[Feature] Support mamba radix cache v0 (#11214)

Co-authored-by: hanming-lu <hanming@x.ai>
Co-authored-by: hzh0425 <hzh0425@apache.org>
Co-authored-by: thalahors <ericalcaide1@gmail.com>

python/sglang/srt/server_args.py

@@ -362,6 +362,7 @@ class ServerArgs:
     # Mamba cache
     max_mamba_cache_size: Optional[int] = None
     mamba_ssm_dtype: str = "float32"
+    mamba_full_memory_ratio: float = 0.2
 
     # Hierarchical cache
     enable_hierarchical_cache: bool = False
@@ -2433,6 +2434,12 @@ class ServerArgs:
             choices=["float32", "bfloat16"],
             help="The data type of the SSM states in mamba cache.",
         )
+        parser.add_argument(
+            "--mamba-full-memory-ratio",
+            type=float,
+            default=ServerArgs.mamba_full_memory_ratio,
+            help="The ratio of mamba state memory to full kv cache memory.",
+        )
         # Args for multi-item-scoring
         parser.add_argument(
             "--multi-item-scoring-delimiter",