tinc - Apr 2013 - Tinc power consuption

hello tincers,

when I on battery on my notebook, running powertop, I get
nic:<tinc-net-name> as the most power consuption resource in my notebook
just after the screen.

This happens no matter if I have traffic or not.

I am not really sure that if this is really a tinc fault or a tap/tun
implementation that never sleeps.

any advice?
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://www.tinc-vpn.org/pipermail/tinc/attachments/20130423/f086f22b/attachment.html>

On Tue, Apr 23, 2013 at 11:58:23AM -0300, Raul Dias wrote:
> when I on battery on my notebook, running powertop, I get
> nic:<tinc-net-name> as the most power consuption resource in my
notebook
> just after the screen.
> 
> This happens no matter if I have traffic or not.
> 
> I am not really sure that if this is really a tinc fault or a tap/tun
> implementation that never sleeps.
> 
> any advice?
Can you copy&paste the output of powertop? It's not always clear what is
being
reported. I'm using PowerTop v2.0, and nowhere does it actually show the
power
consumption in Watts for individual resources. In the "Device stats"
panel, it
shows relative usage for individual resources, but that is either a percentage
(of what is not clear, the sum of all percentages is much more than 100% on my
laptop anyway), or a number of events per second (for example, ops/s for the
GPU, pkts/s for network interfaces). If there is no traffic on the VPN
interface, PowerTop on my laptop says 0 pkts/s. When connected to the VPN but
otherwise idle, the "Overview" panel shows that tincd has only between
0.3 and
1.0 wakeups/s.

-- 
Met vriendelijke groet / with kind regards,
     Guus Sliepen <guus at tinc-vpn.org>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
URL:
<http://www.tinc-vpn.org/pipermail/tinc/attachments/20130423/20bde138/attachment.pgp>

Hello all,

Guus can you tell me if tinc uses openssl and AES algorithm ? (it is
realted to power consumption)

if yes some computers have a special hardware chip to make AES in
hardware, and might be the cause of power consumption ?
http://en.wikipedia.org/wiki/AES_instruction_set

try cat /proc/cpuinfo | grep aes

if there is the aes flag then you are doing probably aes in hardware
in openssl is compiled correctly to do this.
when you use this chip you might expect a bigger power consumption ?

read this paper, they do something else but they describe the SSL AES
stuff in detail
http://staff.science.uva.nl/~delaat/rp/2012-2013/p08/report.pdf

moreover reading this paper could help to make tinc faster :)

ciao,

Saverio

2013/4/23 Raul Dias <raul at dias.com.br>:
> hello tincers,
>
> when I on battery on my notebook, running powertop, I get
> nic:<tinc-net-name> as the most power consuption resource in my
notebook
> just after the screen.
>
> This happens no matter if I have traffic or not.
>
> I am not really sure that if this is really a tinc fault or a tap/tun
> implementation that never sleeps.
>
> any advice?
>
>
> _______________________________________________
> tinc mailing list
> tinc at tinc-vpn.org
> http://www.tinc-vpn.org/cgi-bin/mailman/listinfo/tinc
>

On Wed, Apr 24, 2013 at 09:46:04AM +0200, Saverio Proto wrote:
> Guus can you tell me if tinc uses openssl and AES algorithm ? (it is
> realted to power consumption)
The manual can tell you that tinc 1.0.x indeed uses OpenSSL, but it uses
Blowfish by default. However, you can switch to AES using the Cipher option.

In tinc 1.1, if you enable the new protocol, AES is the default.
> if yes some computers have a special hardware chip to make AES in
> hardware, and might be the cause of power consumption ?
> http://en.wikipedia.org/wiki/AES_instruction_set
> 
> try cat /proc/cpuinfo | grep aes
> 
> if there is the aes flag then you are doing probably aes in hardware
> in openssl is compiled correctly to do this.
> when you use this chip you might expect a bigger power consumption ?
You would expect a lower power consumption if your CPU supports AES
instructions, because otherwise OpenSSL will use less efficient instructions.
> read this paper, they do something else but they describe the SSL AES
> stuff in detail
> http://staff.science.uva.nl/~delaat/rp/2012-2013/p08/report.pdf
> 
> moreover reading this paper could help to make tinc faster :)
I do not see so much in this paper that could be applied to tinc. Of course, if
you have hardware that OpenSSL can take advantage of, this will indeed speed
things up. And if tinc would use multiple threads that would help too (on
multi-core systems), but that is not so trivial to implement.

-- 
Met vriendelijke groet / with kind regards,
     Guus Sliepen <guus at tinc-vpn.org>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
URL:
<http://www.tinc-vpn.org/pipermail/tinc/attachments/20130424/ca016e9e/attachment.pgp>