[ENet-discuss] Channel Incoming Unreliable Commands List Management Question

Alexander Dolgansky alexd221 at gmail.com
Thu Aug 2 06:47:55 PDT 2012


I am O.K. with 36% but I am seeing 0%. Correct me if I am wrong but
even a 200 fragment message should arrive with probability of 13%
assuming 1% packet loss. Obviously this is a very low probability but
still not 0%.

I am trying to reduce data size as much as possible to increase my
chances of getting the data across, but I would like to understand if
I've managed to reach the limit of what unreliable message delivery
can accomplish given my network setup or there is still a bug
somewhere (could be on my side as well) that prevents the most optimal
throughput.

Alexander.

On Thu, Aug 2, 2012 at 4:17 AM, Ruud van Gaal <ruud at racer.nl> wrote:
> Even on a 1Gb LAN, I've experience severe packet drops. However, that was
> with ~30 PC's talking to eachother at around 100Hz on a few channels (so
> quite a lot of data flying around).
> I'd say a method where you get packets that are split into 200 fragments
> might need a bit of redesigning, to send out smaller chunks of data (and
> more chance of them all arriving).
>
> Cheers,
> Ruud
>
> On Thu, Aug 2, 2012 at 2:53 AM, Chris Jurney <jurney at gmail.com> wrote:
>>
>> If you send 100 packets with 99% reliability, there is a 36% chance of
>> getting them all there.
>>
>>
>> On Wed, Aug 1, 2012 at 5:48 PM, Alexander Dolgansky <alexd221 at gmail.com>
>> wrote:
>>>
>>> That's a fair point but what bothers me about this somewhat is the
>>> fact that I was able to get these large messages before (perhaps they
>>> weren't 200 fragments long but definitely twice as large as the kind
>>> of messages that I can send now). Perhaps packet loss is more severe
>>> now but the last time I checked it was less than a percent (I don't
>>> know how accurate that measure was thought).
>>>
>>> Anyways, I'll experiment some more with what I can send unreliably
>>> with the updated ENet given my setup but as I mentioned before, the
>>> update did improve things so I am very happy about that.
>>>
>>> Alexander.
>>> _______________________________________________
>>> 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
>


More information about the ENet-discuss mailing list