Computex 2023 is making headlines at large, and every meeting that Jensen Huang, CEO of NVIDIA, has with different parties, leaves a buzz among the attendees. The last is the fear that NVIDIA will stop focusing on gamers to give top priority to AI, the center of the debate at the moment. The green leader wants calm down and asks for understanding with a key argument: for RTX GPUs, AI is the most important thing.
Don’t worry, NVIDIA is not going to leave the gaming sector aside, rather, it has found a way to carry two at the same time, which are now gaming and AI. Huang made this clear by commenting that “NVIDIA still cares about you.” And it is that if we already talked about the value of the company and its 939 billion, now, after Computex 2023, it has already exceeded one billion and is going for fourth place in the world.
NVIDIA still cares about games, they are a priority for RTX GPUs, just like AI
And they are with an argument that cannot be refuted and that Huang himself put on the table when asked if the players were right to feel like a jealous older brother of the newborn baby in reference to RTX vs. AI. Jensen pretended to have been insulted and took the question as a joke, after which he said the following:
“Number one, RTX was invented for gamers and for RTX technology the most important thing is AI. Without AI, we couldn’t do Ray Tracing in real time. It wasn’t even possible. And the first AI project in our company, the number one AI approach was Deep Learning Super Sampling (DLSS) using deep learning. That is the mainstay of RTX GPUs.
Number two, look at what we did with generative AI, the first application was ACE, Avatar Cloud Engine, you hurt my feelings so deeply…” Huang jokingly replied. The newly revealed ACE for Games is “a custom AI model foundry service that transforms games by bringing intelligence to non-playable characters (NPCs) through AI-powered natural language interactions.”
Is 8 GB enough for gaming in 2023?
If there is a question and an answer that you want to ask and be answered, this is it. We already explained NVIDIA’s reasons for increasing the cache and lowering the bus, keeping 8 GB of VRAM for the mid-range, where only the RTX 4060 Ti 16GB break that paradigm and it will scale a bit in performance (+-5% to 7% at most, according to our calculations at the same frequency). Based on this, the question was formulated to the CEO, and his response was as follows:
“Remember that the frame buffer is not the memory of a PC, it is a cache. And how you manage that cache is a big problem. It is like any other cache. And yes, the bigger the cache, the better. However, you are negotiating with so many things…”
The argument continued by comparing each architecture with an equation (balanced in this case) or Kung Fu, where you have to find the balance by weighing the cost, power and the number of units that have to be included. Finally, Huang insisted again that 8 GB is the right amount for a GPU like the RTX 4060 Ti.
Whether he is right or not, it is clear that NVIDIA is not going to set aside the gaming market in the face of AI, since they are complementary and since at some point Ray Tracing will be what rasterization is today, the company will look for a way not having to focus on one hybrid pipeline as at present. That is to say, we will have a new pipeline where Ray Tracing will be normal, and that undoubtedly happens through AI and Super Sampling technologies such as DLSS.