[ut3] Ryan: Should we look into PhysX cards for our linux servers?

[Apoc]Death death at apoc.org
Wed Oct 17 08:04:23 EDT 2007


One more update, then I'll stop spamming about this:

I did a restart on my server (ApocTestbed1) around midnight eastern,  
and I was able to join it again 7 hours later when I woke up (score  
was 44-8 at the time, so it had been running), so all looks good, or  
as good as can be expected! (unless I join it again...  )

-Rob

On Oct 16, 2007, at 11:43 PM, [Apoc]Death wrote:

> As long as no one leaves the server and then returns, this seems a  
> fine workaround...  I am still hanging the thing in the usual  
> manner when I exit the server and then rejoin it.  If you think  
> about it, this is precisely what a mapchange does, so perhaps this  
> makes sense?  Who knows...  Hope it helps with the debugging.
>
> -Rob
>
> On Oct 16, 2007, at 10:47 PM, [Apoc]Death wrote:
>
>> Just came on a short term workaround for this...  The issues  
>> happen just after a map switch, so I tried to stop them from  
>> happening: I set GoalScore=0 and TimeLimit=1440 (minutes in 1  
>> day).   We've been up for two hours straight this way!  Beats  
>> locking up every 20 minutes.  I expect I'll reset the server once  
>> a day and should be golden.
>>
>> -Rob
>>
>> On Oct 16, 2007, at 9:50 PM, raXimous wrote:
>>
>>> Ryan, any idea when we'll be getting a fix for the timeouts that  
>>> the server seems to be having? alot of people have experianced  
>>> it. just stops accepting connections.
>>>
>>> On 10/17/07, Ryan C. Gordon <icculus at icculus.org> wrote:
>>>
>>> > This question is mostly to Ryan since he's doing the port.  
>>> Should we be
>>> > looking into picking up PhysX cards for our linux servers? Will  
>>> the game
>>> > use them if they're there, and will there be linux support for  
>>> them?
>>> > Will we see perfomance improvements by having them?
>>>
>>> Ageia doesn't produce kernel drivers for Linux, so they wouldn't be
>>> useful in a server.
>>>
>>> I have told them before that it would be nice for just this  
>>> reason, though.
>>>
>>> --ryan.
>>>
>>>
>>> ---
>>> To unsubscribe, send a blank email to ut3-unsubscribe at icculus.org
>>> Mailing list archives: http://icculus.org/cgi-bin/ezmlm/ezmlm-cgi?64
>>>
>>>
>>>
>>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://icculus.org/pipermail/ut3/attachments/20071017/c40e67fe/attachment.htm>


More information about the ut3 mailing list