Version v0.1.0 Release!

Overview

A Python Library provides APIs to move PyTorch Tensors between CPU and NVMe. It provides simple asynchronize I/O API, which helps you easily overlap computation and data moving. Instead of manually installing dependent C++ libraries, it can detect and install them automatically. It's user-friendly and efficient.

What's Changed

• polish code by @feifeibear in #1
• add pyi file by @ver217 in #3
• [hotfix] fix async read and add benchmark by @ver217 in #4
• add aio by @oahzxl in #2
• [feature] support sync IO by @ver217 in #5
• [feature] support register file by @ver217 in #6
• [feature] asyncio support readv/writev by @ver217 in #7
• add test files and fix some bugs by @oahzxl in #8
• [feature] Offloader support readv/writev by @ver217 in #9
• Update test section in readme by @oahzxl in #10
• update test_adam by @ver217 in #11
• test_adam use gpt2-xl by @ver217 in #12
• test_adam use tqdm and update default n_events by @ver217 in #13
• [test] Update catch to v2 and add install guidance by @ver217 in #14
• allow disabling backend when installing by @ver217 in #15
• Add CPUAdam and its test by @oahzxl in #16
• add 3rd party building by @feifeibear in #17
• Re-organize file with cmake by @oahzxl in #18
• Add offload test and backend probe by @oahzxl in #21
• [cli] add cli to check backends by @ver217 in #23
• auto install backends and allow selecting backends by @ver217 in #24
• update version and description by @ver217 in #25
• rename project and update readme by @ver217 in #26
• [hotfix] fix manifest and sdist version by @ver217 in #27
• fix version by @ver217 in #28
• [hotfix] fix libaio.so not found error by @ver217 in #29
• update install guide in readme by @ver217 in #30
• [docker] add dockerfile by @ver217 in #31

New Contributors

• @feifeibear made their first contribution in #1
• @ver217 made their first contribution in #3
• @oahzxl made their first contribution in #2

Full Changelog: https://github.com/hpcaitech/TensorNVMe/commits/v0.1.0