Improve the performance and suitable for NPU computing (#9642)

* Improve the performance and suitable for NPU

* Improve the performance and suitable for NPU computing

* Improve the performance and suitable for NPU

* Improve the performance and suitable for NPU

* Improve the performance and suitable for NPU

* Improve the performance and suitable for NPU

---------
Co-authored-by: 蒋硕 <jiangshuo9@h-partners.com>
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>

examples/text_to_image/train_text_to_image_sdxl.py

@@ -59,6 +59,8 @@ check_min_version("0.31.0.dev0")
 
 logger = get_logger(__name__)
 if is_torch_npu_available():
+    import torch_npu
+
     torch.npu.config.allow_internal_format = False
 
 DATASET_NAME_MAPPING = {
@@ -540,6 +542,9 @@ def compute_vae_encodings(batch, vae):
     with torch.no_grad():
         model_input = vae.encode(pixel_values).latent_dist.sample()
     model_input = model_input * vae.config.scaling_factor
+
+    # There might have slightly performance improvement
+    # by changing model_input.cpu() to accelerator.gather(model_input)
     return {"model_input": model_input.cpu()}
 
 
@@ -935,6 +940,9 @@ def main(args):
     del compute_vae_encodings_fn, compute_embeddings_fn, text_encoder_one, text_encoder_two
     del text_encoders, tokenizers, vae
     gc.collect()
+    if is_torch_npu_available():
+        torch_npu.npu.empty_cache()
+    elif torch.cuda.is_available():
         torch.cuda.empty_cache()
 
     def collate_fn(examples):
@@ -1091,8 +1099,7 @@ def main(args):
                     # Adapted from pipeline.StableDiffusionXLPipeline._get_add_time_ids
                     target_size = (args.resolution, args.resolution)
                     add_time_ids = list(original_size + crops_coords_top_left + target_size)
-                    add_time_ids = torch.tensor([add_time_ids])
-                    add_time_ids = add_time_ids.to(accelerator.device, dtype=weight_dtype)
+                    add_time_ids = torch.tensor([add_time_ids], device=accelerator.device, dtype=weight_dtype)
                     return add_time_ids
 
                 add_time_ids = torch.cat(
@@ -1261,6 +1268,9 @@ def main(args):
                         )
 
                 del pipeline
+                if is_torch_npu_available():
+                    torch_npu.npu.empty_cache()
+                elif torch.cuda.is_available():
                     torch.cuda.empty_cache()
 
                 if args.use_ema: