Skip to content

Commit

Permalink
1.fix pytorch libomp issue 2.update pytorch version list
Browse files Browse the repository at this point in the history
  • Loading branch information
@licyk
licyk committed Sep 8, 2024
1 parent cf725aa commit c177804
Show file tree
Hide file tree
Showing 8 changed files with 418 additions and 237 deletions.
202 changes: 102 additions & 100 deletions help/how_to_use_term_sd.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -53,56 +53,57 @@ Term-SD 在使用安装、管理 AI 软件的功能时，会使用准备功能

### 2、PyTorch 版本选项
有以下版本组合：
- 1、Torch 2.4.0 (CUDA 11.8) + xFormers 0.0.27.post2
- 2、Torch 2.4.0 (CUDA 12.1) + xFormers 0.0.27.post2
- 3、Torch 2.4.0 (CPU)
- 4、Torch 2.3.1 (CUDA 11.8) + xFormers 0.0.27
- 5、Torch 2.3.1 (CUDA 12.1) + xFormers 0.0.27
- 6、Torch 2.3.1 (RoCM 6.0)
- 7、Torch 2.3.1 (DirectML)
- 8、Torch 2.3.1 (CPU)
- 9、Torch 2.3.0 (CUDA 11.8) + xFormers 0.0.26.post1
- 10、Torch 2.3.0 (CUDA 12.1) + xFormers 0.0.26.post1
- 11、Torch 2.3.0 (RoCM 6.0)
- 12、Torch 2.3.0 (CPU)
- 13、Torch 2.2.2 (CUDA 11.8) + xFormers 0.0.25.post1
- 14、Torch 2.2.2 (CUDA 12.1) + xFormers 0.0.25.post1
- 15、Torch 2.2.2 (RoCM 5.7)
- 16、Torch 2.2.2 (CPU)
- 17、Torch 2.2.1 (CUDA 11.8) + xFormers 0.0.25
- 18、Torch 2.2.1 (CUDA 12.1) + xFormers 0.0.25
- 19、Torch 2.2.1 (RoCM 5.7)
- 20、Torch 2.2.1 (DirectML)
- 21、Torch 2.2.1 (CPU)
- 22、Torch 2.2.0 (CUDA 11.8) + xFormers 0.0.24
- 23、Torch 2.2.0 (CUDA 12.1) + xFormers 0.0.24
- 24、Torch 2.2.0 (RoCM 5.7)
- 25、Torch 2.2.0 (CPU)
- 26、Torch 2.1.2 (CUDA 11.8) + xFormers 0.0.23.post1
- 27、Torch 2.1.2 (CUDA 12.1) + xFormers 0.0.23.post1
- 28、Torch 2.1.2 (RoCM 5.6)
- 29、Torch 2.1.2 (CPU)
- 30、Torch 2.1.1 (CUDA 11.8) + xFormers 0.0.23
- 31、Torch 2.1.1 (CUDA 12.1) + xFormers 0.0.23
- 32、Torch 2.1.1 (RoCM 5.6)
- 33、Torch 2.1.1 (CPU)
- 34、Torch 2.1.0 (Intel Arc)
- 35、Torch 2.1.0 (Intel Core Ultra)
- 36、Torch 2.1.0 (RoCM 5.6)
- 37、Torch 2.1.0 (CPU)
- 38、Torch 2.0.1 (CUDA 11.8) + xFormers 0.0.22
- 39、Torch 2.0.1 (RoCM 5.4.2)
