After nearly a decade of development, Ampere release the AmpereOne high-performance server CPU later this year, which uses Arm-based server technology. Arms partners, Marvell and Amazon, use Arms server architecture for their cloud servers.
Ampere’s plans illustrate that Arm server CPUs will widen the gap in performance with the x86 CPUs
During Ampere’s annual update on the company’s future and the rest of this year’s business plan, the company revealed it would be releasing a core 5nm AmpereOne processor later this year.
AmpereOne is also based on the ARM ISA. Customers expect higher performance, better power efficiency and better scalability from CPUs every year. Developing and launching proprietary core CPUs can help us focus all our attention and focus on our customers’ needs and meet their corresponding requirements.
— Jeff Wittich, chief product officer of Ampere Computing, in an interview with Leifeng.com
The company plans to improve power consumption and processor performance for its new server CPU, especially after identifying it as one of the top seven data center customers worldwide — a list includes Alibaba Cloud, Microsoft Azure and Tencent Cloud.
This progress is an excellent sign for Arm, as the company will open up server CPU space even further between the company’s technology and its x86 server processors. Ampere’s processor is three times faster than standard x86 CPUs, while the performance-to-power ratio is four times better. The Ampere Altra CPU series uses half the power of x86 server processors and delivers 200% better performance. The main contributor to the high performance of Altra is the increased number of cores of the CPU.
Our 128 core product is now the industry leader, twice as fast as other CPUs. Only one thread runs on each core and keeps all cores running at a consistently high frequency. Provides a dedicated, high-capacity, low-latency cache and an intelligent, high-bandwidth mesh interconnect structure to connect all high-performance cores together, breaking the traditional CPU utilization bottleneck that yields diminishing returns as demand increases. It also uses advanced DDR and PCIe technology designed to achieve maximum capacity and expand memory and I/O bandwidth. It can effectively avoid mutual interference between users, while providing excellent scalable performance and ultimately achieving maximum utilization.
The Ampere Altra series is built solely for cloud computing. Ampere has intentionally removed features commonly seen with their processors to improve power consumption and surface area to achieve single-core power consumption that is 67% or more inferior to standard CPUs.
One of the biggest drawbacks of traditional x86 is that they are not specifically designed for data centers, nor for cloud and cloud services. They were used in data centers in the past because there were no better options than x86 CPUs at the time.
Cloud native processors are a huge advantage. They typically have a higher number of cores, with each core running only on a separate thread, allowing for greater security and higher performance without sacrificing increased pressure and workload.
Our upcoming proprietary core AmpereOne is also based on Arm ISA. The microarchitecture is very different from the Neoverse provided by Arm, but we can’t give too much information just yet. The self-developed core AmpereOne can be perfect It is compatible with our current products Altra and Altra Max cloud native processors. What the user has optimized for Ampere Altra/Altra Max can also be perfectly applied to our AmpereOne from the user’s point of view. Moreover, we have also cooperated accordingly with compilers such as GCC and LLVM to further ensure compatibility with all our products.
Arm’s powerful processors have a glaring demise: the software ecosystem. Hardware and software need constant improvements to improve the Arm ecosystem. In addition to partnering with cloud platforms and hardware ODMs and OEMs, Ampere has partnered with Gigabyte and ADLINK to design expansion kits for AI and automated vehicles and other products to extend various applications.
We have applied for hundreds of patents covering the performance and functions of our products. The latest product, AmpereOne, uses the proprietary core of Ampere Computing, is based on the 5nm process and supports PCIgen5 and DDR5. The new product has started to send samples. , very curious about customer feedback.
— Renee James, president of Ampere Computing
Ampere also explains that their powerful, energy-efficient edge cloud processors can reduce power to 70W, meaning Ampere’s processor, with 32 cores, can consume nearly five times less power than a comparable x86 processor.
Ampere is exploring more areas, including their CPUs. Still, the company is also exploring hybrid quantum computing for future use, which several other companies, including Intel, AMD, NVIDIA, and Meta, have also looked at. Ampere will use its Developer Program to bring more than 130 applications to developers, ranging from AI inference solutions to advanced cloud databases.
Through regression testing, we can guarantee a wide range of adaptability while guaranteeing high-quality results based on adaptability.
— Jeff Wittich
Statista predicts that the market size of ARM-based data centers and cloud CPUs will reach $58 billion in 2028, 14 times more than in 2019, and that the market share will increase by 20 percent from three years ago. However, Arm is still working on the cloud data center market, so it will be some time before they finally reach that goal.
News source: IT-home