This article may rely excessively on sources too closely associated with the subject, potentially preventing the article from being verifiable and neutral. (March 2021) |
Intel's Deep Learning Boost (DL Boost) is a marketing name for instruction set architecture (ISA) features on the x86-64 designed to improve performance on deep learning tasks such as training and inference.[1]
Features
editDL Boost consists of two sets of features:
- AVX-512 VNNI, 4VNNIW, or AVX-VNNI: fast multiply-accumulation mainly for convolutional neural networks.
- AVX-512 BF16: lower-precision bfloat16 floating-point numbers for generally faster computation. Operations provided include conversion to/from float32 and dot product.
DL Boost features were introduced in the Cascade Lake architecture.
A TensorFlow-based benchmark run on the Google Cloud Platform Compute Engine shows improved performance and reduced cost compared to previous CPUs and to GPUs, especially for small batch sizes.[2]
Notes
editExternal links
edit- Deep Learning Boost at Intel
- Andres Rodrigues et al., "Lower Numerical Precision Deep Learning Inference and Training", Intel White paper [3]
- Intel and ML (2017), from Intel's Developer Relations Division