Add stable diffusion fp32 case

FlagOpen · Aug 31, 2023 · b95556a · b95556a
1 parent f39b1f4
commit b95556a
Show file tree

Hide file tree

Showing 3 changed files with 5 additions and 5 deletions.
diff --git a/inference/configs/stable_diffusion_v1_4/configurations.yaml b/inference/configs/stable_diffusion_v1_4/configurations.yaml
@@ -3,14 +3,12 @@ batch_size: 2
 # Attention! For transformer decoder like bert, 1 token cause 2*param flops, so we need 2*length*params like 2*512*0.33B here
 # format: a_1*a*2*...*a_nea_0,like 2*512*0.33e9(bert) or 4.12e9(resnet50)
 flops: 6.78e11
-fp16: false
-compiler: tensorrt
 num_workers: 8
 log_freq: 5
 repeat: 1
 # skip validation(will also skip create_model, export onnx). Assert exist_onnx_path != null
-no_validation: false
 # set a real onnx_path to use exist, or set it to anything but null to avoid export onnx manually(like torch-tensorrt)
-exist_onnx_path: null
+exist_onnx_path: ../../onnxs/stable_diffusion_v1_4_bs2_pytorch_fp16False.onnx
 # set a exist path of engine file like resnet50.trt/resnet50.plan/resnet50.engine
 exist_compiler_path: null
+
diff --git a/inference/configs/stable_diffusion_v1_4/vendor_config/kunlunxin_configurations.yaml b/inference/configs/stable_diffusion_v1_4/vendor_config/kunlunxin_configurations.yaml
@@ -0,0 +1,3 @@
+fp16: false
+compiler: xtcl
+no_validation: true
diff --git a/inference/inference_engine/kunlunxin/xtcl.py b/inference/inference_engine/kunlunxin/xtcl.py
@@ -23,7 +23,6 @@ def build_engine(self, config, onnx_path):
         for input in onnx_model.graph.input:
             input_shape = input.type.tensor_type.shape.dim
             input_shape = [a.dim_value for a in input_shape]
-            input_shape[0] = config.batch_size
             input_name = input.name #'inputs:0'
             self.input_names.append(input_name)
             shape_dict[input_name] = input_shape