- 40、Torch 2.0.1 (CPU)
- 41、Torch 2.0.0 (CUDA 11.8) + xFormers 0.0.18
- 42、Torch 2.0.0 (Intel Arc)
- 43、Torch 2.0.0 (DirectML)
- 44、Torch 2.0.0 (CPU)
- 45、Torch 1.13.1 (CUDA 11.7) + xFormers 0.0.16
- 46、Torch 1.13.1 (DirectML)
- 47、Torch 1.13.1 (CPU)
- 48、Torch 1.12.1 (CUDA 11.3) + xFormers 0.0.14
- 49、Torch + xFormers
- 50、Torch
- Torch 2.4.0 (CUDA 11.8) + xFormers 0.0.27.post2
- Torch 2.4.0 (CUDA 12.1) + xFormers 0.0.27.post2
- Torch 2.4.0 (CUDA 12.4)
- Torch 2.4.0 (CPU)
- Torch 2.3.1 (CUDA 11.8) + xFormers 0.0.27
- Torch 2.3.1 (CUDA 12.1) + xFormers 0.0.27
- Torch 2.3.1 (RoCM 6.0)
- Torch 2.3.1 (DirectML)
- Torch 2.3.1 (CPU)
- Torch 2.3.0 (CUDA 11.8) + xFormers 0.0.26.post1
- Torch 2.3.0 (CUDA 12.1) + xFormers 0.0.26.post1
- Torch 2.3.0 (RoCM 6.0)
- Torch 2.3.0 (CPU)
- Torch 2.2.2 (CUDA 11.8) + xFormers 0.0.25.post1
- Torch 2.2.2 (CUDA 12.1) + xFormers 0.0.25.post1
- Torch 2.2.2 (RoCM 5.7)
- Torch 2.2.2 (CPU)
- Torch 2.2.1 (CUDA 11.8) + xFormers 0.0.25
- Torch 2.2.1 (CUDA 12.1) + xFormers 0.0.25
- Torch 2.2.1 (RoCM 5.7)
- Torch 2.2.1 (DirectML)
- Torch 2.2.1 (CPU)
- Torch 2.2.0 (CUDA 11.8) + xFormers 0.0.24
- Torch 2.2.0 (CUDA 12.1) + xFormers 0.0.24
- Torch 2.2.0 (RoCM 5.7)
- Torch 2.2.0 (CPU)
- Torch 2.1.2 (CUDA 11.8) + xFormers 0.0.23.post1
- Torch 2.1.2 (CUDA 12.1) + xFormers 0.0.23.post1
- Torch 2.1.2 (RoCM 5.6)
- Torch 2.1.2 (CPU)
- Torch 2.1.1 (CUDA 11.8) + xFormers 0.0.23
- Torch 2.1.1 (CUDA 12.1) + xFormers 0.0.23
- Torch 2.1.1 (RoCM 5.6)
- Torch 2.1.1 (CPU)
- Torch 2.1.0 (Intel Arc)
- Torch 2.1.0 (Intel Core Ultra)
- Torch 2.1.0 (RoCM 5.6)
- Torch 2.1.0 (CPU)
- Torch 2.0.1 (CUDA 11.8) + xFormers 0.0.22
- Torch 2.0.1 (RoCM 5.4.2)
- Torch 2.0.1 (CPU)
- Torch 2.0.0 (CUDA 11.8) + xFormers 0.0.18
- Torch 2.0.0 (Intel Arc)
- Torch 2.0.0 (DirectML)
- Torch 2.0.0 (CPU)
- Torch 1.13.1 (CUDA 11.7) + xFormers 0.0.16
- Torch 1.13.1 (DirectML)
- Torch 1.13.1 (CPU)
- Torch 1.12.1 (CUDA 11.3) + xFormers 0.0.14
- Torch + xFormers
- Torch

