GitLab

Support `--host-list` for deploy and run commands.

Before this change, an inventory file is needed to use `sb deploy/run`.
Now, `--host-list localhost` or `-l localhost` is sufficient for quick try.

* Support custom output directory.
* Update document.

* Add introduction and release documents.
* Fix some typos in documents.

* fix bug for nvidia v100
* hard code the supported dict for different arch.

* add config file for ndv4.

* replace torch.optim.AdamW with transformers.AdamW.

* Initialize SuperBench website.
* Add GitHub Actions for automatically build and publish.

Update SuperBench documents.

* Add `sb deploy` command content.
* Fix inline if-expression syntax in playbook.
* Fix quote escape issue in bash command.
* Add custom env in config.
* Update default config for multi GPU benchmarks.
* Update MANIFEST.in to include jinja2 template.
* Require jinja2 minimum version.
* Fix occasional duplicate output in Ansible runner.
* Fix mixed color from Ansible and Python colorlog.
* Update according to comments.
* Change superbench.env from list to dict in config file.

Benchmarks: Add Feature - Add DistributedImpl and DistributedBackend arguments for micro benchmark. (#100)

rename bin name and result metric of cublas and cudnn microbenchmark

Update packages and add build cache for CUDA 11.1.1 Dockerfile:

* Remove duplicate cmake and ompi, which are already in base image
* Add hpcx and sharp lib
* Add cache for gitmodules build
* Sort apt-get packages

Fix bugs and refine log in single GPU benchmarks:

* Fix none framework issue
* Fix empty parameter bug
* Remove missed mobilenet_v3 models
* Change benchmark registration log to debug level
* Add pid in logging
* Add missing benchmarks in default config
* Fix deprecated logging warn

* Clean up the cache.

* fix return code reset issue

* Support local mode in runner.

* Change default shape of sharding-matmul.

* add cuda flops performance benchmark.

* add python related cudnn microbenchmark

* add error return code for cublas micro benchmark

* add cutlass as submodule.
* add build script for cutlass.
* only support compute capability 7.0(V100) and 8.0(A100)

* Benchmarks: Add benchmark - add source code of cudnn function micro benchmark

* Save benchmark results to json file.

* add benchmark for cublas test

* format

* revise error handling and test

* add interface to read json file, revise json file path and include .json in packaging

* add random_seed in arguments

* revise preprocess of cublas benchmark

* fix lint error and note error in source code

* update according comments

* revise input arguments from json file to custom str and convert json file to built-in dict list

* restore package config

* fit lint issue

* update platform and comments

* rename files to match source code dir and fix comments error
Co-authored-by: root <root@sb-validation-000001.51z1chmys5fuzfqyo4niepozre.bx.internal.cloudapp.net>

* Support `torch.distributed` mode in runner.
* Support given `proc_num` and `node_num` in `torch.distributed` mode.

* Superbenchmark: Add benchmarks - add cublas function micro benchmark

* format

* add python benchmark for cublas functions, example and test file

* detele python related and rename some files

* revise cmd_helper and move json package to cmake

* resolve conflict

* revise error handing to try-catch and update some code style

* revise cmd_helper.h, cublas_helper.h, cublas_helper.cpp

* revise structure of the cublas function

* add some comments and move cuda_init and cuda_free

* add comments for class member

* add ramdom seed, revise input from file to json string, simplify cmake

* delete json file in source code of cublas

* update according comments

* limit batchcount=1 in initialization of cublas function which do not use batch count

* revise and fix some errors of annotations

* update according comments and revise construction of CublasFunction
Co-authored-by: root <root@sb-validation-000001.51z1chmys5fuzfqyo4niepozre.bx.internal.cloudapp.net>

* Add integration tests for Ansible playbooks
* Add `gpu_vendor` var to bypass gpu mount

* Unify SB_MICRO_PATH and SB_MICRO_LIB

* fix bug of lib path

* Update Readme file on main page

Implement ansible client and runner:
* add ansible client
* add deploy and check_env playbooks

* add kernel launch overhead benchmark.

* add cuda kernel launch overhead benchmark - source part.
* can customize the nvcc_archs_support.
* set SB_MICRO_PATH for azure pipeline tests.

* call build script in Makefile.
* add cppbuild command for testing and docker env.

* use absolute path of input file
* parse registry uri from image
* merge common parts for arguments processing

* add GitHub Action to build and push image
* update Dockerfile to copy from context