diff --git a/extensions-builtin/multidiffusion-upscaler-for-automatic1111 b/extensions-builtin/multidiffusion-upscaler-for-automatic1111
index 51cb83ce2..70b3c5ea3 160000
--- a/extensions-builtin/multidiffusion-upscaler-for-automatic1111
+++ b/extensions-builtin/multidiffusion-upscaler-for-automatic1111
@@ -1 +1 @@
-Subproject commit 51cb83ce2a53bf147a9091ed5269dcce8f1662d7
+Subproject commit 70b3c5ea3c9f684d04e7ff59167565974415735c
diff --git a/modules/devices.py b/modules/devices.py
index 843ed4dd8..b49745bd3 100644
--- a/modules/devices.py
+++ b/modules/devices.py
@@ -81,6 +81,8 @@ def torch_gc(force=False):
 
 
 def test_fp16():
+    if shared.cmd_opts.experimental:
+        return True
     try:
         x = torch.tensor([[1.5,.0,.0,.0]]).to(device).half()
         layerNorm = torch.nn.LayerNorm(4, eps=0.00001, elementwise_affine=True, dtype=torch.float16, device=device)
@@ -114,7 +116,7 @@ def set_cuda_params():
                 pass
     global dtype, dtype_vae, dtype_unet, unet_needs_upcast # pylint: disable=global-statement
     ok = test_fp16()
-    if shared.cmd_opts.use_directml: # TODO DirectML does not have full autocast capabilities
+    if shared.cmd_opts.use_directml and not shared.cmd_opts.experimental: # TODO DirectML does not have full autocast capabilities
         shared.opts.no_half = True
         shared.opts.no_half_vae = True
     if ok and shared.opts.cuda_dtype == 'FP32':