选择版本时需要根据系统类型和显卡选择
- 在 Windows 系统中，Nvidia 显卡选择 Torch（CUDA）+ xFormers 的版本，AMD 显卡选择 Torch (Directml) 的版本，Intel 显卡选择 Torch (Intel Arc / Core Ultra) 的版本（核显选择 Intel Core Ultra 版本，独显选择 Intel Arc 版本）
Expand Down Expand Up @@ -255,56 +256,57 @@ Term-SD 支持 Stable Diffusion WebUI，ComfyUI，InvokeAI，Fooocus，lora-scri

#### 2、PyTorch 版本选择
第二个是PyTorch版本的选择界面，有以下版本组合
- 1、Torch 2.4.0 (CUDA 11.8) + xFormers 0.0.27.post2
- 2、Torch 2.4.0 (CUDA 12.1) + xFormers 0.0.27.post2
- 3、Torch 2.4.0 (CPU)
- 4、Torch 2.3.1 (CUDA 11.8) + xFormers 0.0.27
- 5、Torch 2.3.1 (CUDA 12.1) + xFormers 0.0.27
- 6、Torch 2.3.1 (RoCM 6.0)
- 7、Torch 2.3.1 (DirectML)
- 8、Torch 2.3.1 (CPU)
- 9、Torch 2.3.0 (CUDA 11.8) + xFormers 0.0.26.post1
- 10、Torch 2.3.0 (CUDA 12.1) + xFormers 0.0.26.post1
- 11、Torch 2.3.0 (RoCM 6.0)
- 12、Torch 2.3.0 (CPU)
- 13、Torch 2.2.2 (CUDA 11.8) + xFormers 0.0.25.post1
- 14、Torch 2.2.2 (CUDA 12.1) + xFormers 0.0.25.post1
- 15、Torch 2.2.2 (RoCM 5.7)
- 16、Torch 2.2.2 (CPU)
- 17、Torch 2.2.1 (CUDA 11.8) + xFormers 0.0.25
- 18、Torch 2.2.1 (CUDA 12.1) + xFormers 0.0.25
- 19、Torch 2.2.1 (RoCM 5.7)
- 20、Torch 2.2.1 (DirectML)
- 21、Torch 2.2.1 (CPU)
- 22、Torch 2.2.0 (CUDA 11.8) + xFormers 0.0.24
- 23、Torch 2.2.0 (CUDA 12.1) + xFormers 0.0.24
- 24、Torch 2.2.0 (RoCM 5.7)
- 25、Torch 2.2.0 (CPU)
- 26、Torch 2.1.2 (CUDA 11.8) + xFormers 0.0.23.post1
- 27、Torch 2.1.2 (CUDA 12.1) + xFormers 0.0.23.post1
- 28、Torch 2.1.2 (RoCM 5.6)
- 29、Torch 2.1.2 (CPU)
- 30、Torch 2.1.1 (CUDA 11.8) + xFormers 0.0.23
- 31、Torch 2.1.1 (CUDA 12.1) + xFormers 0.0.23
- 32、Torch 2.1.1 (RoCM 5.6)
- 33、Torch 2.1.1 (CPU)
- 34、Torch 2.1.0 (Intel Arc)
- 35、Torch 2.1.0 (Intel Core Ultra)
- 36、Torch 2.1.0 (RoCM 5.6)
- 37、Torch 2.1.0 (CPU)
- 38、Torch 2.0.1 (CUDA 11.8) + xFormers 0.0.22
- 39、Torch 2.0.1 (RoCM 5.4.2)
- 40、Torch 2.0.1 (CPU)
- 41、Torch 2.0.0 (CUDA 11.8) + xFormers 0.0.18
- 42、Torch 2.0.0 (Intel Arc)
