- In a bid to help developers meet these challenges Intel is releasing new technologies under its Developer Cloud, which allows early access to some of its latest technology.
- Intel’s next-generation Intel Data Center GPU is now being sent to Argonne National Laboratory, where it will power the Aurora supercomputer.
At its second annual Intel Innovation event, Intel Corporation announced that some of its most sophisticated technologies would be added to the Intel Developer Cloud platform. Concurrently, the business disclosed several enhancements throughout its graphics processing unit portfolio.
Intel CEO Pat Gelsinger unveiled the upgrades to the Intel Developer Cloud during his keynote talk at Intel Innovation 2022. The 13th Gen Intel Core family, the company’s next-generation desktop central processing unit, promises up to 15% greater single-threaded performance and 41% better multi-threaded performance than the company’s previous-generation silicon.
In his keynote address, Gelsinger discussed the myriad issues that developers face today, including vendor lock-in, access to the most recent hardware, and security.
In a bid to help developers meet these challenges, Gelsinger stated that Intel is releasing new technologies under its Developer Cloud, which allows early access to some of its latest technology. The new technologies are now accessible through a restricted beta test.
Select customers can now test new platforms, such as the 4th Gen Intel Xeon Scalable CPUs based on its innovative Sapphire Rapids architecture, the 4th Gen Intel Xeon Processor with high bandwidth memory, and the Ice Lake D-based Intel Xeon D processors. Gelsinger said that Intel’s future Habana Gaudi 2 Deep Learning accelerators and the Ponte Vecchio-codenamed Intel Data Center GPU are also available.
Gelsinger said developers could also be interested in checking out the new Intel Geti computer vision platform, which is open to anybody seeking a simple approach to creating powerful Artificial Intelligence (AI) models. Intel Geli offers a unified interface for data upload, annotation, model training, and retraining, which reduces the time, money, and skill necessary to construct new AI models.
Gelsinger said, “In the next decade, we will see the continued digitization of everything. Five foundational technology superpowers — compute, connectivity, infrastructure, AI, and sensing — will profoundly shape how we experience the world.”
Intel is also making significant advancements in GPU technology. Gelsinger stated that the company’s next-generation Intel Data Center GPU is now being sent to Argonne National Laboratory, where it will power the Aurora supercomputer.
In addition, the business has updated its Flex Series GPUs, which offer a single GPU solution for visual cloud applications. Gelsinger stated that they now support prominent AI frameworks such as OpenVINO, TensorFlow, and PyTorch.
In addition, Intel is deploying its software-based Xe Super Sampling gaming performance accelerator to more than 20 of the most popular games. Gelsinger announced the availability of new accelerators for its 4th Generation Intel Xeon Scalable Processors for specialized applications, including AI, data analytics, networking, and storage.
Intel competes with other chipmakers but knows that sometimes the industry can gain from cooperation. Gelsinger demonstrated this when he invited representatives from Samsung and Taiwan Semiconductor Manufacturing Co. to discuss the work of the Universal Chiplet Interconnect Express consortium. The UCIe collaboration establishes an open ecosystem to integrate improved packaging technologies with chiplets made on diverse process technologies by competing manufacturers.
Gelsinger claimed that the UCIe is rapidly becoming a reality, given the support of the three top chipmakers worldwide and more than 80 additional semiconductor companies. He pledged that Intel and Intel Foundry Services, the company’s chip manufacturing branch, will pioneer a new era known as the “systems foundry,” comprised of four fundamental components. Gelsinger stated that these include wafer fabrication, packaging, software, and an open chiplet ecosystem.
Gelsinger cited one of the first inventions to result from this cooperation: A breakthrough pluggable co-package photonics solution. Optical communications, such as photonics, can potentially improve chip-to-chip bandwidth, which might be particularly beneficial in the data center. However, the technology has not lived up to these promises since manufacturing challenges have rendered it prohibitively expensive.
However, Gelsinger stated that Intel has finally resolved this issue by developing a reliable, high-yield, glass-based solution with a pluggable connection. The CEO noted that it promises to minimize production difficulties and save costs, creating new opportunities for interoperable system and semiconductor package designs.
Finally, Intel presented an update on its funding initiatives for the future ecosystem. Intel introduced a USD 1 billion IFS Innovation Fund earlier last year, intending to support early-stage startups and more established companies developing revolutionary foundry ecosystem technology. Since then, Intel has granted money to Astera Labs Inc., SiFive Inc., and Movellus Inc., three firms pioneering across the whole semiconductor stack.