Home

Machine Learning Concepts

• Types: Supervised Learning | Unsupervised Learning | Reinforcement learning
• Networks: Convolutional Neural Network | Recurrent Neural Network
• Activation function | ReLU | SoftMax
• Data formats - FP16 | FP32 | bfloat16 | INT8

ML benchmarks

• Image classification: ResNet | VGG | Inception | MobileNets | AlexNet | ResNeXt
• NLP: BERT
• MLPerf - MLPerf Inference | MLPerf Training
• Data sets - MNIST database | ImageNet | COCO | PASCAL VOC

Hardware

• x86 CPU : Intel | AMD
• Arm CPU: AWS | Marvell Technology Group | Fujitsu | Huawei | NVIDIA | Ampere Computing
• GPU : AMD | NVIDIA
• ML Accelerators: GraphCore | Habana Labs | Groq
• Connectors: PCIe | NVLink | Infiniband | Ethernet

ML Software

• Frameworks: TensorFlow | PyTorch | MXNET | Keras
• Inference engines: ArmNN | OpenVINO | TensorRT | TensorFlow Lite | Core ML
• Libraries: cuDNN | MKL-DNN | Arm Compute Library
• BLAS libraries: Eigen | OpenBLAS | MKL | BLIS | cuBLAS
• Compiler tech: MLIR | TensorFlow XLA | TVM

Arm Architecture

• A profile - ARMv8.0-A | ARMv8.1-A | ARMv8.2-A | ARMv8.3-A | ARMv8.4-A | ARMv8.5-A | ARMv8.6-A
• R profile - ARMv7-R | ARMv8-R
• M profile - ARMv6-M | ARMv7-M | ARMv7E-M | ARMv8-M

Languages and tools

• Languages - Python | Julia | R
• Compiler - GNU Toolchain | LLVM
• Others - Valgrind | QEMU

Python related

• Interpreters: CPython
• Packaging: pip (package manager) | Conda
• Distributions: Anaconda
• Repository: PyPi
• Libraries: NumPy | SciPy

Cloud

• Vendors: AWS | Microsoft Azure | Google Cloud
• Linux distributions: RHEL | SLES | CentOS | OpenSUSE | Ubuntu | Amazon Linux 2
• Containers: Docker | Kubernetes | KVM

AWS

• AWS SageMaker | AWS Lambda | AWS EC2 | AWS Elastic Inference
• AWS Graviton | AWS Inferentia

General Benchmarks

• SPEC - SPEC CPU 2017 | SPEC OMP 2012 | SPEC MPI 2007