and yet you keep ignoring things like this:
http://www.slideshare.net/DICEStudio/directx-11-rendering-in-battlefield-3
https://twitter.com/repi
but you know honest representation of the situation isn't something your actually after right .
who cares if AMD DX11 isn't multi threaded when it doesn't actually buy you anything........
so NV supports DCL, so does frostbite and yet mantel still shows massive performance improvements above and beyond it....
Just wondering why you stick to the arbitrary number "300", surely, had it been directx optimized it could easily be 700% right? Or are you just making up numbers.
When DICE made those slides, neither GPU brand supported DCL, but you know honest representation of the situation isn't something your actually after right .
In any case, "Mantel" only works for AMD cards which do not support DCL so Frostbite has to use manual threading token+replay which is much slower than native multithreading.
I wouldn't call it flawed quite, but SLI performing worse than not, the way in which it gets limited and what happens to the performance and the heavy cost of a single effect do limit its uses as a benchmark.
We know that Mantle can use multiple render threads. We also know that DX 11 Star Swarm only uses 1 rendering thread. Do I need to keep spoonfeeding?
Oxide's engine, Nitrous biggest advantage is that it is completely asynchronous in its rendering. The engine is not tied to a thread so the more cores you have, the faster it gets.
The biggest advantage of Mantle is that unlike DirectX, Mantle is truly multicore aware. With DirectX 11, more cores don't buy you nearly as much. That said, Nitrous, even on DirectX, is still two orders of magnitude faster than say a typical DirectX 9 engine.
A Mantle optimized game can show a massive performance gain depending on how many cores the user has on their CPU. Contrary to what I read on some forums, most games remain CPU or video driver bound (i.e. the GPU is waiting to be fed). Mantle lets you get a lot more stuff onto the GPU.
Repi said yesterday that DCL is fundamentally broken in DX.
Yesterday.
Also, AC III was confirmed to use DCL, and likely AC IV does as well. NVidia has a large lead in that game, with a GTX 770 roughly equal to a 290x:
This test was made day one of AC:IV
Very good, no more favors Nvidia.
snip
This looks the same for borderlands 1/2, batman and any other TWIMTBP title. Some call it concrete tessellation performance advantage
From the technical standpoint, an engineer would not care even if the demo was random dots popping on the screen. All he cares is batch count and performance. Leave the 'looks' for the artists.I think the issue with throwing out that series of quotes is that it is a PR interview piece, talking about the theoretical advantages of an, at that time, unreleased game engine.
Yet here we are with an actual application, which when you look at it from numerous angles functions in the real world like it is incredibly multi-core unaware, does not handle thread scheduling well, and in fact creates massive object and draw call overhead for an effect that looks worse than traditional techniques.
The reality of the software and the ideals set forth in that dated interview don't gel.
It should also be noted that in the same period of time, the engine went from being the core of a new, developmental engine for RTS and RPG games to being a canned time demo benchmarking application to hype Mantle features.
I think that the point is that the synchronous nature of those events is a bit suspicious.
So whilst Repi might claim the technology is broken, other developers have been, or are making use of it.
When DICE made those slides, neither GPU brand supported DCL, but you know honest representation of the situation isn't something your actually after right .
o'rly.jpg answer point 1 then.In any case, "Mantel" only works for AMD cards which do not support DCL so Frostbite has to use manual threading token+replay which is much slower than native multithreading.
From the technical standpoint, an engineer would not care even if the demo was random dots popping on the screen. All he cares is batch count and performance. Leave the 'looks' for the artists.
it is not timed, and not canned. Try running it not in 'benchmark' mode
And while the software may be doing dynamic AI and interaction calculations on the fly, that shows what, exactly? As an early look at what a game might some day be able to do? Cool. But based on the timing and the amount of PR that surrounds this release, I have a hard time thinking that that is what most peope are either intending or taking away from this. This is designed, functionally, as a PR benchmark to show how amazing Mantle rendering is compared to DX 11. People trying to understand how and why that may be the case is a worthy exercise.
This looks the same for borderlands 1/2, batman and any other TWIMTBP title. Some call it concrete tessellation performance advantage
Nope, the fact that DCL is used in less games than PhysX is proof of its massive failure.
If Mantle is implemented in just 5 games this year it will get past DCL.
Just how broken is it to be ignored by developers left and right?
That's the point. That "some day" is now; with Mantle. There is no more waiting. We can have that once Mantle is out.
To add to that every object has it's own independant "physics, AI, pathfinding, and threat assessment". That is down to the individual turrets on the capital ships.
I'm sure Repi has forgotten more about programming than most (or any) of us here, but anyone that has a free thinking brain will find issues with that statement.
If DCL is fundamentally broken, why did it work in Civ 5? The last benchmark of Civ 5 on Anandtech was in the GTX 770 review, and you can see the GTX 780, the most CPU limited GPU, is nearly 50% ahead of the 7970 Ghz, despite being on average 20% faster:
Also, AC III was confirmed to use DCL, and likely AC IV does as well. NVidia has a large lead in that game, with a GTX 770 roughly equal to a 290x:
And then the biggest knock against Repi's statement, is that Project C.A.R.S has been confirmed to use DX11 multithreading. In fact, the entire renderer has been built around it. I think this game will finally give the technology some good press, because it's a really big next gen title.
So whilst Repi might claim the technology is broken, other developers have been, or are making use of it.
But we can't. The game doesn't really function, as a game. It does a fantastic job of throwing a huge number of objects up on the screen, but there is no benefit from those huge numbers of units outside of making the spectacle of it nicer. It doesn't appear that something is going on such that there is any benefit, or even practical manner, for you to individually control each individual fighter out of the thousands that might spawn per side. So all of those fighters and all of those calculations effectively end up working like a fighter wing in any other RTS game, except with more effective animation variety.
Again, maybe I am just being dense, and the fact that I have yet to use any of my four local systems that can get respectable frame rates on the thing. That may change this weekend when I get a chance to get back to my main gaming rig. At the moment though it seems to be designed to highlight a massive number of calculations that, depending on how the game is actually designed, could end up trivial in the long run, but at a huge hardware resource cost.