2023-08-12 10:03:23
Ray-tracing, Resizable BAR and Direct Storage. Behind these three names of technologies hides a huge revolution in PC video games that risks upsetting current configurations.
The word “revolution” is used to being thrown around in the middle of high tech, since the marketing teams of the big brands took possession of it. However, we are truly in front of a revolution in the middle of the video game. A silent revolution, because it primarily impacts the development of future titles.
A revolution that might have a significant impact. It is therefore more than important that consumers and PC gamers become aware of this wave which is coming soon. Otherwise, their gamer config might get old in a snap. To better explain it, we were able to interview John Benard, known as Linuxydable, a graphics engineer specialized in game engines, which is notably a contributor to the free game engine Godot Engine and works in collaboration with Microsoft and SEGA.
The arrival of the revolution
Before even talking regarding three new technologies, let’s go back to the current situation of video games. While publishers specializing in PC video games have gradually disappeared over the years in the AAA segment, consoles have taken an increasingly important place in development. If they have always been the main development platforms, for the simple fact that they are the most popular and therefore represent the greatest return on investment, they have also become the source of innovations in the sector.
And that, we might observe from the first presentation of the PlayStation 5 by Mark Cerny, the architect of the Sony console. During this presentation, the engineer was able to highlight many differentiating points of his machine, but three in particular caught our attention. First, the integration of ray tracing, which confirmed for many the future of the new technology. Then, the 16 GB memory in GDDR6 shared between the processor and the graphics card, a massive pool.
But finally, and above all, the famous SSD storage space passing through a PCIe 4.0 bus, which might be freely called by the CPU like the GPU. And tied to that, an awesome demo where Marvel’s Spider-Man load times have virtually disappeared. But that’s far from the only technical impact that the direct connection between the SSD and the CPU/GPU pair has brought; we will come back to it.
This architecture is not exclusive to Sony. The Xbox Series X philosophically shares the same approach. But what is important to observe is that it has not yet been really used during the first three years of life of the “new generation”. As tradition dictates, the first titles released were cross-gen: they were released on PS5/Xbox Series X as well as on PS4 and Xbox One. After these first three years, we can probably expect that games fully using this architecture will start to point their noses.
3 technologies to remember for PC gamers
To be able to follow these innovations brought by the new generation of consoles, the PC platform had to adapt and develop its own technologies. And while one of them is shared with consoles, the other two are important to remember.
Le ray tracing
We have already explained to you extensively how the ray tracing works and what impact it has on the rendering of video games. So here we will focus on the impact it brings to the development of video games with the help of Linuxydable, which characterizes the technology of “ huge innovation with a lot of potential on open world and lane games. »
It must be understood that today, development is influenced by decades of rendering techniques that have now become very cumbersome to manage. This is called rasterization, and it’s a trick to try to copy reality in a very artificial way, requiring a lot of work from the developers. In this context, ray tracing must be seen as a reset and a simplification of the way of designing video games.
In green: the lines of light seen by the camera. In red: those excluded.
Previously, to display an object, you had to assign it several different layers. You had the object itself, then a layer to modify its shape on the fly, then a layer to establish its lights, then another to establish its shadows, then its reflections, and so on . With ray tracing comes a new form of object: the mesh shader. The latter combines the previous techniques into one, making it possible to simplify the creation of the 3D object, and therefore free up the resources of the machine.
This is where ray tracing comes into its own. Combined with the mesh shader, the new technology eases the rendering process once once more by drastically reducing the number of calls to the graphics card that need to be executed. ” Previously, the more shadows had to be rendered on a shadow map, the more it had to be recalculated. The interest of ray tracing is to be able to calculate all the lighting of a world without overloading the process. Linuxydable tells us. Basically, rather than multiple small calls to all resources, ray tracing really only has to do its math once. ” We just have to send the 3D models, the mesh shader, the post-processing, the FSR, and it’s over!»
Le Resizable BAR (ou Smart Access Memory)
AMD called it Smart Access Memory, the others call it Resizable BAR, but the technology remains the same. Previously, to access the VRAM of a graphics card, the processor was limited to simple segments of 256 MB. This creates real bottlenecks, which we absolutely do not want. It’s not hard to imagine what a limitation this is in the era of 4K, or even 8K, for video games that you want to run at ever higher frame rates.