- 43、Torch 2.0.0 (DirectML)
- 44、Torch 2.0.0 (CPU)
- 45、Torch 1.13.1 (CUDA 11.7)+ xFormers 0.0.16
- 46、Torch 1.13.1 (DirectML)
- 47、Torch 1.13.1 (CPU)
- 48、Torch 1.12.1 (CUDA 11.3)+ xFormers 0.0.14
- 49、Torch + xFormers
- 50、Torch
- Torch 2.4.0 (CUDA 11.8) + xFormers 0.0.27.post2
- Torch 2.4.0 (CUDA 12.1) + xFormers 0.0.27.post2
- Torch 2.4.0 (CUDA 12.4)
- Torch 2.4.0 (CPU)
- Torch 2.3.1 (CUDA 11.8) + xFormers 0.0.27
- Torch 2.3.1 (CUDA 12.1) + xFormers 0.0.27
- Torch 2.3.1 (RoCM 6.0)
- Torch 2.3.1 (DirectML)
- Torch 2.3.1 (CPU)
- Torch 2.3.0 (CUDA 11.8) + xFormers 0.0.26.post1
- Torch 2.3.0 (CUDA 12.1) + xFormers 0.0.26.post1
- Torch 2.3.0 (RoCM 6.0)
- Torch 2.3.0 (CPU)
- Torch 2.2.2 (CUDA 11.8) + xFormers 0.0.25.post1
- Torch 2.2.2 (CUDA 12.1) + xFormers 0.0.25.post1
- Torch 2.2.2 (RoCM 5.7)
- Torch 2.2.2 (CPU)
- Torch 2.2.1 (CUDA 11.8) + xFormers 0.0.25
- Torch 2.2.1 (CUDA 12.1) + xFormers 0.0.25
- Torch 2.2.1 (RoCM 5.7)
- Torch 2.2.1 (DirectML)
- Torch 2.2.1 (CPU)
- Torch 2.2.0 (CUDA 11.8) + xFormers 0.0.24
- Torch 2.2.0 (CUDA 12.1) + xFormers 0.0.24
- Torch 2.2.0 (RoCM 5.7)
- Torch 2.2.0 (CPU)
- Torch 2.1.2 (CUDA 11.8) + xFormers 0.0.23.post1
- Torch 2.1.2 (CUDA 12.1) + xFormers 0.0.23.post1
- Torch 2.1.2 (RoCM 5.6)
- Torch 2.1.2 (CPU)
- Torch 2.1.1 (CUDA 11.8) + xFormers 0.0.23
- Torch 2.1.1 (CUDA 12.1) + xFormers 0.0.23
- Torch 2.1.1 (RoCM 5.6)
- Torch 2.1.1 (CPU)
- Torch 2.1.0 (Intel Arc)
- Torch 2.1.0 (Intel Core Ultra)
- Torch 2.1.0 (RoCM 5.6)
- Torch 2.1.0 (CPU)
- Torch 2.0.1 (CUDA 11.8) + xFormers 0.0.22
- Torch 2.0.1 (RoCM 5.4.2)
- Torch 2.0.1 (CPU)
- Torch 2.0.0 (CUDA 11.8) + xFormers 0.0.18
- Torch 2.0.0 (Intel Arc)
- Torch 2.0.0 (DirectML)
- Torch 2.0.0 (CPU)
- Torch 1.13.1 (CUDA 11.7) + xFormers 0.0.16
- Torch 1.13.1 (DirectML)
- Torch 1.13.1 (CPU)
- Torch 1.12.1 (CUDA 11.3) + xFormers 0.0.14
- Torch + xFormers
- Torch

