Merge pull request #557 from deepmodeling/github-actions-selfhosted-runner

CI: Change actions to self-hosted runner.

Author: maorz1998

.github/workflows/CPU_inferencce_validation.yml

@@ -8,53 +8,30 @@ on:
      - master
 jobs:
   build_validation:
-    runs-on: [self-hosted, linux, x64]
+    runs-on: [self-hosted, linux, x64, 4060Ti]
     if: github.repository_owner == 'deepmodeling'     
-    container: 
-      image: ubuntu:20.04
-      options: --shm-size=5g  # shared memory size = 5GB
     steps:
      - name: Checkout
        uses: actions/checkout@v3
-     - name: install dependencies
-       env:
-        DEBIAN_FRONTEND: noninteractive
-       run: |
-        apt-get update 
-        apt-get install -y sudo wget unzip git software-properties-common make cmake g++ mpich openmpi-bin libopenmpi-dev libscalapack-mpi-dev vim git-core
-        sudo sh -c "wget -O - https://dl.openfoam.org/gpg.key | apt-key add -" 
-        sudo add-apt-repository http://dl.openfoam.org/ubuntu
-        sudo apt-get update
-        export DEBIAN_FRONTEND=noninteractive
-        apt-get -y install openfoam7 
-        wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
-        bash Miniconda3-latest-Linux-x86_64.sh -b
-        . ~/miniconda3/etc/profile.d/conda.sh
-        conda create -n libcantera python=3.8
-        conda activate libcantera
-        conda install -c cantera libcantera-devel=2.6
-        conda install -c cantera cantera
-        conda install pytorch pybind11
-        conda install pkg-config
-        conda install --channel https://conda.anaconda.org/zhaofeng-shu33 easydict
-        git clone https://github.com/deepmodeling/deepflame-dev.git
-        cd deepflame-dev
 
- 
      - name: build and validation with CPU inference 
        env: 
         OMPI_ALLOW_RUN_AS_ROOT: 1 
         OMPI_ALLOW_RUN_AS_ROOT_CONFIRM: 1 
         OMPI_MCA_btl_vader_single_copy_mechanism: none
        run: 
-        /bin/bash -c " echo $PWD && ls $PWD
-        && wget --content-disposition https://aisquare.oss-us-east-1.aliyuncs.com/data/datasets/14b50df5-dbe9-4f1c-bf58-032b8bc40a20
-        && unzip flare_CH4_SandiaD_4D.zip
+        /bin/bash -c " echo $PWD 
         && ls $PWD
-        && cp -r flare_CH4_SandiaD_4D.tbl examples/dfLowMachFoam/fgm/twoD_SandiaD_flareFGM
-        && source ~/miniconda3/etc/profile.d/conda.sh && conda activate libcantera && source /opt/openfoam7/etc/bashrc 
-        && . configure.sh --use_pytorch&& source ./bashrc && . install.sh 
-        && cd test && ./Allrun && conda deactivate "
+        && cp -r /root/actions-runner/data/flare_CH4_SandiaD_4D.tbl examples/dfLowMachFoam/fgm/twoD_SandiaD_flareFGM
+        && source ~/miniconda3/etc/profile.d/conda.sh 
+        && conda activate libcantera 
+        && source /opt/openfoam7/etc/bashrc
+        && . configure.sh --use_pytorch
+        && source ./bashrc 
+        && . install.sh 
+        && cd test 
+        && ./Allrun 
+        && conda deactivate "
 
      - name: test 
        run: | 

examples/dfLowMachFoam/fgm/twoD_SandiaD_flareFGM/Allrun

@@ -1,7 +1,7 @@
 #!/bin/sh
 cd ${0%/*} || exit 1    # Run from this directory
 
-if [ -e flare_CH4_SandiaD_4D.tbl]
+if [ -e flare_CH4_SandiaD_4D.tbl ]
 then
     echo "flare_CH4_SandiaD_4D.tbl exists. Make sure correct table has been used!"
 else

test/Tu500K-Phi1/DNN_model.pt

@@ -5,7 +5,7 @@
 import cantera as ct
 
 device_main = "cuda:0"
-device_list = range(torch.cuda.device_count())
+device_list = [0] #range(torch.cuda.device_count())
 
 torch.set_printoptions(precision=10)
 
@@ -95,16 +95,18 @@ def forward(self, x):
     Ymu2  = Ymu0
     Ystd2 = Ystd0
 
+    
+    """
     #load model  
     layers = [n_species +2, 1600, 800, 400, 1]
-
+    
     model0list = []
     for i in range(n_species-1):
         model0list.append(NN_MLP(layers))
     
     for i in range(n_species-1):
         model0list[i].load_state_dict(state_dict[f'net{i}'])
-
+    
 
     for i in range(n_species-1):
         model0list[i].eval()
@@ -113,7 +115,23 @@ def forward(self, x):
     if len(device_ids) > 1:
         for i in range(n_species-1):
             model0list[i] = torch.nn.DataParallel(model0list[i], device_ids=device_ids)
+    """
+    
+    #load model  
+    layers = [2+n_species]+[400]*4+[ n_species-1]
+    # layers = [2+n_species]+[800,400,200,100]+[n_species-1]
+    
+    
+    model = NN_MLP(layers)
+    
+    model.load_state_dict(state_dict['net'])
+    
+    model.eval()
+    model.to(device=device)
 
+    if len(device_ids) > 1:
+        model = torch.nn.DataParallel(model, device_ids=device_ids)
+    
 except Exception as e:
     print(e.args)
 
@@ -126,6 +144,8 @@ def inference(vec0):
     '''
     vec0 = np.abs(np.reshape(vec0, (-1, 3+n_species))) # T, P, Yi(7), Rho
     vec0[:,1] *= 101325
+    # vec0[:,1] *= 0
+    # vec0[:,1] += 101325
     mask = vec0[:,0] > frozenTemperature
     vec0_input = vec0[mask, :]
     print(f'real inference points number: {vec0_input.shape[0]}')
@@ -148,9 +168,11 @@ def inference(vec0):
             #inference
 
             output0_normalized = []
-            for i in range(n_species-1):
-                output0_normalized.append(model0list[i](input0_normalized))
-            output0_normalized = torch.cat(output0_normalized, dim=1)
+            
+            #for i in range(n_species-1):
+            #    output0_normalized.append(model0list[i](input0_normalized))
+            #output0_normalized = torch.cat(output0_normalized, dim=1)
+            output0_normalized = model(input0_normalized)
 
             # post_processing
             output0_bct = output0_normalized * Ystd0 + Ymu0 + input0_bct[:, 2:-1]