Intel AXG optimizes data center GPU roadmap underlined with a 2-year cadence

Late last year and early this year, Intel finally released its full line of data center-focused Arctic Sound and Ponte Vecchio GPUs, as well as its Sapphire Rapids CPUs. These chips have been a long time coming, but they offer solid performance for high-performance computing, as evidenced by their adoption for the Aurora supercomputer at Argonne National Labs. Last we heard Intel’s Accelerated Computing Group (AXG) would follow these GPU parts with Lancaster Sound and Rialto Bridge. The Lancaster sound was intended to be an incremental improvement on the Arctic Sound, using a revised Xe-HP architecture, while the Rialto Bridge was intended to be a refined Ponte Vecchio with improved density resulting in a higher core count.

It turns out none of these parts are destined to see the light of day. In an editorial published on Intel’s website, Jeff McVeigh, corporate vice president and interim general manager of AXG, announced the discontinuation of both product lines. However, this isn’t the death of Intel’s GPU efforts; far from it. Instead, Intel is making what’s probably a smart move over the long term.

One of Intel’s current Data Center Flex series GPUs intended for cloud video processing.

It’s the same thing 3dfx should have done: recognize that your current product needs more than one makeover to keep up with your competitors, cancel the updated and revised version, and refocus your efforts on entirely new products. To that end, McVeigh says Intel’s next offerings in these two areas will be the previously announced Falcon Shores, as well as a new product codenamed Melville Sound. This announcement fits perfectly with recent comments from Tom “TAP” Petersen about Intel simplifying its approach to consumer graphics. Intel’s original plans for Xe included many different architectures targeting a variety of specific applications, and the company found it was just too difficult to manage and inefficient. Instead, the company takes a much simpler approach, similar to NVIDIA and AMD: Build one or two graphics architectures at once and customize those pieces for multiple applications.

READ :  Oosto’s Latest Software Release Pushes Real-time Identification and Threat Detection to the Edge

Falcon Shores hosts any mix of x86 and Xe cores for versatile computing power.

We don’t have any information about Melville Sound, but we assume it will be a datacenter adaptation of the Battlemage GPU that should power the next generation of Arc graphics cards. Meanwhile, Falcon Shores will be Intel’s first “XPU,” a tile-based piece with the ability to mix Xe graphics cores and x86 CPU cores in a single package.

But that won’t be seen until 2025. According to McVeigh, Intel is moving to a two-year cadence for GPU parts in data centers. Given that the company’s Flex series and Max series GPUs have only just launched, we can expect to wait another 18 to 24 months for the next generation to launch.