Skip to content

Latest commit

 

History

History
152 lines (129 loc) · 8.92 KB

CHANGELOG.md

File metadata and controls

152 lines (129 loc) · 8.92 KB
Raw

NVIDIA CUTLASS Changelog

CUTLASS 2.x

2.4.0 (2020-11-19)

  • Implicit GEMM convolution kernels supporting CUDA and Tensor Cores on NVIDIA GPUs
    • Operators: forward (Fprop), backward data gradient (Dgrad), and backward weight gradient (Wgrad) convolution
    • Data type: FP32, complex, Tensor Float 32 (TF32), BFloat16 (BF16), Float16, Int4, Int8, Int32
    • Spatial dimensions: 1-D, 2-D, and 3-D
    • Layout: NHWC, NCxHWx
  • Implicit GEMM convolution components:
    • Global memory iterators supporting Fprop, Dgrad, and Wgrad
    • MmaMultistage for implicit GEMM convolution for NVIDIA Ampere architecture
    • MmaPipeline for implicit GEMM convolution for NVIDIA Volta and Turing architectures
    • Documentation describing Implicit GEMM Convolution algorithm and implementation

2.3.0 (2020-09-23)

2.2.0 (2020-06-08)

  • NVIDIA Ampere Architecture features
    • Fast Tensor Core operations:
    • Maximum performance via mma.sync
    • Tensor Float 32, BFloat16, and double-precision data types
    • Mixed integer data types (int8, int4, bin1)
    • Asynchronous copy for deep software pipelines via cp.async
    • Described in GTC 2020 Webinar (SR 21745) (free registration required)
  • Features:
    • SDK examples showing GEMM fused with bias+relu and fused GEMM+GEMM
    • Complex-valued GEMMs targeting NVIDIA Ampere Tensor Cores in double-precision and Tensor Float 32
    • Gaussian complex GEMMs using 3m complex multiply algorithm
    • Universal GEMM kernel supporting two batch modes and two algorithms for parallel reductions
  • Policy updates:
    • CUDA 11 Toolkit needed to enable NVIDIA Ampere Architecture features
    • Disabled F16C by default for compatibility - enable on cmake command line with -DCUTLASS_ENABLE_F16C=ON

2.1.0 (2020-04-06)

  • BLAS-style host-side API added to CUTLASS Library
    • API to launch compiled kernel instances for GEMM and planar complex GEMM
  • Planar Complex GEMM kernels targeting Volta and Turing Tensor Cores
  • Minor enhancements and bug fixes

2.0.0 (2019-11-19)

  • Substantially refactored for
    • Better performance, particularly for native Turing Tensor Cores
    • Robust and durable templates spanning the design space
    • Encapsulated functionality embodying modern C++11 programming techniques
    • Optimized containers and data types for efficient, generic, portable device code
  • Updates to:
  • Native Turing Tensor Cores
    • Efficient GEMM kernels targeting Turing Tensor Cores
    • Mixed-precision floating point, 8-bit integer, 4-bit integer, and binarized operands
  • Coverage of existing CUTLASS functionality
    • GEMM kernels targeting CUDA and Tensor Cores in NVIDIA GPUs
    • Volta Tensor Cores through native mma.sync and through WMMA API
    • Optimizations such as parallel reductions, threadblock rasterization, and intra-threadblock reductions
    • Batched GEMM operations
    • Complex-valued GEMMs
  • Note: a host compiler supporting C++11 or greater is required.

CUTLASS 1.x

1.3.2 (2019-07-09)

  • Performance improvement for Volta Tensor Cores TN and TT layouts.

1.3.1 (2019-04-09)

  • Corrected NVRTC unit tests.

1.3.0 (2019-03-20)

  • Efficient GEMM kernel targeting Volta Tensor Cores via mma.sync instruction added in CUDA 10.1.

1.2.0 (2018-10-26)

  • Parallelized reductions across threadblocks ("Split-K")
    • Improved IGEMM performance
  • Batched strided WMMA GEMMs

1.1.0 (2018-09-19)

  • Turing Features
    • WMMA GEMM targeting TensorCores - INT8, INT4, 1-bit
  • Batched Strided GEMM
  • Threadblock rasterization strategies
    • Improved performance for adverse problem sizes and data layouts
  • Extended CUTLASS Core comonents
    • Tensor views support arbitrary matrix and tensor layouts
    • Zip iterators for structuring multiple data streams
  • Enhanced CUTLASS utilities
    • Reference code for tensor operations in host and device code
    • Added HostMatrix<> for simplified matrix creation
  • Examples
    • Basic GEMM, tensor views, CUTLASS utilities, batched GEMM, WMMA GEMM

1.0.1 (2018-06-11)

  • Intra-threadblock reduction added for small threadblock tile sizes
    • sgemm_64x128x16, sgemm_128x128x16, sgemm_128x64x16, sgemm_128x32x16, sgemm_64x64x16, sgemm_64x32x16
    • igemm_32x32x128
  • GEMM K residue handled during prologue prior to mainloop
  • Replaced Google Test copy with submodule. Use git submodule init --recursive --update

1.0.0 (2018-05-16)

  • Substantial rewrite to accommodate new architecture
  • Kernels: SGEMM, DGEMM, IGEMM, HGEMM, WMMA GEMM
  • Unit and performance tests

0.0.1 (2017-12-04)

  • Initial release

Copyright

Copyright (c) 2017-2021, NVIDIA CORPORATION. All rights reserved.

  Redistribution and use in source and binary forms, with or without modification, are permitted
  provided that the following conditions are met:
      * Redistributions of source code must retain the above copyright notice, this list of
        conditions and the following disclaimer.
      * Redistributions in binary form must reproduce the above copyright notice, this list of
        conditions and the following disclaimer in the documentation and/or other materials
        provided with the distribution.
      * Neither the name of the NVIDIA CORPORATION nor the names of its contributors may be used
        to endorse or promote products derived from this software without specific prior written
        permission.

  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR
  IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
  FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL NVIDIA CORPORATION BE LIABLE
  FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
  BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
  OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
  STRICT LIABILITY, OR TOR (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
  OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.