FP64, FP32, FP16, BFLOAT16, TF32, and other members of the ZOO | by Grigory Sapunov | Medium
Running theano with float16 + tensor core operations
TVM: An Automated End-to-End Optimizing Compiler for Deep Learning | DeepAI
Train With Mixed Precision :: NVIDIA Deep Learning Performance Documentation
Video Series: Mixed-Precision Training Techniques Using Tensor Cores for Deep Learning | NVIDIA Technical Blog
Float16 status/follow-up · Issue #2908 · Theano/Theano · GitHub
PDF) Theano: A Python framework for fast computation of mathematical expressions
Caffe2: Portable High-Performance Deep Learning Framework from Facebook | NVIDIA Technical Blog
使用NVIDIA NGC加速AI开发流程
félteke Mus Üdvözlet theano extend gpu operator A nevében tüsszent Alice
TVM: An Automated End-to-End Optimizing Compiler for Deep Learning | DeepAI
Theano | PDF | C (Programming Language) | Program Optimization
TVM: An Automated End-to-End Optimizing Compiler for Deep Learning | DeepAI
New Features in CUDA 7.5 | NVIDIA Technical Blog
félteke Mus Üdvözlet theano extend gpu operator A nevében tüsszent Alice
lower precision computation floatX = float16, why not adding intX param in theano.config ? · Issue #5868 · Theano/Theano · GitHub
ValueError: Cannot construct a ufunc with more than 32 operands (similar to #2870) · Issue #3052 · Theano/Theano · GitHub
NVIDIA DGX-1 with Tesla V100 System Architecture White paper
Float16 status/follow-up · Issue #2908 · Theano/Theano · GitHub
Mixed-Precision Programming with CUDA 8 | NVIDIA Technical Blog
Float16 status/follow-up · Issue #2908 · Theano/Theano · GitHub
Accelerating AI Inference Workloads with NVIDIA A30 GPU | NVIDIA Technical Blog
Mixed-Precision Programming with CUDA 8 | NVIDIA Technical Blog
Train With Mixed Precision :: NVIDIA Deep Learning Performance Documentation
