Automobilista 2 V1.4.3.3 & Racin´ USA Pt3 Officially Released!

Setting aside who has got the definitions right, then obviously he is asking for a Pocono or Lausitzring type.

Me, I'm dreaming of Trenton and the suitable cars. I wonder if the Trenton oval is not copyrighted anymore either, as its been demolished long ago so not in use and therefore "free game"?

 

It worked so well it was removed from the game.

It was essentially the same scheme that rFactor and rF2 had from ISI where the AI would randomly try slightly different lines each corner/each lap and a system recorded the best outcomes. The AI would then use the best/better of the lines in subsequent usage (e.g., the next time you used that car/track combination).

Why was it removed? One reason is that the AI were not capable of distinguishing the line choice relative to their own tire temperatures/grip levels. The best lines and subsequent lap times would, as expected, get generated when tires were fully up to temp. But when the AI tried to follow some of those lines right off the mark, the results could be chaotic.

Now imagine that system with dynamic weather and track grip added into the equation... Perhaps why it has not worked out well in rF2?

 

The only game I remember that used multiple cores and could take full advantage of, say, an i7, was Battlefield 4. Probably there are more, but they are the minority.

Bear in mind that I'm not trying to bash Automobilista 2. This simply was a test to put out some evidence on a debate that came from a couple pages back regarding AI. My point was that sim games are not taking advantage of the CPU power available, concentrating on single core punch but leaving lots of cores unused, and it's perhaps a reason why we have not seen big advancements in, say, AI development. Or maybe the reason is lack of resources, which is why they cannot develop a truly multithreaded code. Never tried to imply that it's an easy thing to do.

 

And in Reiza's defence we as consumers have no idea what limitations on that front the Madness engine has...

 

900hp in 1978...

My goodness, I already was wanting 80's CART/IndyCar, but now I REALLY want more classic Indy content.

 

It seems unrealistic for developers to adapt and optimise their games for the ever changing number of CPU cores available. Why isn't this handled by the operating system or some sort of middleware?

(this is a genuine question from a layman, and not directed at you OP )

 

I don't think so. He asked for Nazareth, which was an actual tri-oval about 45 minutes away from Pocono.

Pocono and Lausitzring are actual triangles - they have three turns.

A tri-oval has five turns, with turn 5 usually not counted because you usually take it flat out. Turn 5 is usually in the middle of the front stretch and is usually where the Start/Finish line is. (Putting a "turn" on the front stretch creates passing opportunities, and having the front stretch have a 90 degree angle at it means you can set up better sightlines so fans can see the S/F line.)

I'd love to see Trenton in the game, but it has more of a reputation with early NASCAR, so you'd want it with some of those cars.

Also, Reiza doesn't need the real rights to a track in order to put it in the game. Azure Circuit is a thinly disguised Monaco, for example. If they wanted to, Reiza could add a kidney-bean oval to the game with the same lengths and banking as Trenton, but just call it a different name.

 

It's because adding different threads (which can run on different processors) adds a lot of overhead and that overhead slows the program down.

If you divide up the threads correctly, the performance gain will make up for the performance penalty from the overhead. But if you don't, your program will run worse than if it's single threaded.

Ultimately, identifying which tasks can be parallelized (and run on different cores at the same time) is one of the hard tasks that goes along with writing a racing game engine. The parallelizable tasks in AMS2 might be different than the parallelizable tasks in RF2 or ACC based on how the three different games model the world.

That doesn't even consider first person shooters, real time strategy, Madden, etc, all of which have different tasks that can be parallelized than in a racing game.

The reason you can't have middleware or the OS decide what to parallelize is because the OS team and the middleware team don't work on racing games and don't understand which tasks are parallelizable in a racing game.

 

sorry, that's right, Daytone is one. But so wide, it feels more like a long,long,long and hilly straight with no turns (as long as there is no g--force simulator)

By the way, Eurospeedway / Lausitzring, Triangle, would be a fantastic addition as well....
Lausitzring

 

Hmmm. Don't know any game that has the Eurospeedway Oval.
But the infield itself is very cool too!
So i would love to get both into AMS2

Just seen the Video..
Very bad thing that Alex Zanardi lost his both legs at this awful crash
But hey, he made his way and i remember a guest start in DTM a few years later, where he blew some "pros" in the rain.....without legs!!!

 

Lausitzring would be a lot of fun, many configurations as well...

 

Also worth adding that just having more compute resources available doesn't magically make the A.i. better either. It takes considerable development bandwidth to put those resources to good use. Not even referring to thread synchronization, referring to actually coding in more complex A.i. algorithms/decisions per tick.

 

consider that the big development of the Madness Engine was conceived in 2012, with the addition of the new SETA tire model, and at the time the flagship processors were the quad-core, but there were still many dual-core in the mass.
I remember that at the time SMS or rather Andrew Weber Physics load, argued that for the calculation of SETA two cores were needed for processing.

The first pCARS had to come out also for the then Nintendo Wii U console, and I remember that they had problems on the CPU side, for processing. Apart from this I recall that there were during development and even after release commands to also exploit logical coores in CPU processing, or turn off public to reduce the load on the CPU side.

 

also, and not only that, but just think that only in recent years have we had a significant number of cores for processing, the general average is around 6-8 cores, thanks to AMD which has upset the market by launching the Ryzen family in late 2017 with the first 8-core processors, otherwise we still went ahead with 4 Intel processing cores.

I'm not a programming expert but even if a graphics engine were rewritten from scratch today, I don't think it's so easy to combine such a large amount of data to be processed also for the AI without there being problems on the engine itself between physics and graphics.

The only ones who can enlighten us in this regard are the Reiza team

 

Pocono would be cool too.

 

This was a very hard moment. One of the badest I've ever seen in motorracing.
But when CART / ChampCar raced there in 2003, Zanardi finished the last laps which he missed in 2001 for (maybe) victory.
That was one of my most heart-touching moment. Zanardi raced so many more races in WTCC (and hand-bike), all with his personal issue. Really a great man

 

Zanardi is a legend in so many ways...

 

I guess it does not hurt as bad in GT7 with their SOPHY AI solution?

 

We’ll see when GT7 comes to PC