Skip to main content

Arm server CPU challenges Intel and AMD

This week, the good news about Arm-based CPUs for non-mobile phones, especially Arm server chips, came together. Among them, the following three have received high attention: First, Apple officially announced that it will abandon the use of traditional Intel CPUs in its Mac series computers and switch to self-developed Arm-based processors; secondly, it is ranked in the latest global supercomputer TOP500. On the list, Fugaku equipped with Arm processors won the championship, which also created the history of Arm chips; in addition, the cutting-edge Arm server CPUs launched new products with 128 cores.

Compared with the various negative news in previous years, the good news about Arm server CPU has continued in the last two years. It seems that after a technical iteration and performance improvement, and downstream users' desire for non-x86 architecture server CPU demand, Arm server CPU is on the rise in both technology and market. Although there is a significant gap in the overall competitiveness of the x86 series, its development momentum seems to have increased a lot and it has entered a period of rapid development.

The reason for this is that, first of all, many large customers particularly need the performance provided by Arm-based CPUs. In addition, the Arm-based software ecosystem, including APP and operating systems, is also maturing. Moreover, the innovation of the x86 series is also decelerating.

The development of cloud computing also gives Arm architecture processors development opportunities. CPUs need some special properties and characteristics to meet the characteristics of cloud-native, including 1. predictable high performance; 2. high scalability; 3. High energy efficiency.
128-core Arm server chip was born
Based on such market demand, in March of this year, Ampere launched the 80-core Arm server chip Altra series, which features high memory bandwidth and high density.

In the past few months, the company has been cooperating closely with several world-class cloud service providers in China and the United States, focusing on studying how to improve the performance of customer applications. After knowing the customer’s application pain points, Ampere launched an enhanced version of Altra Max, its biggest feature is that it has 128 cores.

From 80 to 128, how does the expansion of the kernel help the overall performance of the data center? The first is scalability, which is the most concerned by the industry. To this end, the company compared its Altra series with Intel and AMD processors.

As the number of cores increases, performance expansion also shows linear growth. Jeff Wittich, senior vice president of Ampere products, said: "As the number of cores increases to 160, we can achieve 98% of the theoretically optimal performance."

As shown in Figure 2, compared with AMD's EPYC and Intel's Xeon, Altra's scalability has obvious advantages. In terms of EPYC, with the increase of threads, performance has not been increased accordingly. Xeon's performance is not as good as EPYC. It has not shown a corresponding increase in performance. In the early days, when the number of threads reached more than 20, the performance had already shown a decline. Since Intel was not primarily focused on performance improvement before, as threads increase, power will gradually decrease. Jeff Wittich believes that, especially in the cloud environment, the shortcomings of these two processors are more obvious, because as more and more users, they may enjoy the speed of performance degradation very fast. On the contrary, when using the Altra processor, as new users increase in the cloud, the processor performance that the new users enjoy after the workflow goes to the cloud is the same as before, and there will be no degradation.

Ampere Altea provides ideal performance expansion


It is estimated that by 2023, 80% of workloads will be migrated to microservices, which means that in the future, better single-core predictability will be required, and the performance of each core will be predictable; higher core density is required, A single core is required to have lower power consumption and better efficiency, and this trend will continue to accelerate in the future. This is the reason why the 128-core Altra Max processor was introduced.

Altra Max is also based on TSMC’s 7nm process technology, and its samples will be available in the fourth quarter of 2020. According to Jeff Wittich, the Altra Max and Altra products are slot compatible and have the same memory. This means that the development platform used by customers for Altra can also be used for Altra Max. In the future, Ampere will have a series of product launches, including a Future with a 5nm process technology, which will be available in 2022. The 5nm product has completed the tape-out of the test chip.

Jeff Wittich said that Ampere's design philosophy is to make cloud service providers and ODMs very convenient to use. Each category supports two slots and has the same IO scalability. We have a total of 11 categories of products. Cloud and ODM customers can choose according to their application types and TDP preferences.