选择版本时需要根据系统类型和显卡选择
- 在 Windows 系统中，Nvidia 显卡选择 Torch（CUDA）+ xFormers 的版本，AMD 显卡选择 Torch (Directml) 的版本，Intel 显卡选择 Torch (Intel Arc / Core Ultra) 的版本（核显选择 Intel Core Ultra 版本，独显选择 Intel Arc 版本）
Expand Down
2 changes: 1 addition & 1 deletion modules/install_invokeai.sh
View file
Original file line number Diff line number Diff line change
Expand Up @@ -177,7 +177,7 @@ install_pypatchmatch_for_windows() {
fi
}

# 获取 PyPatchMatch 路径
# 获取 PyPatchMatch 路径(Python)
py_get_pypatchmatch_path() {
cat<<EOF
import importlib.metadata
Expand Down
29 changes: 25 additions & 4 deletions modules/install_prepare.sh
View file
Original file line number Diff line number Diff line change
Expand Up @@ -146,8 +146,10 @@ download_mirror_select() {
fi
}

# TODO 优化pytorch安装，如添加标志自动切换源，添加新版本的pytorch
# PyTorch 安装版本选择
# 选择后设置 INSTALL_PYTORCH_VERSION 全局变量
# 选择后设置 INSTALL_PYTORCH_VERSION 全局变量保存 PyTorch 版本信息
# PYTORCH_TYPE 保存使用的 PyTorch 镜像源种类, 如果为空则使用默认的镜像源
pytorch_version_select() {
local dialog_arg
unset INSTALL_PYTORCH_VERSION
Expand All @@ -158,8 +160,9 @@ pytorch_version_select() {
--ok-label "确认" --no-cancel \
--menu "请选择要安装的 PyTorch 版本, 注:\n1. Nvidia 显卡选择 CUDA 的版本\n2. AMD 显卡选择 RoCM(Linux) / DirectML(Windows) 版本\n3. Intel 显卡选择 IPEX Arc(独显) / Core Ultra(核显)\n4. Apple M 系列芯片选择无特殊标识版本\n5. 使用 CPU 运算选择 CPU 版本" \
$(get_dialog_size_menu) \
"50" "> Torch 2.4.0 (CUDA 11.8) + xFormers 0.0.27.post2" \
"49" "> Torch 2.4.0 (CUDA 12.1) + xFormers 0.0.27.post2" \
"51" "> Torch 2.4.0 (CUDA 11.8) + xFormers 0.0.27.post2" \
"50" "> Torch 2.4.0 (CUDA 12.1) + xFormers 0.0.27.post2" \
"49" "> Torch 2.4.0 (CUDA 12.4)" \
"48" "> Torch 2.4.0 (CPU)" \
"47" "> Torch 2.3.1 (CUDA 11.8) + xFormers 0.0.27" \
"46" "> Torch 2.3.1 (CUDA 12.1) + xFormers 0.0.27" \
Expand Down Expand Up @@ -241,6 +244,7 @@ pytorch_version_select() {
;;
9)
INSTALL_PYTORCH_VERSION="torch(ipex_Arc) 2.0.0"
PYTORCH_TYPE="ipex"
;;
10)
INSTALL_PYTORCH_VERSION="torch==2.0.0+cu118 torchvision==0.15.1+cu118 torchaudio==2.0.0+cu118 xformers==0.0.18"
Expand All @@ -262,9 +266,11 @@ pytorch_version_select() {
;;
16)
INSTALL_PYTORCH_VERSION="torch(ipex_Core_Ultra) 2.1.0"
PYTORCH_TYPE="ipex"
;;
17)
INSTALL_PYTORCH_VERSION="torch(ipex_Arc) 2.1.0"
PYTORCH_TYPE="ipex"
;;
18)
INSTALL_PYTORCH_VERSION="torch==2.1.1+cpu torchvision==0.16.1+cpu torchaudio==2.1.1+cpu"
Expand All @@ -286,6 +292,7 @@ pytorch_version_select() {
;;
24)
INSTALL_PYTORCH_VERSION="torch==2.1.2+cu121 torchvision==0.16.2+cu121 torchaudio==2.1.2+cu121 xformers==0.0.23.post1"
PYTORCH_TYPE="cu121"
;;
25)
INSTALL_PYTORCH_VERSION="torch==2.1.2+cu118 torchvision==0.16.2+cu118 torchaudio==2.1.2+cu118 xformers==0.0.23.post1+cu118"
Expand All @@ -298,6 +305,7 @@ pytorch_version_select() {
;;
28)
INSTALL_PYTORCH_VERSION="torch==2.2.0+cu121 torchvision==0.17.0+cu121 torchaudio==2.2.0+cu121 xformers==0.0.24"
PYTORCH_TYPE="cu121"
;;
