Samsung is using AI to accelerate chip development

As the first SoC equipped with AMD RDNA 2 architecture GPU, Samsung Exynos 2200 related news has been circulating in the market for some time. This product will support ray tracing and variable-rate shading, rumored to easily beat Samsung Exynos 2100, Qualcomm Snapdragon 888, and Apple’s A14 bionic in the 3DMark Wild Life test, which is higher than any current Android-based device and only lower than the device based on the M1 chip.

There have been rumors that Samsung will release this SoC in June or July, but it seems to be postponed. According to Twitter user @FrontTron, Samsung has encountered some setbacks, mainly from manufacturing. Exynos 2200 will use Samsung’s 4nm LPE process, which is different from the 4nm LPX process used by Snapdragon 898. This indicates that there are differences in the manufacturing of the two chips, and the high probability is related to the yield rate. This is likely to also affect the release of Galaxy S22 series smartphones equipped with Exynos 2200. The insufficient production may even cause Samsung to cancel the Exynos 2200 version in some regions.
Samsung Exynos W920

Image: Samsung

Earlier news pointed out that Exynos 2200 still has a power consumption problem, which is higher than expected. Samsung has been looking for a solution to achieve a balance between performance and power consumption. The development of smartphone chips is a time-consuming process. In order to reduce the pressure, Samsung has introduced artificial intelligence technology to develop the Exynos series of products.

The products of the chip design software company Synopsys are used by many companies in the industry, and its DSO.ai tool can accelerate the development process through AI algorithms. According to Wired reports, Samsung confirmed the use of Synopsys’ artificial intelligence software to design Exynos series products but did not confirm which product it is and whether Exynos 2200 also uses this tool. It is foreseeable that in addition to Samsung, more chip design companies will adopt this method in the future to shorten the development cycle and reduce the difficulty of design.