AMD's next-gen Ryzen 8000-series 'Strix Point' accelerated processing unit may pack quite a punch with twelve Zen 5-class cores and a reworked RDNA 3.5-based integrated GPU if a leak from Golden Pig Upgrade Pack is to be believed (via HXL). While Golden Pig Upgrade Pack has a good reputation, the information is unofficial and should be taken with discretion.
According to the leak, AMD's Ryzen 8000 'Strix Point' APU will pack four fully-fledged Zen 5 cores with 16 MB L3 cache, eight Zen 5c cores with 8 MB L3 cache, and an integrated GPU with 8 WGPs (1024 stream processors) based on the RDNA 3.5 architecture into a big monolithic die. The CPU cores will purportedly be distributed over two core complexes (CCX).
The Strix Point APU will be AMD's first hybrid processor featuring 'big' and 'little' cores, so it will be interesting to see how the processor will stack up against AMD’s existing offerings as well as Intel’s Meteor Lake offerings, which are said to come with up to 14 cores that will process up to 20 threads simultaneously. Meanwhile, since both Zen 5 and Zen 5c cores allegedly support threading, Strix Point will be able to process up to 24 threads at once.
The information about 12 Zen 5 cores in the upcoming APU has been reported before, so some might consider the new information as yet another sign that AMD is working on such an APU. Meanwhile, it is entirely possible that the company will have Strix Point APUs with different configurations, though we'll refrain from making assumptions about possible configs.
Arguably the most intriguing part of the Strix Point leak is that the APU will feature a graphics processor with eight RDNA 3.5-class WGPs, which points to 1024 ALUs. We can only guess about the clock rate of the iGPU as well as its intended performance, but the fact that the company decided to increase the number of stream processors in its built-in graphics processor indicates that there is a plan to offer higher performance with this unit.
AMD's Ryzen 8000-series Strix Point processors for laptops are expected to hit the market in 2024, so it is highly likely that some of AMD's partners are already test-driving them. Therefore, it is not surprising that there are leaks of their specifications, and chances that the APUs will indeed feature 12 cores and a potent iGPU are fairly high.
Nonetheless, at this point, take the information with a grain of salt since it comes from an unofficial source and can be inaccurate.
Stay on the Cutting Edge
Join the experts who read Tom's Hardware for the inside track on enthusiast PC tech news — and have for over 25 years. We'll send breaking news and in-depth reviews of CPUs, GPUs, AI, maker hardware and more straight to your inbox.
Anton Shilov is a Freelance News Writer at Tom’s Hardware US. Over the past couple of decades, he has covered everything from CPUs and GPUs to supercomputers and from modern process technologies and latest fab tools to high-tech industry trends.
I really hope AMD and Intel APUs get to a point where Nvidia has no reason to release price gouged low end cards that aren’t any more powerful than the last model but come with DLSS 4’s “Artificial Frame Injection” tech lol, or is it “Frame Synthesis” tech, or “frame Infusion” tech??? HahaReply
Having the 4 fastest cores in one CCX by themselves could be a problem. Should be interesting to see the benchmarks.Reply
Intel should make a desktop APU.The Historical Fidelity said:I really hope AMD and Intel APUs get to a point where Nvidia has no reason to release price gouged low end cards that aren’t any more powerful than the last model but come with DLSS 4’s “Artificial Frame Injection” tech lol, or is it “Frame Synthesis” tech, or “frame Infusion” tech??? Haha
Looking good just hope AMD do not find a new way to shoot themselves in the foot.Reply
This is dual-CCX with heterogeneity, in contrast to the last few 8-core APUs to come before it. So if they've found a way, we probably know what it is already. You also have to wonder what they will do for the cut down models. 2+6?thisisaname said:Looking good just hope AMD do not find a new way to shoot themselves in the foot.
Why did AMD just killed Jaguar cores? Think could be usefull now at big.LITTLE.Reply
This is the way they are going for now. Big cores, and medium cores with the same IPC (when no cache penalty), lower clocks, less L3 cache (optional, depends on the design), and an area reduction (depends, they can bin the big cores). It avoids some of the problems of Intel's approach and it's easier to design.adunlucas said:Why did AMD just killed Jaguar cores? Think could be usefull now at big.LITTLE.
Whenever they start using FinFlex, they should be able to have performance optimized and area optimized cores on the same die. I don't know if Strix Point will do that yet. Whatever the case, AMD's "big" cores are very tiny these days, so including 8 or 12 of them is no problem.
Sounds like an OK compromise. If they can keep the power to performance ratio good, it might be a good thing.Reply
I don't need a laptop for a while, but I'd love to get a long lasting Linux beast as the next one.
It'll be interesting to see how they approach the "c" cores with Zen 5. Ian Cutress did an interview with I think PC World where he talked about the clockspeed scaling with Zen 4c. With Zen 4c the power efficiency is absolutely shot compared to Zen 4 once you reach a certain clockspeed so you'd be better off just going with full cores and maximizing turbo. He was pretty confident this was why we wouldn't see any hybrid CPUs with Zen 4, but if they're planning one with Zen 5 something must have changed.Reply
Servethehome.com reviewed the 128c/256t Bergamo. They loaded a dual socket system with all cores at 100% load. The Bergamo stayed rock steady at 3.1GHz across 256c for hours on end. Note 3.1GHz is the all core boost as well.thestryker said:Ian Cutress did an interview with I think PC World where he talked about the clockspeed scaling with Zen 4c. With Zen 4c the power efficiency is absolutely shot compared to Zen 4 once you reach a certain clockspeed so you'd be better off just going with full cores and maximizing turbo.
Originally the Bobcat core was designed to give 80% of the performance as a K8 at the same clock speed but at only an 18W envelope. However, after several iterations it was never scaling as well in performance. The final core design, Puma, was pretty anemic by 2014 even with the top end chip A8-6410. Even with 4 cores it was slower than the i3-4030u and both were 15W chips. Zen4c on the other hand keeps close to Zen4's IPC and still has SMT but lower clock rates to keep power down while increasing core count. Not to mention the cores are all the same uArch. This makes it easier for the CPU scheduler and programmers.adunlucas said:Why did AMD just killed Jaguar cores? Think could be usefull now at big.LITTLE.