29)
INSTALL_PYTORCH_VERSION="torch==2.2.0+cu118 torchvision==0.17.0+cu118 torchaudio==2.2.0+cu118 xformers==0.0.24+cu118"
Expand All @@ -313,6 +321,7 @@ pytorch_version_select() {
;;
33)
INSTALL_PYTORCH_VERSION="torch==2.2.1+cu121 torchvision==0.17.1+cu121 torchaudio==2.2.1+cu121 xformers==0.0.25"
PYTORCH_TYPE="cu121"
;;
34)
INSTALL_PYTORCH_VERSION="torch==2.2.1+cu118 torchvision==0.17.1+cu118 torchaudio==2.2.1+cu118 xformers==0.0.25+cu118"
Expand All @@ -325,6 +334,7 @@ pytorch_version_select() {
;;
37)
INSTALL_PYTORCH_VERSION="torch==2.2.2+cu121 torchvision==0.17.2+cu121 torchaudio==2.2.2+cu121 xformers==0.0.25.post1"
PYTORCH_TYPE="cu121"
;;
38)
INSTALL_PYTORCH_VERSION="torch==2.2.2+cu118 torchvision==0.17.2+cu118 torchaudio==2.2.2+cu118 xformers==0.0.25.post1+cu118"
Expand All @@ -337,6 +347,7 @@ pytorch_version_select() {
;;
41)
INSTALL_PYTORCH_VERSION="torch==2.3.0+cu121 torchvision==0.18.0+cu121 torchaudio==2.3.0+cu121 xformers==0.0.26.post1"
PYTORCH_TYPE="cu121"
;;
42)
INSTALL_PYTORCH_VERSION="torch==2.3.0+cu118 torchvision==0.18.0+cu118 torchaudio==2.3.0+cu118 xformers==0.0.26.post1+cu118"
Expand All @@ -352,18 +363,26 @@ pytorch_version_select() {
;;
46)
INSTALL_PYTORCH_VERSION="torch==2.3.1+cu121 torchvision==0.18.1+cu121 torchaudio==2.3.1+cu121 xformers==0.0.27"
PYTORCH_TYPE="cu121"
;;
47)
INSTALL_PYTORCH_VERSION="torch==2.3.1+cu118 torchvision==0.18.1+cu118 torchaudio==2.3.1+cu118 xformers==0.0.27+cu118"
PYTORCH_TYPE="cu118"
;;
48)
INSTALL_PYTORCH_VERSION="torch==2.4.0+cpu torchvision==0.19.0+cpu torchaudio==2.4.0+cpu"
;;
49)
INSTALL_PYTORCH_VERSION="torch==2.4.0+cu121 torchvision==0.19.0+cu121 torchaudio==2.4.0+cu121 xformers==0.0.27.post2"
INSTALL_PYTORCH_VERSION="torch==2.4.0+cu124 torchvision==0.19.0+cu124 torchaudio==2.4.0+cu124"
PYTORCH_TYPE="cu124"
;;
50)
INSTALL_PYTORCH_VERSION="torch==2.4.0+cu121 torchvision==0.19.0+cu121 torchaudio==2.4.0+cu121 xformers==0.0.27.post2"
PYTORCH_TYPE="cu121"
;;
51)
INSTALL_PYTORCH_VERSION="torch==2.4.0+cu118 torchvision==0.19.0+cu118 torchaudio==2.4.0+cu118 xformers==0.0.27.post2+cu118"
PYTORCH_TYPE="cu118"
;;
esac
}
Expand Down Expand Up @@ -584,6 +603,7 @@ clean_install_config() {
term_sd_echo "PIP_FORCE_REINSTALL_ARG: ${PIP_FORCE_REINSTALL_ARG}"
term_sd_echo "PIP_UPDATE_PACKAGE_ARG: ${PIP_UPDATE_PACKAGE_ARG}"
term_sd_echo "PIP_PREFER_BINARY_ARG: ${PIP_PREFER_BINARY_ARG}"
term_sd_echo "PYTORCH_TYPE: ${PYTORCH_TYPE}"
fi

unset PIP_INDEX_MIRROR # 指定 Pip 镜像源的参数
Expand All @@ -600,6 +620,7 @@ clean_install_config() {
unset PIP_FORCE_REINSTALL_ARG # 是否在 Pip 使用 --force-reinstall 参数
unset PIP_UPDATE_PACKAGE_ARG # 是否更新软件包
unset PIP_PREFER_BINARY_ARG # 使用 --prefer-binary 使 Pip 优先使用编译好的 Python 软件包进行安装
unset PYTORCH_TYPE # PyTorch 种类, 用于切换 PyTorch 镜像源
}

# 如果启用了 Pip 镜像源, 则返回0
Expand Down
Loading

0 comments on commit c177804

Please sign in to comment.