Intel has made several new disclosures in a blog written by the GPU boss and current interim General Manager of AXG (since Raja moved on to become the Chief Architect of Intel), Jeff McVeigh. Distilling the blog, we can find several nuggets of new information that would be very interesting for investors.
Intel is skipping intermediary products: Jumping straight to Falcon Shores MCM-based GPU and moving to a consistent two-year cadence
1. Intel has given us a tentative launch date for the multi-exaflop Argonne supercomputer. Featuring 60,000 Max Series GPUs and 20,000 Max Series CPUs, the installation will feature performance levels over 2 exaflops and will become available to researchers by the third quarter of 2023. The Borealis test system for Argonne has already gone live so there is every reason to believe Intel will hit this milestone as well.
2. The company has announced two more design wins for its datacenter series of products:
- Lawrence Livermore National Laboratories (LLNL) and Sandia National Laboratories are installing thousands of nodes of 4th Gen Intel Xeons in their CTS-2 systems – the supercomputing workhorse of the Department of Energy (DOE). LLNL’s Intel Xeon-powered predecessor, JADE, recently contributed to thebreakthrough in fusion energy, helping to design the optimal package for laser induction.
- Los Alamos National Laboratory (LANL), another DOE research center, is installing more than 10,000 Max Series CPUs for itsCrossroads supercomputer, which will powernational securityandwildfire research.
3. Intel is also taking aim at NVIDIA's data center products, specifically the A100. As Jeff points out in his blog, for material sciences, nuclear engineering, cosmology and plasma physics codes, researchers have measured a 30% to 260% speedup over leading alternative GPUs (from NVIDIA).
4. Finally, we have what is probably the most exciting (at least for us tech enthusiasts) news of the day. There had been so many rumors that Ponte Vecchio (Intel Max Series GPU) might be the last datacenter GPU on the horizon but I am happy to report that this is not the case. Absolutely crushing those rumors, Intel is proudly announcing its datacenter roadmap which will have a two year cadence and will feature Falcon Shores arriving sometime in 2025 to succeed Ponte Vecchio.
Targeted for introduction in 2025, Falcon Shores’ flexible chiplet-based architecture will address the exponential growth of computing needs for HPC and AI. We are working on variants for this architecture supporting AI, HPC and the convergence of these markets. This foundational architecture will have the flexibility to integrate new IP (including CPU cores and other chiplets) from Intel and customers over time, manufactured using our IDM 2.0 model. Rialto Bridge, which was intended to provide incremental improvements over our current architecture, will be discontinued.
The Flex Series product family will also move to a two-year cadence. We will discontinue the development of Lancaster Sound, which was intended to be an incremental improvement over our current generation. This allows us to accelerate development on Melville Sound, which will be a significant architectural leap from the current generation in terms of performance, features and the workloads it will enable.
5. Intel is accelerating their roadmap by cancelling products with incremental updates. Rialto Bridge has been cancelled and so has Lancaster Sound. Intel will instead accelerate development on Melville Sound - which is a code name for the successor to Lancaster Sound. As you can see from the graphic above, Intel has decided to skip an entire step in its Super Compute Silicon Roadmap which is an excellent decision as it will allow it to offer a far more competitive product to the likes of NVIDIA.
Both Rialto and Lancaster Sound were supposed to be products with iterative improvements and with Intel jumping directly to Falcon Shores, Intel is positioning itself to be a force to be reckoned with by 2025. Intermediary products are usually required to iron out the kinks but Intel seems to be extremely confident in their ability to forge ahead and considering they are ahead of schedule on manufacturing too, a turnaround might just be on its way.
The post Intel Disclosure: Jumping Directly To Falcon Shores GPU In 2025, Argonne Supercomputer Ready by 2023 by Usman Pirzada appeared first on Wccftech.
Refference- https://wccftech.com
0 Comments