Integrate RAM cache with model RAM management (#13173)

2026-03-27 18:34:16 -07:00
parent 3696c5bad6
commit b353a7c863
9 changed files with 61 additions and 43 deletions
@@ -141,3 +141,17 @@ def interpret_gathered_like(tensors, gathered):
    return dest_views

 aimdo_enabled = False
+
+extra_ram_release_callback = None
+RAM_CACHE_HEADROOM = 0
+
+def set_ram_cache_release_state(callback, headroom):
+    global extra_ram_release_callback
+    global RAM_CACHE_HEADROOM
+    extra_ram_release_callback = callback
+    RAM_CACHE_HEADROOM = max(0, int(headroom))
+
+def extra_ram_release(target):
+    if extra_ram_release_callback is None:
+        return 0
+    return extra_ram_release_callback(target)