diff --git a/TODO.md b/TODO.md
index 7349fcc..22ef037 100644
--- a/TODO.md
+++ b/TODO.md
@@ -14,9 +14,13 @@
     * Run multiple threads to test performance when contested?
     * Iterate over whole returned buffer and run e.g. xor on it (.iter().fold())
     * Generic bench function since current functions are almost identical
+    * State is now accumulated over runs. Is this good?
 * Performance
     * cpu-target=native good?
     * mialloc good?
+    * TorrentMap: mutable access only to insert peer, then drop reference
+      and get read-only access to gather peers. This could speed up
+      multi-threaded performance a lot
     * Use less bytes from PeerId for hashing? Would need to implement
       "faulty" PartialEq too (on PeerMapKey, which would be OK)
 * bittorrent_udp
@@ -41,4 +45,4 @@
 * Use `bytes` crate for bittorrent_udp: seems to worsen performance somewhat
 * Zerocopy (https://docs.rs/zerocopy/0.3.0/zerocopy/index.html) for requests
   and responses? Doesn't work on Vec etc
-* New array buffer each time in response_to_bytes: doesn't help performance
\ No newline at end of file
+* New array buffer each time in response_to_bytes: doesn't help performance