v1.4.0

• Comm|Scope 0.9.0-30c3570f-master

• cudnn|Scope 1.0.0-3a643afb-master

• Example|Scope 1.0.1-0f5db5c0-master

• NCCL|Scope 84da171d6fd15ce647159b1bb5c67170af6b573a

• Friendlier error formatting for VS Code

• Fix an issue with Benchmark submodule and CMake 3.17.2

• Add USE_NVTX option for Nvidia Tools Extensions

v1.3.2

• Comm|Scope 0.8.2-b342d608-master

• cudnn|Scope 1.0.0-3a643afb-master

• Example|Scope 1.0.1-0f5db5c0-master

• NCCL|Scope 84da171d6fd15ce647159b1bb5c67170af6b573a

• Move Benchmark library from Hunter to submodule

v1.3.1

• Comm|Scope 0.8.1-2d1a32e8-master
• cudnn|Scope 1.0.0-3a643afb-master
• Example|Scope 1.0.1-0f5db5c0-master
• NCCL|Scope 84da171d6fd15ce647159b1bb5c67170af6b573a

v1.3.0

• Comm|Scope 0.8.0-82f0364a-master
• cudnn|Scope 1.0.0-3a643afb-master
• Example|Scope 1.0.1-0f5db5c0-master
• NCCL|Scope 84da171d6fd15ce647159b1bb5c67170af6b573a

v1.2.0

• Comm|Scope 0.7.1-48e1ce32-master
• cudnn|Scope 1.0.0-3a643afb-master
• Example|Scope 1.0.1-0f5db5c0-master
• NCCL|Scope 84da171d6fd15ce647159b1bb5c67170af6b573a
• silence some warnings

v1.1.0

• Comm|Scope 0.6.3-bd264d4d-master
• cudnn|Scope 1.0.0-3a643afb-master
• Example|Scope 1.0.1-0f5db5c0-master
• NCCL|Scope 84da171d6fd15ce647159b1bb5c67170af6b573a
• CMake CUDA language support instead of FindCUDA

v1.0.0

• Comm|Scope v0.6.3
• Example|Scope v1.0.1
• NCCL|Scope v0.1.1
• vastly improve documentation
• pin dependencies to spdlog 0.16.3-p1, fmt 4.1.0, benchmark 1.4.0, sugar 1.3.0
• allow benchmarks to be dynamically registered
• generate benchmark scripts for ICPE 2019

v0.4.0

• Automatically load scopes from scope/scopes
• Use Clara for command line flags
• add scope_fatal for CMake
• add a website
• add an after_init lifecycle phase
• provide target_include_scope_directories and target_link_scope_libraries for CMake
• change docker images to be base docker images. Scopes should build their own Docker images.
• Don't hard-code nvcc location.
• add unique_cuda_ids to init/cuda
• add tools/generate_runscripts.py to generate scripts to run different benchmarks in separate processes.

v0.3.0

• Comm|Scope v0.2.0
• Add support for parsing lists of ints from command line
• Add --cuda_device_ids flag

v0.2.0

Bump version: 0.1.0 → 0.2.0