diff --git a/extensions-builtin/Lora/lora_patches.py b/extensions-builtin/Lora/lora_patches.py
index 7085d251b..12986e8f9 100644
--- a/extensions-builtin/Lora/lora_patches.py
+++ b/extensions-builtin/Lora/lora_patches.py
@@ -2,6 +2,7 @@ import torch
 import networks
 from modules import patches, shared
 
+
 class LoraPatches:
     def __init__(self):
         self.active = False
diff --git a/modules/devices.py b/modules/devices.py
index 1102b9a74..78a20c4e6 100644
--- a/modules/devices.py
+++ b/modules/devices.py
@@ -134,7 +134,7 @@ def torch_gc(force=False):
     if oom > previous_oom:
         previous_oom = oom
         log.warning(f'GPU out-of-memory error: {mem}')
-    if used > shared.opts.torch_gc_threshold:
+    if used >= shared.opts.torch_gc_threshold:
         log.info(f'GPU high memory utilization: {used}% {mem}')
         force = True
     if not force:
diff --git a/modules/shared.py b/modules/shared.py
index 259b04e34..04a03d9b4 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -300,7 +300,7 @@ options_templates.update(options_section(('cuda', "Compute Settings"), {
     "other_sep": OptionInfo("<h2>Execution precision</h2>", "", gr.HTML),
     "opt_channelslast": OptionInfo(False, "Use channels last as torch memory format "),
     "cudnn_benchmark": OptionInfo(False, "Enable full-depth cuDNN benchmark feature"),
-    "torch_gc_threshold": OptionInfo(90, "VRAM usage threshold before running Torch GC to clear up VRAM", gr.Slider, {"minimum": 0, "maximum": 100, "step": 1}),
+    "torch_gc_threshold": OptionInfo(80 if devices.backend == "ipex" else 90, "VRAM usage threshold before running Torch GC to clear up VRAM", gr.Slider, {"minimum": 0, "maximum": 100, "step": 1}),
 
     "cuda_compile_sep": OptionInfo("<h2>Model Compile</h2>", "", gr.HTML),
     "cuda_compile": OptionInfo(True if cmd_opts.use_openvino else False, "Compile UNet"),