G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.games.action (More info?)

On Fri, 2 Sep 2005 19:58:00 +0100, "Derek Baker"
<me@xyzderekbaker.eclipse.co.uk> wrote:

>http://www.extremetech.com/article2/0,1697,1855163,00.asp

AI chip is next to useless. The problem isn't with AI being
computationally expensive, it has to do with the fact that most AI players
are brain-dead.

The only case where it would be needed is if an engine is straining on a
large nember of AIs - and that usually indicates an inefficient game
design.

A dual-core CPU should have the same effect - have the AI plan the actions
while the frame is being rendered.
 

Andrew

Distinguished
Mar 31, 2004
2,439
0
19,780
Archived from groups: comp.sys.ibm.pc.games.action (More info?)

On Fri, 2 Sep 2005 19:58:00 +0100, "Derek Baker"
<me@xyzderekbaker.eclipse.co.uk> wrote:

>http://www.extremetech.com/article2/0,1697,1855163,00.asp

Hopefully they can upgrade the SteamKiller Usenet spambot with it, the
current one doesn't demonstrate any degree of intelligence, artificial
or not ;-)
--
Andrew, contact via interpleb.blogspot.com
Help make Usenet a better place: English is read downwards,
please don't top post. Trim replies to quote only relevant text.
Check groups.google.com before asking an obvious question.
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.games.action (More info?)

Andrew <spamtrap@localhost.> once tried to test me with:

> On Fri, 2 Sep 2005 19:58:00 +0100, "Derek Baker"
><me@xyzderekbaker.eclipse.co.uk> wrote:
>
>>http://www.extremetech.com/article2/0,1697,1855163,00.asp
>
> Hopefully they can upgrade the SteamKiller Usenet spambot with it, the
> current one doesn't demonstrate any degree of intelligence, artificial
> or not ;-)

SteamKILLER is a new experiment in A.S. not A.I.

--

Knight37 - http://knightgames.blogspot.com

Once a Gamer, Always a Gamer.
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.games.action (More info?)

On 3 Sep 2005 00:53:23 GMT, Knight37 <knight37m@gmail.com> wrote:

>Andrew <spamtrap@localhost.> once tried to test me with:
>
>> On Fri, 2 Sep 2005 19:58:00 +0100, "Derek Baker"
>><me@xyzderekbaker.eclipse.co.uk> wrote:
>>
>>>http://www.extremetech.com/article2/0,1697,1855163,00.asp
>>
>> Hopefully they can upgrade the SteamKiller Usenet spambot with it, the
>> current one doesn't demonstrate any degree of intelligence, artificial
>> or not ;-)
>
>SteamKILLER is a new experiment in A.S. not A.I.

An apparently successful one at that.
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.games.action (More info?)

I'm somewhat optimistic about a physics chip for games- after all,
everybody knows roughly how a box should fall, so I think it would be easier
to agree on a standard implementation of physics. But this AI chip seems
less than sound. It only sounds like it would work for a first person
shooter, and even then, it sounds like it's setup for UT 2004 style bots.
There are alot of other types of AI in games that are less ammenable to this
approach. OTOH, a generalized CPU will be able to run just about any kind
of scripting or AI you could want.

The solution to AI is simply to just write better AI. Multi-core
processors should mean that there should be more than enough computing power
for this task.
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.games.action (More info?)

On Sat, 3 Sep 2005 03:43:49 -0400, "Magnulus" <Magnulus@bellsouth.net>
wrote:

> I'm somewhat optimistic about a physics chip for games- after all,
>everybody knows roughly how a box should fall, so I think it would be easier
>to agree on a standard implementation of physics. But this AI chip seems
>less than sound. It only sounds like it would work for a first person
>shooter, and even then, it sounds like it's setup for UT 2004 style bots.
>There are alot of other types of AI in games that are less ammenable to this
>approach. OTOH, a generalized CPU will be able to run just about any kind
>of scripting or AI you could want.
>
> The solution to AI is simply to just write better AI. Multi-core
>processors should mean that there should be more than enough computing power
>for this task.
>
>
Well, Magnalus, you finally have your physics wish granted.

See:-

http://ageia.vnewscenter.com/press.jsp?id=1125064384975

However, the $64 question will be:- Which games will support
the card ??

John Lewis
- Technology early-birds are flying guinea-pigs.
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.games.action (More info?)

"John Lewis" <john.dsl@verizon.net> wrote in message
news:431cc396.16547545@news.verizon.net...
> http://ageia.vnewscenter.com/press.jsp?id=1125064384975
>
> However, the $64 question will be:- Which games will support
> the card ??

The next Unreal Tournament game will support it, for sure, and it's likely
any future Unreal-based games will as well. Several other companies have
said they will support it (after all, Ageia is also a software solution that
can run on a CPU). It looks like support might be comparable to Creative's
EAX.

It will remain to be seen if a PCI solution is really what customers
should buy, though. Obviously, PCI Express slots are not common on
motherboards (outside of the graphics slot and maybe a useless network card
slot- most graphics cards now days don't fit within one slot), so Ageia
doesn't have alot of choice- they have to get it to work with regular PCI.
The regular PCI bus may face bandwith issues, especially if the number of
voices in games are increased (the X-Fi supports over 120 hardware voices,
whereas the original Live only supported about 16, the Audigy about 64-
realisticly, 64 is probably more than enough, and most games don't use more
than 16-32 at most), and this is combined with a physics card on the same
bus, though Ageia claims that the PCI bus will be adequate. Even a
soundcard and a hard drive, and a lousy chipset driver, can create problems
on a PCI bus, even with otherwise good hardware.