Lower TDP power consumption is also very beneficial. Compared with AMD's EPYC, which is also 64 cores, Altra's TDP can achieve below 100. As the TDP decreases, the performance advantages shown will be more obvious. In some application scenarios with very low power consumption requirements, such as edge computing, it can provide 32 high-performance cores, and its TDP is only 45 to 58Watts. Jeff Wittich said: "No other similar product can provide so many high-performance cores and such low power consumption as ours, and does not require the use of fans."
Regarding the problem of customers migrating from x86 to Arm, Jeff Wittich said that in this process, especially the end-users applying to the Altra processor need to do less and less work. We have some software engineers to support this aspect. Work to ensure the smooth progress of basic work. We have also optimized the system and the kernel, and have done a lot of basic work in Java. We will integrate and update some iterative updates of basic software every day. This basic system software runs on the Altra platform without any problems.
Partner
According to Jeff Wittich, Ampere already has a number of partners, such as packet, CLOUDFLARE, Scaleway, Phoenics, GenyMotion, and Nvidia.

It is reported that Ampere has been cooperating with a packet for many years, which is a provider of infrastructure and services for cloud developers, and the packet has begun to use Altra processors.

CLOUDFLARE is mainly doing web page storage. They appreciate the power consumption and performance of the Altra processor very much. It is reported that they will release some relevant data indicators for using the Altra processor at the end of this year.

In addition, the company also cooperates with other types of partners in the entire ecosystem, such as some ODMs, and they also integrate Ampere's products to provide services to end-users. Phoenics will add the Altra series of products to the series they provide in the next few months. Jeff Wittich said: "In the next few months, we will also announce a series of new system integrator partners, especially Chinese partners."

GenyMotion will provide related services to the market based on the Altra series of products. Genymobile mainly provides Android virtual facilities and services, and its development prospects in China are very broad.

Recently, NVIDIA announced that it has provided a series of tools and CUDA libraries, all based on Arm versions. In the past few years, Ampere has been working with NVIDIA to use Altra processors to help them develop a software stack based on graphics cards. In the future, AI and image processing can all run on the Altra processor.

Comments

Popular posts from this blog

AMD's GPU technology enters the mobile phone chip market for the first time

In addition to the release of the Exynos2100 processor, Samsung also confirmed a major event at this Exynos event, that is, the custom GPU that they have worked with AMD for many years will soon appear and will be used on the next flagship machine. The current Exynos2100 processor uses ARM’s Mali-G78GPU core with a total of 14 cores, so the GPU architecture developed by Samsung will be the next Exynos processor, and the GPU will be the focus. This is probably the meaning of Exynos2100’s GPU stacking. The key reason. Dr. InyupKang, president of Samsung’s LSI business, confirmed that the next-generation mobile GPU in cooperation with AMD will be used in the next flagship product, but he did not specify which product. Samsung is not talking about the next-generation flagship but the next one, so it is very likely that a new Exynos processor will be available this year, either for the GalaxyNote21 series or the new generation of folding screen GalaxyZFold3. In 2019, AMD and Samsung reached

Apple and Intel want to join the game, what happened to the GPU market?

Intel recently announced that it will launch Xe-LP GPU at the end of this year, officially entering the independent GPU market, and will hand over to TSMC for foundry. At the 2020 WWDC held not long ago, Apple also revealed that it is possible to abandon AMD's GPU and use a self-developed solution based on the ARM architecture. It will launch a self-developed GPU next year. What happened to the GPU market? Why are the giants entering the game?    Massive data calls for high-performance GPU    Why has the demand for GPUs increased so rapidly in recent years? Because we are entering an era where everything needs to be visualized. Dai Shuyu, a partner of Aiwa (Beijing) Technology Co., Ltd., told a reporter from China Electronics News that visualization requires a large amount of graphics and image computing capabilities, and a large amount of high-performance image processing capabilities are required for both the cloud and the edge.    Aiwa (Beijing) Technology Co., Ltd. is an enterp

NVIDIA officially launches RTX 30 series mobile graphics cards

In the early morning of January 13, NVIDIA officially launched the RTX30 series of mobile graphics cards at the CES2021 exhibition. Ampere-based GPUs have also reached the mobile terminal, mainly including RTX3080, RTX3070 and RTX3060 models. In addition to improving game performance, the RTX30 series of mobile graphics cards have twice the energy efficiency of the previous generation, and support the third-generation Max-Q technology, mainly supporting DynamicBoost2.0 dynamic acceleration technology, WisperMode2.0 noise control, ResizableBAR (similar to AMD’s SAM technology) and DLSS. The third-generation Max-Q technology uses AI and new system optimization to make high-performance gaming laptops faster and more powerful than ever. These technologies include: ·DynamicBoost2.0: The CPU and GPU powers of traditional gaming notebooks are fixed, while games and creative applications are dynamic, and the requirements for the system will vary with the number of frames. With DynamicBoost2.0,