Report: AMD May Not Launch Zen 4, RDNA3 Until Q4 2022

Report: AMD May Not Launch Zen 4, RDNA3 Until Q4 2022

A report is making the rounds claiming AMD will not launch Zen 4 until Q4 2022, with RDNA3 tipping up at the same time. It’s not the first time AMD would have synchronized a CPU and GPU launch, but it implies a longer-than-expected delay between Zen 3 and Zen 4.

Q4 2022

— vegeta (@Broly_X1) June 7, 2021

One thing we’ve learned in recent days is that AMD will launch Ryzen CPUs equipped with a large, vertically mounted L3 cache later this year. It’s not clear which CPUs the feature will come to — AMD showed off a 5900X 12-core thusly equipped — but we expect to see these CPUs launch before the end of the year. It is not clear how far down the product stack the feature would stretch, but AMD has implied we could see a 15 performance uplift at the same frequency compared with a 5900X with a standard 32MB L3 cache per chiplet.

We know less about what RDNA3 might look like. Texts from late May hinted that we might see a dual-chiplet architecture with 80 CUs per chiplet, or 10,240 cores in total assuming 64 GPU cores per CU.

-5nm(no matter TSMC or SEC)-AD102 in transition or GH202 in revolution?-RDNA3, GFX11, how does AMD reach its perf goal? Double the SIMD in CU?-How about Intel?

— kopite7kimi (@kopite7kimi) May 22, 2021

At a guess, SEC is Samsung Electronics Corporation, not the Securities and Exchange Commission. Some aspects of this make sense. RDNA3 being on 5nm along with Zen 4 by the end of 2022? Easily believable. But if this timeline is accurate, and always take reports like this with a grain of salt, AMD won’t refresh RDNA2 for quite a long time.

AMD has not kept pace with Nvidia’s GPU launch schedule for at least five years. RDNA and RDNA2 were pitched as closing the gap, with RDNA debuting in July 2019 and RDNA2 launched November 2020. Holding RDNA3 until Q4 2022 would be a roughly two-year gap between RDNA2 and RDNA3. That’s longer than we’ve seen previously; the gap between Vega 64 (August 2017) and Radeon VII (February 2019) was 18 months, not 24.

AMD has struggled to supply the GPU market since pandemic-related shortages began, probably due to the number of chips it’s been building for Sony and Microsoft. The PlayStation 5 and Xbox Series X SoCs are not as large as GPUs like the Radeon 6800 XT, but they aren’t exactly small, either. The Xbox Series X is a 360.4mm sq chip, while the PS5 measures 308 mm sq. That’s roughly on par with the 6700 XT, which fields a 335mm sq chip.

This rumor isn’t necessarily true, of course. Even if it is, AMD might possibly have an RDNA2 refresh planned with tweaks to clocks or the underlying microarchitecture of the GPU. It seems unlikely that AMD would play the same V-Cache trick again. We don’t even know if it would benefit from doing so, seeing as the RDNA2 family has much more cache available by default than your typical high-end Ryzen 5000 (32MB versus 128MB).

The biggest feature RDNA3 could offer today — regardless of when it launches — is availability at MSRP.

Continue reading

Intel Launches AMD Radeon-Powered CPUs
Intel Launches AMD Radeon-Powered CPUs

Intel's new Radeon+Kaby Lake hybrid CPUs are headed for store shelves. Here's how the SKUs break down and what you need to know.

SpaceX Launches ‘Better Than Nothing’ Starlink Beta
SpaceX Launches ‘Better Than Nothing’ Starlink Beta

Those lucky few who have gotten invitations to try the service will have to pay a hefty up-front cost, and the speeds aren't amazing. Still, it's a new generation of satellite internet.

AMD May Allow Custom RX 6900 XT Cards, Launch Stock May Be Limited
AMD May Allow Custom RX 6900 XT Cards, Launch Stock May Be Limited

There are rumors that Nvidia may not be the only company facing production shortages this holiday season. High-end GPUs might just be very hard to find in general.

Intel Launches New Xe Max Mobile GPUs for Entry-Level Content Creators
Intel Launches New Xe Max Mobile GPUs for Entry-Level Content Creators

Intel has launched a new consumer, mobile GPU — but it's got a very specific use-case, at least for now.