Since the launch of Zen 2 in November 2019, AMD has taken processor performance to new heights with TSMC's 7nm process, which shows that the competitiveness of process nodes is very important at any time. The migration of TSMC's N7 process also benefited from the use of the same small size chip for desktop processors at that time, ensuring sufficient capacity and supply of desktop and enterprise processors.
With the passage of time, now that this problem has continued to TSMC 5nm and 4nm, coupled with the 3nm process that will be launched in the next 24 months, and TSMC has received orders such as NVIDIA, Apple and other influxes to lock in production capacity, AMD really has to consider how to act next.
2022 is an interesting node for both PC mobile terminals and mobile phones, in the first half of this year, based on TSMC 4nm (N4) Tianji 9000 SoC has been mass-produced, and AMD's customized TSMC 5nm process will be used in AMD's enterprise-level products in the second half of 2022 after the extension, plus TSMC N3 (3nm) plans to mass production at the end of 2022, AMD's process nodes have at least 2 process node gaps.
In an interview with CES2022, Lisa Su said there is no need to worry about this. For AMD, smaller chip technology really helps package more of the processor's performance together. In particular, AMD has eaten dividends on TSMC's strong 7nm delivery to further improve, and for the upcoming 6nm, the 5nm of Zen 4 will be the key to steady progress. Of course, while the process is improving, AMD has also correctly mastered the use of 2D and 3D chips and made the correct response plan, which is exclusive to AMD.
Therefore, the 5nm process process used by AMD is actually deeply customized by AMD. This can often be seen on NVIDIA and Apple.
Although the chip's lead does not depend entirely on the chip process, in the era of small chips, the rich combination and packaging methods mean that smaller process nodes are required to achieve. For example, as early as 2017, Intel packaged 24 to 26 modules in a consumer-grade desktop processor through the form of small chip modularization, and obtained good cost and performance performance.
In the next few years, Intel will become one of TSMC's 3nm major customers through arc-specific display GPUs, and soon we will be able to see whose small chip IP accumulation and packaging tools are more efficient between AMD and Intel under the truly same process.
Of course, AMD is not only focused on the development of consumer products. In fact, AMD has been emphasizing that the RDNA2 graphics architecture has great scalability, whether desktop, laptop or mobile devices, servers, embedded systems, industrial systems and automobiles, can become the future direction of AMD.
Now that AMD has secured orders for infotainment systems including Tesla's Model X, Model S, and Model 3, if all goes well, it will be the first version of a vehicle to use Zen and Vega architectures. At the same time, Lisa Su also said that Tesla's choice of AMD technology in the new solution is not a deep agreement relationship between the two companies, but AMD's Zen and Vega architecture is indeed more suitable to improve the performance of these two models.
At the very least, it is certain that in the high-end models of Tesla's Model X and Model S delivered in the future, Ryzen Embedded + Navi (RDNA2) products will be seen, while the Model 3 and Model Y will use Ryzen SoCs. Yes, friends who shout AMD YES, the range of products available for purchase has expanded again.