BigDL v0.8.0 Release Notes

Release Date: 2019-03-28 // about 5 years ago
  • Highlights

    • โž• Add MKL-DNN Int8 support, especially for VNNI acceleration support. Low precision inference accelerates both latency and throughput significantly
    • โž• Add support for runnning MKL-BLAS models under MKL-DNN. We leverage MKL-DNN to speed up both training and inference for MKL-BLAS models
    • โž• Add Spark 2.4 support. Our examples and APIs are fully compatible with Spark 2.4, we released the binary for Spark 2.4 together with other Spark versions

    Details

    • ๐Ÿ‘ [New Feature] Add MKL-DNN Int8 support, especially for VNNI support
    • ๐Ÿ‘ [New Feature] Add support for runnning MKL-BLAS models under MKL-DNN
    • ๐Ÿ‘ [New Feature] Add Spark 2.4 support
    • [New Feature] Add auto fusion to speed up model inference
    • ๐Ÿ‘ [New Feature] Memoery reorder support for low precision inference
    • ๐Ÿ‘ [New Feature] Add bytes support for DNN Tensor
    • [New Feature] Add SAME padding in MKL-DNN layers
    • [New Feature] Add combined (add/or) triggers for training completion
    • ๐Ÿ‘ [Enhancement] Inception-V1 python training support enhancement
    • โšก๏ธ [Enhancement] Distributed Optimizer enhancement to support customized optimizer
    • ๐Ÿ‘ [Enhancement] Add compute output shape for DNN supported layers
    • [Enhancement] New MKL-DNN computing thread pool
    • ๐Ÿ‘ [Enhancement] Add MKL-DNN support for Predictor
    • ๐Ÿ“š [Enhancement] Documentation enhancement for Sparse Tensor, MKL-DNN support, etc
    • [Enhancement] Add ceilm mode for AvgPooling and MaxPooling layers
    • ๐Ÿ‘ [Enhacement] Add binary classification support for DLClassifierModel
    • ๐Ÿ‘ [Enhacement] Improvement to support conversion between NHWC and NCHW for memory reoder
    • [Bug Fix] Fix SoftMax layer with narrowed input
    • ๐Ÿ‘ [Bug Fix] TensorFlow loader to support checking all data types
    • ๐Ÿ‘ [Bug Fix] Fix Add operation bug to support double type when loading TensorFlow graph
    • โšก๏ธ [Bug Fix] Fix one-step weight update missing issue in validation during training
    • ๐Ÿ”’ [Bug Fix] Fix scala compiler security issue in 2.10 & 2.11
    • [Bug Fix] Fix model broadcast cache UUID issue
    • [Bug Fix] Fix predictor issue for batch size == 1