From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: tcondeixa@veniamworks.com Received: from krantz.zx2c4.com (localhost [127.0.0.1]) by krantz.zx2c4.com (ZX2C4 Mail Server) with ESMTP id a98ec166 for ; Mon, 26 Jun 2017 12:14:14 +0000 (UTC) Received: from mail-wm0-f51.google.com (mail-wm0-f51.google.com [74.125.82.51]) by krantz.zx2c4.com (ZX2C4 Mail Server) with ESMTP id c3f38263 for ; Mon, 26 Jun 2017 12:14:14 +0000 (UTC) Received: by mail-wm0-f51.google.com with SMTP id w126so5912385wme.0 for ; Mon, 26 Jun 2017 05:30:37 -0700 (PDT) Return-Path: From: Tiago Condeixa Content-Type: text/plain; charset=utf-8 Mime-Version: 1.0 (Mac OS X Mail 10.3 \(3273\)) Subject: Wireguard performance Message-Id: Date: Mon, 26 Jun 2017 13:30:35 +0100 To: "Jason A. Donenfeld" , WireGuard mailing list List-Id: Development discussion of WireGuard List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Hi, I have been doing more experiments with heterogenous technologies and = connections and there is a scenario where I believe that something = strange is happening with wireguard. When my board is using a 100 Mbps ethernet connection in a local network = the iperf3 TCP is able to reach 95Mbps with the direct IP route, but = when using wireguard the throughput reduces to 65/70 Mbps.=20 It=E2=80=99s not a bad result at all, but I noticed that with wireguard = the iperf3 CPU consumption goes from 10% to 70% when comparing with = direct IP route. Do you have any idea why is this happening? ...the = overall system CPU can slightly increase but I do not expect that happen = with iperf3. In other scenarios with other technologies and networks wireguard = usually reduce 5% of the throughput but in this case it goes beyond 25%. Thanks a lot, Tiago=