fea(): Updates the func inputs. Added G1 test

tests/test_diffusers.py

@@ -1256,11 +1256,16 @@ def test_stable_diffusion_xl_inference_script(self):
             # Ensure the run finished without any issue
             self.assertEqual(return_code, 0)
 
-    _sdxl_inferece_throughput_data = (("ddim", 0.301), ("euler_discrete", 0.301))
-
-    @parameterized.expand(_sdxl_inferece_throughput_data)
-    def test_stable_diffusion_xl_generation_throughput(self, scheduler, baseline):
-        def _sdxl_generation(self, scheduler: str, baseline: float):
+    if IS_GAUDI2:
+        _sdxl_inferece_throughput_data = (("ddim", 1, 10, 0.301), ("euler_discrete", 1, 10, 0.301))
+    else:
+        _sdxl_inferece_throughput_data = (("ddim", 1, 10, 0.074),)
+
+    @parameterized.expand(_sdxl_inferece_throughput_data, skip_on_empty=True)
+    def test_stable_diffusion_xl_generation_throughput(
+        self, scheduler: str, batch_size: int, num_images_per_prompt: int, baseline: float
+    ):
+        def _sdxl_generation(self, scheduler: str, batch_size: int, num_images_per_prompt: int, baseline: float):
             kwargs = {"timestep_spacing": "linspace"}
             if scheduler == "euler_discrete":
                 scheduler = GaudiEulerDiscreteScheduler.from_pretrained(
@@ -1281,18 +1286,18 @@ def _sdxl_generation(self, scheduler: str, baseline: float):
                 "stabilityai/stable-diffusion-xl-base-1.0",
                 **kwargs,
             )
-            num_images_per_prompt = 10
+            num_images_per_prompt = num_images_per_prompt
             res = {}
             outputs = pipeline(
                 prompt="Sailing ship painting by Van Gogh",
                 num_images_per_prompt=num_images_per_prompt,
-                batch_size=1,
+                batch_size=batch_size,
                 num_inference_steps=30,
                 **res,
             )
             self.assertGreaterEqual(outputs.throughput, 0.95 * baseline)
 
-        _sdxl_generation(self, scheduler, baseline)
+        _sdxl_generation(self, scheduler, batch_size, num_images_per_prompt, baseline)
 
 
 class GaudiStableDiffusion3PipelineTester(TestCase):