch / Tensorflow / mxnet / onnx	Y	-
MsnhNet	CPU (ARM optimized) / Mali GPU / x86 / TensorRT	PyTorch	Y	Link
ONNX-Runtime	CPU / Nvidia GPU	onnx	Y	-
HiAI	Kirin CPU / NPU	Caffe / Tensorflow	Y	-
NNIE	NPU	Caffe	Y	1TOPs
Intel-Caffe	CPU (Intel optimized)	Caffe	Y	Link
FeatherCNN	CPU (ARM optimized)	Caffe	N	Link / unofficial Link
TEngine	CPU (ARM A72 optimized)	Caffe / mxnet	Y	Link
Tensorflowlite	CPU (Android optimized)	Caffe2 / Tensorflow / onnx	Y	Link
TensorRT	GPU (Volta optimized)	Caffe / Tensorflow / onnx	Y	Link
TVM	CPU (ARM optimized) / Mali GPU / FPGA	onnx	Y	-
SNPE	CPU (Qualcomm optimized) / GPU / DSP	Caffe / Caffe2 / Tensorflow/ onnx	Y	Link
Pocket-Tensor	CPU (ARM/x86 optimized)	Keras	N	Link
ZQCNN	CPU	Caffe / mxnet	Y	Link
ARM-NEON-to-x86-SSE	CPU (Intel optimized)	Intrinsics-Level	-	-
Simd	CPU (all platform optimized)	Intrinsics-Level	-	-
clDNN	Intel® Processor Graphics / Iris™ Pro Graphics	Caffe / Tennsorflow / mxnet / onnx	Y	Link

FIX16-Support

Framework	Main Platform	Model Compatibility	Detection-Support	Speed Benchmarks
Bolt	CPU (ARM optimized) / x86 / Mali GPU	Caffe / Tensorflow / PyTorch	Y	Link
ARM32-SGEMM-LIB	CPU (ARM optimized)	GEMM Library	N	Link
TNN	CPU (ARM optimized) / Mali Adreno Apple GPU	Caffe / Tensorflow / PyTorch	Y	Link
Yolov2-Xilinx-PYNQ	FPGA (Xilinx PYNQ)	Yolov2-only	Y	Link

INT8-Support

Framework	Main Platform	Model Compatibility	Detection-Support	Speed Benchmarks
Bolt	CPU (ARM optimized) / x86 / Mali GPU	Caffe / Tensorflow / PyTorch	Y	Link
Intel-Caffe	CPU (Intel Skylake)	Caffe	Y	Link
TNN	CPU (ARM optimized) / Mali Adreno Apple GPU	Caffe / Tensorflow / PyTorch	Y	Link
NCNN	CPU (ARM)	Caffe / pytorch / mxnet / onnx	Y	Link
Paddle-Light	CPU (ARM optimized) / Mali GPU / FPGA	Paddle / Caffe / onnx	Y	Link
MNN	CPU (ARM optimized) / Mali GPU	Caffe / Tensorflow / onnx	Y	Link
Tensorflowlite	CPU (Android)	Caffe2 / Tensorflow / onnx	Y	Link
TensorRT	GPU (Volta)	Caffe / Tensorflow / onnx	Y	Link
Gemmlowp	CPU (ARM / x86)	GEMM Library	-	-
SNPE	DSP (Quantized DLC)	Caffe / Caffe2 / Tensorflow/ onnx	Y	Link
MACE	CPU (ARM optimized) / Mali GPU / DSP	Caffe / Tensorflow / onnx	Y	Link
TF2	FPGA	Caffe / PyTorch / Tensorflow	Y	Link

TERNARY-Support

Framework	Main Platform	Model Compatibility	Detection-Support	Speed Benchmarks
Gemmbitserial	CPU (ARM / x86)	GEMM Library	-	Link

BINARY-Support

Framework	Main Platform	Model Compatibility	Detection-Support	Speed Benchmarks
Bolt	CPU (ARM optimized) / x86 / Mali GPU	Caffe / Tensorflow / PyTorch	Y	Link
BMXNET	CPU (ARM / x86) / GPU	mxnet	Y	Link
DABNN	CPU (ARM)	Caffe / Tensorflow / onnx	N	Link
Espresso	GPU	-	N	Link
BNN-PYNQ	FPGA (Xilinx PYNQ)	-	N	Link
FINN	FPGA (Xilinx)	-	N	Link

NLP-Support

Framework	Main Platform	Model Compatibility	Speed Benchmarks
TurboTransformers	CPU / Nvidia GPU	PyTorch	Link
Bolt	CPU / Mali GPU	Caffe / onnx	Link

*: Conv-BN-Scale-fused