Nano: update os support (#7491)

This commit is contained in:
Yishuo Wang 2023-02-09 15:48:30 +08:00 committed by GitHub
parent 9c129ec158
commit 56d699d935

View file

@ -23,8 +23,8 @@
| **Jemalloc** | ✅ | ✅ | ✅ | ❌ | ❌ | | **Jemalloc** | ✅ | ✅ | ✅ | ❌ | ❌ |
| **Tcmalloc** | ✅ | ❌ | ❌ | ❌ | ❌ | | **Tcmalloc** | ✅ | ❌ | ❌ | ❌ | ❌ |
| **Neural-Compressor** | ✅ | ✅ | ❌ | ❌ | ? | | **Neural-Compressor** | ✅ | ✅ | ❌ | ❌ | ? |
| **OpenVINO** | ✅ | ① | ❌ | ❌ | ? | | **OpenVINO** | ✅ | ① | ❌ | ❌ | |
| **ONNXRuntime** | ✅ | ① | ✅ | ❌ | ? | | **ONNXRuntime** | ✅ | ① | ✅ | ❌ | |
## PyTorch Feature Support ## PyTorch Feature Support
@ -33,16 +33,16 @@
| **CUDA patch** | ✅ | ✅ | ✅ | ? | ✅ | | **CUDA patch** | ✅ | ✅ | ✅ | ? | ✅ |
| **JIT** | ✅ | ✅ | ✅ | ? | ✅ | | **JIT** | ✅ | ✅ | ✅ | ? | ✅ |
| **Channel last** | ✅ | ✅ | ✅ | ? | ✅ | | **Channel last** | ✅ | ✅ | ✅ | ? | ✅ |
| **BF16** | ✅ | ✅ | ⭕ | ⭕ | ? | | **BF16** | ✅ | ✅ | ⭕ | ⭕ | |
| **IPEX** | ✅ | ✅ | ❌ | ❌ | ❌ | | **IPEX** | ✅ | ✅ | ❌ | ❌ | ❌ |
| **Multi-instance** | ✅ | ✅ | ② | ② | ? | | **Multi-instance** | ✅ | ✅ | ② | ② | |
## TensorFlow Feature Support ## TensorFlow Feature Support
| Feature | Ubuntu (20.04/22.04) | CentOS7 | MacOS (Intel chip) | MacOS (M-series chip) | Windows | | Feature | Ubuntu (20.04/22.04) | CentOS7 | MacOS (Intel chip) | MacOS (M-series chip) | Windows |
| ------------------ | -------------------- | ------- | ------------------ | --------------------- | ------- | | ------------------ | -------------------- | ------- | ------------------ | --------------------- | ------- |
| **BF16** | ✅ | ✅ | ⭕ | ⭕ | ? | | **BF16** | ✅ | ✅ | ⭕ | ⭕ | |
| **Multi-instance** | ③ | ③ | ②③ | ②③ | ? | | **Multi-instance** | ③ | ③ | ②③ | ②③ | |
## Symbol Meaning ## Symbol Meaning
@ -54,4 +54,5 @@
| ① | This feature is only supported when used together with jemalloc | | ① | This feature is only supported when used together with jemalloc |
| ② | This feature is supported but without any performance guarantee | | ② | This feature is supported but without any performance guarantee |
| ③ | Only Multi-instance training is supported for now | | ③ | Only Multi-instance training is supported for now |
| ④ | This feature is only supported when using PyTorch |
| ? | Not tested | | ? | Not tested |