Unify server-side and model-side Config(Part-5) (#3497)

* move config

* fix xpu

* fix

* fix vl

* fix vl

* fix unitest

* fix args

* add unitest

* fix test

tests/utils/test_config.py

@@ -0,0 +1,81 @@
+import unittest
+
+from fastdeploy.config import (
+    CacheConfig,
+    FDConfig,
+    GraphOptimizationConfig,
+    ParallelConfig,
+)
+
+
+class TestConfig(unittest.TestCase):
+    def test_fdconfig_nnode(self):
+        parallel_config = ParallelConfig({"tensor_parallel_size": 16, "expert_parallel_size": 1})
+        graph_opt_config = GraphOptimizationConfig({})
+        cache_config = CacheConfig({})
+        fd_config = FDConfig(
+            parallel_config=parallel_config,
+            graph_opt_config=graph_opt_config,
+            cache_config=cache_config,
+            ips=["1.1.1.1", "0.0.0.0"],
+            test_mode=True,
+        )
+        assert fd_config.nnode == 2
+        assert fd_config.is_master is False
+
+    def test_fdconfig_ips(self):
+        parallel_config = ParallelConfig({})
+        graph_opt_config = GraphOptimizationConfig({})
+        cache_config = CacheConfig({})
+        fd_config = FDConfig(
+            parallel_config=parallel_config,
+            graph_opt_config=graph_opt_config,
+            cache_config=cache_config,
+            ips="0.0.0.0",
+            test_mode=True,
+        )
+        assert fd_config.master_ip == "0.0.0.0"
+
+    def test_fdconfig_max_num_tokens(self):
+        parallel_config = ParallelConfig({})
+        graph_opt_config = GraphOptimizationConfig({})
+        cache_config = CacheConfig({})
+        cache_config.enable_chunked_prefill = True
+        fd_config = FDConfig(
+            parallel_config=parallel_config,
+            graph_opt_config=graph_opt_config,
+            cache_config=cache_config,
+            ips="0.0.0.0",
+            test_mode=True,
+        )
+        assert fd_config.max_num_batched_tokens == 2048
+
+        cache_config.enable_chunked_prefill = False
+        fd_config = FDConfig(
+            parallel_config=parallel_config,
+            graph_opt_config=graph_opt_config,
+            cache_config=cache_config,
+            ips="0.0.0.0",
+            test_mode=True,
+        )
+        assert fd_config.max_num_batched_tokens == 8192
+
+    def test_fdconfig_init_cache(self):
+        parallel_config = ParallelConfig({})
+        graph_opt_config = GraphOptimizationConfig({})
+        cache_config = CacheConfig({})
+        cache_config.cache_transfer_protocol = "rdma,ipc"
+        cache_config.pd_comm_port = "2334"
+        fd_config = FDConfig(
+            parallel_config=parallel_config,
+            graph_opt_config=graph_opt_config,
+            cache_config=cache_config,
+            splitwise_role="prefill",
+            test_mode=True,
+        )
+        fd_config.init_cache_info()
+        assert fd_config.disaggregate_info["role"] == "prefill"
+
+
+if __name__ == "__main__":
+    unittest.main()