News AMD Unveils Zen 4 CPU Roadmap: 96-Core 5nm Genoa in 2022, 128-Core Begamo in 2023

[kwohlt]

I don't think big little is a good idea for servers.. Better power stepping would do the same thing without needing to sacrifice performance on any of the cores.

Big little on desktop also doesn't seem like a good idea. I just don't see the attraction. Saving $20 a year on my electric bill isn't really a concern and I wouldn't be buying high end components if it was. Even at mid range it doesn't seem necessary. A "big" core could run at a lower clock and essentially save as much energy while also able to clock up if needed.

Ryzen is technically ready for big little and was from gen 1. They could easily put "little" cores in a CCX and call it a day... But why?.

Now if we are talking laptops & netbooks it's a different story and makes complete sense. It also would kinda make sense for the low end desktop APU's.

Honestly I think Intel did it just to increase core counts so they didn't look bad compared to Ryzen.

They beat Ryzen... Which is a year old but AMD is releasing the refresh with the 3d vcache which will likely give it the advantage again and with a year old core.

If AMD puts just a little more focus on IPC then Intel would be even further behind. Imagine AMD matching IPC with Intel on every generation going forward... It would do some serious damage to Intel across all sectors. There would be no reason to pick Intel over AMD for any application, big little or not...

"Big little on desktop also doesn't seem like a good idea. I just don't see the attraction. Saving $20 a year on my electric bill isn't really a concern "
If you think that's the purpose heterogeneous architecture ("big.little" is a trademarked by Arm), then I can see why you'd think it's useless.

If you think of it rather as you can design 1 core architecture that is focused on maximum single threaded performance, at the expense of heat and power draw, and then a second core architecture that is incredibly tiny and energy efficient, but it a few years behind on single threaded performance, you could then maximize your single threaded AND multithreaded performance in a given die space and power budget. Under full load, the only about 40 of the 240 watts the 12900K is pulling is for the 8 Gracemont cores, yet these gracemont cores are adding a tremendous boost to multithreading performance for their relatively low power and heat requirements. 8+8 12900K might seem pointless versus the 16+0 5950X that draws less power, but Alder Lake is only the first step. E cores can (and will) be added to desktop chips in a large amount (13900K is going to have 8+16, 13700K will be 8+8, 13600K will be 6+8.). And so on. Intel can just keep adding E cores to their designs to optimize for workloads that benefit most from having more threads, and can have a few dedicated cores for maximum lightly threaded workload performance.

This opens the door for future products, such as a 0 + 64 CPU that can have a lot of multithreaded performance at much lower power draw, price. and heat than a comparable Threadripper, while only slightly losing in single threaded performance. Besides, Desktop SKUs have been scaled up mobile designs for a while now. Desktop ranks 3rd behind laptop and server for importance - using scaled up mobile designs for desktop (as intel has been doing for years) is just the way to go.

 

[TerryLaze]

E cores can (and will) be added to desktop chips in a large amount (13900K is going to have 8+16, 13700K will be 8+8, 13600K will be 6+8.). And so on. Intel can just keep adding E cores to their designs to optimize for workloads that benefit most from having more threads, and can have a few dedicated cores for maximum lightly threaded workload performance.

That's extremely doubtful because there needs to be a big enough user group as a target market, you have to convince all your uncle and aunt type people that they need this level of performance ..... for reasons.
Making overkill products just forces more people to the low end and unless both AMD and intel can keep up with not releasing any low end CPUs anymore, this just isn't going to fly.

This opens the door for future products, such as a 0 + 64 CPU that can have a lot of multithreaded performance at much lower power draw, price. and heat than a comparable Threadripper, while only slightly losing in single threaded performance.

intel already tried xeon phi, it's terrible for desktop.
They even stopped making them at all because they are not even that great for server, hence why Xe is suddenly pushed so hard, it's much better for parallel workloads then any cpu will ever be.

 

[kwohlt]

That's extremely doubtful because there needs to be a big enough user group as a target market, you have to convince all your uncle and aunt type people that they need this level of performance ..... for reasons.
Making overkill products just forces more people to the low end and unless both AMD and intel can keep up with not releasing any low end CPUs anymore, this just isn't going to fly.

intel already tried xeon phi, it's terrible for desktop.
They even stopped making them at all because they are not even that great for server, hence why Xe is suddenly pushed so hard, it's much better for parallel workloads then any cpu will ever be.

The increased E core count on Raptor Lake has been leaked by the same sources that perfectly predicted Alder Lake 2 years in advance. And Not all parallelized work loads - I'm not going to run a bunch of VMs off my VM host's GPU.
Xeon Phi's failure is different - Gracemont is quite capable in single core performance. A 3 Ghz gracemont-based Xeon would certainly offer single core performance on par with most of the servers we're running in our datacenter right now - A Gracemont based, 64 core Xeon would make for a great VM host.

 

[Atom Symbol]

I want to see AMD adopt a big.LITTLE strategy in the future. They have those Jaguar/Bobcat etc cores, it would be nice to see them used across the board for the low/mid end. Having this many cores, it makes a lot of sense to have even 10 or so of them little efficiency cores. Threadripper is aimed at a big core userbase that presumably wants max threads at all times and has the specialty software to leverage it.

I would think that servers would also be a big beneficiary of efficiency cores given the electric-bill sensitivity of many data centers.

Bergamo (Zen 4c) appears to be AMD's version of Intel's E-cores. One difference in comparison to Intel's strategy will probably be that AMD won't be combining Zen 4 cores (aka P-cores) and Zen 4c cores (aka E-cores) on a single chip.

They have those Jaguar/Bobcat etc cores, it would be nice to see them used across the board for the low/mid end.

Jaguar/Bobcat are already in production for potential use in low-end machines. The historical record is showing that there isn't enough demand for such CPUs because they are too slow compared to alternatives. Thus, seeing them used across the board is highly unlikely.

 

[Historical Fidelity]

Bergamo (Zen 4c) appears to be AMD's version of Intel's E-cores. One difference in comparison to Intel's strategy will probably be that AMD won't be combining Zen 4 cores (aka P-cores) and Zen 4c cores (aka E-cores) on a single chip.



Jaguar/Bobcat are already in production for potential use in low-end machines. The historical record is showing that there isn't enough demand for such CPUs because they are too slow compared to alternatives. Thus, seeing them used across the board is highly unlikely.

According to my sources they are thinking about combining an 8 core zen 4 chiplet with a 16 core zen4c chiplet on one chip, so if they do decide to do it it’ll be pretty easy with AMD’s multi-chiplet packaging scheme

 

[lastot069]

AMD is going variable core voltage, so any core can be big or little based on power sent to it.