2023-06-24 16:52:30
Exclusive to Nvidia and AMD
Generative artificial intelligence has made exciting progress, and many technological, industrial and commercial companies have quickly adopted this technology, while “graphics chips” have become the most demanded component in the market today, as these chips are considered the “oxygen” that the burgeoning system of artificial intelligence breathes today. .
And any company that really wants to compete in the new world of artificial intelligence must resort to graphics processing units, because without the power provided by these chips, most of the services of generative artificial intelligence will not exist.
At the present time, Nvidia dominates the market for artificial intelligence chips, with a market share of more than 80 percent, as this superiority is due to the company’s possession of two super chips, which are the A100 chip and the H100 chip, which is the most powerful of all in the artificial intelligence chips currently available in the market. He described it as legendary, being the best produced by the human mind in this field, and its price reaches $40,000.
But Nvidia seems to be facing the demise of its absolute domination of the artificial intelligence chip market, due to the MI300X graphics chip from AMD that will make its way to the market later this year.
AMD has always been famous for its production of traditional computer processors, so if the MI300X chip succeeds in gaining market confidence, the company will have entered a new market for it, in an effort to break Nvidia’s dominance in this field.
But the question remains, does the MI300X, which AMD said is the most advanced in the field of artificial intelligence so far, really have the specifications that enable it to break Nvidia’s dominance?
Digital transformation expert Rudy Shoshani said, in an interview with “Sky News Arabia Economy”, that the background from which to evaluate the super MI300X chip is not only related to the technical specifications of the chip, but it must also be taken into account that it is the first chip for AMD in the field of chipsets. Artificial intelligence, which means that the company does not have sufficient experience in this field, compared to Nvidia, indicating that the MI300X is superior in some specifications to the H100 chip from Nvidia, but no one knows yet how the performance of the MI300X will be and its interaction with operational systems, and this is a point It is considered crucial, in the process of market adoption of the AMD chip.
According to Shoshani, the MI300X uses up to 192 GB of memory, which means that it is more suitable for artificial intelligence programs than the H100 chip, which uses up to 80 GB of memory, pointing out that this matter gives the AMD chip precedence in theory and not in reality, and that The fact that this chip does not have what is known as the Transformer Engine, which is owned by the competing H100 chip, which gives it up to 6 times higher performance, and the capabilities to significantly accelerate the performance of artificial intelligence.
The actual advantage of “Nvidia”
Shoshani explains that so far MI300X has only a “preference” in descriptions, as there are many issues related to the performance of the chip that have not yet been published, while Nvidia chips have a “preference” in terms of market confidence and speed, and a history full of successes in this field, expecting Nvidia intends, before the MI300X hits the market, to update the specifications of the H100 chip, in a way that prevents any other company from claiming that it has the strongest specifications.
Shoshani confirms that all that has been said does not mean at all that the MI300X will have no hope of success, as this matter depends on the ability of the chip to convince companies, in the event that it proves through its performance that it is a strong chip, as it is possible to know the extent of the market’s adoption of it, during The first half of 2024, considering that breaking Nvidia’s dominance is a “big phrase” that will not be achieved in the foreseeable future, as companies that are racing with each other, to reach innovations in artificial intelligence technologies, will not abandon Nvidia products in favor of a product that knows nothing about its performance, at least in perspective range.
For his part, technical analyst Khaled Moussa confirms, in an interview with “Sky News Arabia Economy”, that talking about the extent of AMD’s ability to compete with Nvidia in the manufacture of artificial intelligence chips is premature, as there is great difficulty in pulling the rug of this industry from under it. Legs of Nvidia, which excels in many technical elements in this area.
Musa believes that the MI300X chip from AMD could be the number two option in the market after the Nvidia chipset, especially if it is available in the market at a cheaper price than the H100, which costs $40,000, noting that many startups cannot afford it. It bears the cost of Nvidia chips, as this problem appeared in the market recently, and therefore AMD’s provision of artificial intelligence chips at an affordable price may help the company reserve a firm seat for it in this industry, if the MI300X proves its ability to compete operationally, knowing that AMD has not yet revealed the price of its new product.
1687670771
#AMD #ability #break #dominance #Nvidia