[ENet-discuss] ENet bandwidth problems

[Syed Setia Pernama]

Pablo,
You said it is now faster without fragmentation, so how fast is it now?


________________________________
 From: Pablo de Heras Ciechomski <pablo.deheras at gmail.com>
To: Discussion of the ENet library <enet-discuss at cubik.org> 
Sent: Wednesday, August 29, 2012 5:47 AM
Subject: Re: [ENet-discuss] ENet bandwidth problems
 

Thanks Lee and thanks Dennis! It was fragmentation and now the
sleep command makes a difference again. Fixed.

Pablo
 
On Tue, Aug 28, 2012 at 4:42 PM, Lee Salzman <lsalzman at gmail.com> wrote:

You could try the ENET_PACKET_UNRELIABLE_FRAGMENT option if you really need fragmentation but your packet sizes are still only a handful of multiples of MTU.
>
>I would make the packets smaller than 1400 if you want to avoid all fragmentation, though, 1300ish would be safer since it leaves room for headers. 
>
>
>On 08/29/2012 12:01 AM, Pablo de Heras Ciechomski wrote:
>
>Kind of solved,
>>Made the packets into less than 1400 bytes and now it's faster. I guess it
>>has to do with fragmentation.
>>Pablo
>>
>>
>>On Tue, Aug 28, 2012 at 10:29 PM, Pablo de Heras Ciechomski <pablo.deheras at gmail.com <mailto:pablo.deheras at gmail.com>> wrote:
>>
>>    Hello,
>>
>>    I am playing around with unreliable unordered packets in ENet and I am
>>    trying to figure out why I am not transferring fast enough. My packets are
>>    all around 1800 bytes and are sent continuously. I seem to get the same
>>    speed on most any machine independent of wifi/Gigabit LAN/local intra-
>>    process communication. This is disturbing to me as I don't understand
>>    why. I am using the same loop as in the tutorial and I am getting no more
>>    than 250kB/s transfer rates, when it should be around 10MB/s in the best
>>    scenario. Is it due to all the packet_create calls? Is it due to some internal
>>    throttling? Packets don't seem to be lost so I am at loss :-)
>>
>>    I changed the timer to 0 ms wait on the host loop function if that makes
>>    any difference, but it doesn't seem so. Adding or removing a 5m Sleep
>>    (windows function so working on the whole process) doesn't seem to make
>>    any change other than making the whole system unresponsive if removed.
>>    Pablo
>>
>>
>>
>>_______________________________________________
>>ENet-discuss mailing list
>>ENet-discuss at cubik.org
>>http://lists.cubik.org/mailman/listinfo/enet-discuss
>>
>_______________________________________________
>ENet-discuss mailing list
>ENet-discuss at cubik.org
>http://lists.cubik.org/mailman/listinfo/enet-discuss
>
_______________________________________________
ENet-discuss mailing list
ENet-discuss at cubik.org
http://lists.cubik.org/mailman/listinfo/enet-discuss
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.cubik.org/pipermail/enet-discuss/attachments/20120828/e8dbd430/attachment-0001.html>