Was the problem with Project Cars ever detailed?

[3DVagabond]

Why is this game talked about so much on PC then? It's got 30k sales.... 30k? So why is everyone in a rush to talk about a game so few of us were willing to buy?

I may have had my gripes about AC Unity but at least people bought it.

Actually, this is what gets me with a lot of the PC community. It seems PC community gets caught up on these smaller titles and it's like wait a minute... barely ANY of us have this game. Why is this such a large part of reviews when this game is a game none of us play.

Games like Project Cars should get reviewed, maybe even follow up reviews on performance. But including them in GPU reviews is just senseless.

I find it more concerning that a game with so few sales even gets used to review cards. They certainly can't use the "popular" reasoning.

 

[Flapdrol1337]

Arma 3 has consistently been in the top 20 for YEARS yet so few review sites use it as a benchmark. Very popular game, crushes GPUs/CPU, good game too.

Bit tech used to bench arma games with cpu reviews.

For gpu reviews I don't think arma is that useful. Much more likely to be cpu bound.

I find it more concerning that a game with so few sales even gets used to review cards. They certainly can't use the "popular" reasoning.

It looks pretty, is relatively new and quite demanding.

Some sites test with world of warcraft, sure it's popular, but old cards get 100+ fps so what's the point.

 

[SniperWulf]

Alright, so I haven't read all the posts in this thread, but I will tell you what I have found with Project Cars on my own rigs.

Contrary to what Slightly Mad has stated about the game not being a gameworks sponsored title, I believe it's using an awful lot more of Nvidia tech than just PhysX. An that tech is actually offloading computations to the GPU.

The reason why I say this is that my Steambox and laptop both have nv gpus. The Steambox is basically a bunch of old hardware that I've purchased used from family and friends. I has an old GTX 680 for rendering and a GTX 660 dedicated to physx (and cuda calculations). I typically use MSI Afterburner's OSD to keep an eye on stats when playing. What I've noticed is that although the physx OSD indicator states that the calculations are being done on the CPU (and the CPU utilization shows it), the GTX 660 gets spooled up to its load operating frequencies. The same happens on the laptop (it has 680M SLI). So something is getting done on the GPU. Whether they will admit to it (or even know what it is if nv's engineers coded it in) is another story.

I haven't seen any other game that uses CPU based physx alone do this. Only gameworks titles.

My main rig is AMD based. Needless to say the game runs, but not great.

Without the proper tools, source code, or a Slightly Mad dev coming clean, it's almost impossible to tell for sure. If someone knows of a way to see what workloads in a game are going where, feel free to chime in.

<Hypothesis>
Slightly Mad took the cheap/quick way out and and used nvidia's half assed tech and engineering time instead of taking the time and coding a proper physics engine (and everything related) themselves. Couple that with the heavy Tessellation on the cars (which I'm pretty sure nv's engineers cranked) and you get the debacle that is Project CARS.
</Hypothesis>

 

[3DVagabond]

They don't need to offload anything to the GPU to cause issues with AMD. They can just code their drivers to distribute the PhysX to the CPU "more optimally" than AMD drivers do.

As an example (just an example not making this as a factual claim to what they are doing) if nVidia required a particular call from the driver or else the PhysX and Rendering all occurred on a single core then it would tank performance in any device that simply didn't use the correct call.

I would hate to think nVidia would be blatant enough to claim no PhysX on GPU when in actuality, there is. I suppose they could say it was miscommunication, but you can't continually use that excuse.