Google Cloud will adopt ARM -based chips for use on servers, making them the latest company to transition from x86 chips to ARM.
This transition should be a pressure for Intel and AMD, the rulers of x86 chips, whose markets are predicted to be increasingly eroded, we were quoted as saying by Reuters, Monday (7/18/2022).
The ARM chip itself is so far better known as a chip for mobile devices, such as mobile phones and tablets. But in 2018, ARM began to introduce chip technology for use in data centers, which until now are still controlled by Intel and AMD.
Then in the last four years, ARM technology, used by many chipmakers, has become more common in various data centers around the world, including data centers owned by Amazon, Microsoft, and Oracle in the United States, as well as Alibaba, Baidu, and Tencent in China.
The use of chips for this data center is not playful, which is very large. Because through the data center, the provider companies then re -sell the computing capabilities to their consumers.
Indeed, they still offer chip -based services from Intel and AMD. However, with the integration of Google Cloud into the ranks of ARM users, it will certainly increase the length of large data center providers that expand the use of the chip.
Cloud computing companies like Amazon and Alibaba even designed their own ARM-based chips, and then manufactured them by chipmakers like TSMC and Samsung.
While a number of other companies, such as Microsoft, Oracle, Google and others, use the services of Ampere Computing founded by former Intel executives. Google said it will use a new chip from Ampere called Altra.
ARM itself recently showcased its latest mobile GPU that supports ray tracing hardware. This ray tracing was previously only available on PC graphics cards.
This GPU is called Immortalis and has at least 10 cores. Immortalis will support ray tracing hardware and a new execution engine for high -end Android phones. There is also Variable Rate Shading to improve performance and battery life.
Ray tracing actually requires a fairly high computing performance, which makes its presence delayed in mobile devices. But now it seems that the ability of mobile chips can meet that need.