Untested: total non-garbage-collected streams across all devices

CHANGES.md

@@ -3,7 +3,7 @@
 ### Added
 
 - Debug number of total and unreleased events in a stream.
-
+- Debug the total number of non-garbage-collected streams across all devices.
 
 ### Fixed
 

cudajit.ml

@@ -1716,13 +1716,18 @@ module Stream = struct
     | false -> Unsigned.UInt.of_int64 cu_stream_default
     | true -> Unsigned.UInt.of_int64 cu_stream_non_blocking
 
+  let total_live_streams = Atomic.make 0
+  let get_total_live_streams () = Atomic.get total_live_streams
+
   let destroy stream =
     release_stream stream;
+    Atomic.decr total_live_streams;
     check "cu_stream_destroy" @@ Cuda.cu_stream_destroy stream.stream
 
   let create ?(non_blocking = false) ?(lower_priority = 0) () =
     let open Ctypes in
     let stream = allocate_n cu_stream ~count:1 in
+    Atomic.incr total_live_streams;
     check "cu_stream_create_with_priority"
     @@ Cuda.cu_stream_create_with_priority stream
          (uint_of_cu_stream_flags ~non_blocking)

cudajit.mli

@@ -777,6 +777,9 @@ module Stream : sig
   val total_unreleased_unfinished_delimited_events : t -> int * int * int
   (** Debug information about delimited events carried by the stream: total, unreleased (i.e. not
       destroyed), unfinished. *)
+
+  val get_total_live_streams : unit -> int
+  (** The total non-garbage-collected streams across all devices. *)
 end
 
 (** CUDA events can be used for synchronization between streams without blocking the CPU, and to