The Resizable BAR finally puts an end to this limit and allows processors to access the entire memory range of a graphics card in a single call. Again, the main purpose of this technology is to reduce the weight of instructions on the processor, to free up resources. On our modern gaming PCs and consoles, the graphics card really outperforms the CPU on certain types of calculations. This old behavior might force the graphics card to wait for the processor to finish processing. This change also changes the way VRAM is used, since it can now be loaded and unloaded on the fly: since it is a direct pipeline, we can send the 3D object live without having to store it in advance. »
Note that graphics cards are compatible with this new technology only from generation 30 from NVIDIA and the 6000 family from AMD. Intel graphics cards almost need this technology to work.
Direct Storage
Windows has also updated to enable a new API for developers: DirectStorage. This is once once more central to lighten the load on the PC processor. The latter was until then the conductor of the transfer of data from the storage space (hard disk) to the graphics card.
DirectStorage allows the graphics card to go directly to the SSD of the machine, without having to have the validation of the processor to do so. This technology will simply put an end to VRAM size limits: “ since the SSD is literally the VRAM, we allocate and deallocate on the fly. With increasing texture compression in games, it was important to allow the graphics card to use its computing power to decompress data on the fly, rather than the CPU.
The real impact of these technologies
Let’s remember the facts. We are now at three years of next generation consoles, so developers will start to exploit the new console architecture. To be able to adapt to it, PCs have finally deployed and/or refined the technologies that will be able to adapt to this new era. And the game engines of this new generation, such as Unreal Engine 5 or Unity version 2022, are officially deployed and are improving day by day.
The three technologies mentioned above will be important to follow this paradigm shift, where the “old way” will be gradually replaced to counter the current problem of PC video games: the fact that the processor creates a bottleneck. We want to call textures from storage to send them to the map? The processor slows down. We want to load these textures on the graphics card? The processor slows down. We want to create the final image for the player? The processor slows down.
Direct Storage gives control back to the graphics card. The Resizable BAR ensures that the latter correctly uses all the VRAM. And as for ray tracing… It changes the very foundations of our games, so that the data is less heavy to load, and the graphics card is finally called upon to do more of the work. The finesse of future developments will be found in the ability of developers to properly balance the load between CPU and GPU, according to the capabilities of each. At least, if we leave the field open to developers.
The problem of the PC port
Because there is one last element of the puzzle to consider: PC ports. Since specialized PC publishers have become rare, most ports are developed by external players like Linuxydable, often independent or belonging to small structures. And to successfully convert a PS5 title on Windows, the task is far from being as simple as when converting a .DOC to .PDF: “ sometimes you really have to call for a complete redevelopment of the engine. »
Yes, porting a game to PC can be considered as redeveloping the game. This is explained by several facts. Internal teams sometimes use their own technologies, and do not particularly want to share them with external people. Console development kits, useful for understanding how a game works or the libraries it uses, are not necessarily shared. And above all, you have to succeed in converting certain functions to completely different tools, such as the free Vulkan library or DirectX 12, to name a few. And this when those responsible for the port have access to the source code, which is not necessarily won either according to Linuxydable.
« The main cases of a bad port are often the lack of budget and deadlines that are too close.. » Lack of time, lack of money, lack of information sharing; for publishers, the PC versions are very often the weak links in the chain, which are only taken care of remotely and only watched with one eye. But the problem goes even further, since the actors with the knowledge to carry out effective portages are becoming increasingly rare: “ There is simply a lack of engineers. Apprenticeship, the school curriculum, do not exist. […] Many are self-taught. »
The thorny situation of PC gamers
And that’s all that might cause your PC gamers to become obsolete very quickly if you are not aware of the small revolution underway in this segment. While the PS5 and Xbox Series X have the stranglehold on innovation and are the leading development platforms, PCs are only just beginning to catch up. But for that, they must necessarily go through an almost complete architectural update of their configuration, where the ray tracing is minimally optimized, the graphics card is Resizable BAR compatible, and the storage space is fast enough to be used as a data source by the GPU. As a bonus, we can’t particularly count on any improvement in PC ports, an area of development that doesn’t have access to the resources, time and budget to do a good job. Not to mention the lack of engineers specialized in this particular field.
This same domain which will soon receive console versions optimized for this new architecture boosted to PCIe 4.0 and ray tracing, and who will have to succeed in the herculean task of porting this new rendering to older platforms. You only have to see the state of the latest PC versions released to date to determine it: the job will not be properly done. In which case, PCs already compatible with the ray tracingthe Resizable BAR and having ultra-fast storage will necessarily be favored for the next video games to be released on the platform.
The unanswered question is as simple as it can be devastating: is this the case with your current config?
Want to join a community of enthusiasts? Our Discord welcomes you, it is a place of mutual aid and passion around tech.
1691838132
#technologies #gaming #PCs #obsolete
