• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Raptor Lake Leaks + Intel 4 developments

Games don't need more than 8 performance cores and that will be the case until consoles get more than 8 cores - so just adding efficiency cores for multithreaded performance makes sense

+1. Having the E cores to handle all those small background processes (Windows defender, Discord, whats app, audio apps etc etc) while the 8 P cores power their way through the actual game is clearly working well, in terms of performance and energy efficiency.
 
+1. Having the E cores to handle all those small background processes (Windows defender, Discord, whats app, audio apps etc etc) while the 8 P cores power their way through the actual game is clearly working well, in terms of performance and energy efficiency.
Big cores can handle all of that too cuz of much higher performance, and not only that, but smaller latency too.
 
Big cores can handle all of that too cuz of much higher performance, and not only that, but smaller latency too.

A small number of E cores to boost efficiency makes some sense but Intel are using them to plaster over their inability to scale up performance core numbers.

As things stand I prefer AMDs approach, using multiple chiplets to scale up full fat cores that wouldn't be possible on a monolithic die, it seems better than tacking on E cores.
 
+1. Having the E cores to handle all those small background processes (Windows defender, Discord, whats app, audio apps etc etc) while the 8 P cores power their way through the actual game is clearly working well, in terms of performance and energy efficiency.
I do not see the benefits of more e cores, how many do they realistically need to handle backgroud processes. I think thsi just shows that Intel are maxed out on P cores, extra e cores will equal small performance gain <10%
 
A small number of E cores to boost efficiency makes some sense but Intel are using them to plaster over their inability to scale up performance core numbers.

As things stand I prefer AMDs approach, using multiple chiplets to scale up full fat cores that wouldn't be possible on a monolithic die, it seems better than tacking on E cores.
Intels architecture is clearly working though as just 8 high performance P cores are enough to beat ryzen in games and ST workloads while with the E cores tacked on ADL beats ryzen in MT workloads.

Raptorlake will further enhance the P cores to increase ST / gaming and then bolt on another 8 E cores to add more MT so it should result in decent gains again.

AMD will certainly have to increase core counts on Zen 4 to keep up in MT performance as CPUs like the 5600X and 5800X are already 40% behind in what was tradionally a strong area for Ryzen.
 
Last edited:






A
t just 75w the 12900k is on par with the 5950x in gaming. Going above 100w does almost nothing for games and going above 125w does absolutely nothing for games. Also while not in this graph, on the TPU site there is no difference in single core performance anywhere between 75w and 240w.


If all you care about is single core you can lock the 12900k to 75w and call it a day. If you care just about gaming lock it to 100w or 125w and call it a day. The only people who should keep the power limit unlocked on a 12900k is those who can make use of all core workloads like video rendering
 
Last edited:
Intels architecture is clearly working though as just 8 high performance P cores are enough to beat ryzen in games and ST workloads while with the E cores tacked on ADL beats ryzen in MT workloads.

Raptorlake will further enhance the P cores to increase ST / gaming and then bolt on another 8 E cores to add more MT so it should result in decent gains again.

AMD will certainly have to increase core counts on Zen 4 to keep up in MT performance as CPUs like the 5600X and 5800X are already 40% behind in what was tradionally a strong area for Ryzen.
Well Zen 3 is 1 year old architecture, nothing miracle. Zen 4 will have IPC in range of RPL, plus more advanced v-cache. They will increase core count to keep up in MT no doubt, but they won't need that much cores because one Zen 4 core will be much stronger than RPL e core. Zen 3 server had max 64 cores, while Zen 4 will have 96, that will translate to desktop too. Chiplets allows them to pump cores while keeping good yields.
 
Well Zen 3 is 1 year old architecture, nothing miracle. Zen 4 will have IPC in range of RPL, plus more advanced v-cache. They will increase core count to keep up in MT no doubt, but they won't need that much cores because one Zen 4 core will be much stronger than RPL e core. Zen 3 server had max 64 cores, while Zen 4 will have 96, that will translate to desktop too. Chiplets allows them to pump cores while keeping good yields.
Increased core count, Vcache and no ddr4 compatibility will make zen 4 ridiculously expensive so it will need to beat RPL by a big margin to make it worthwhile.
 
Well Zen 3 is 1 year old architecture, nothing miracle. Zen 4 will have IPC in range of RPL, plus more advanced v-cache. They will increase core count to keep up in MT no doubt, but they won't need that much cores because one Zen 4 core will be much stronger than RPL e core. Zen 3 server had max 64 cores, while Zen 4 will have 96, that will translate to desktop too. Chiplets allows them to pump cores while keeping good yields.

Zen 4 IPC is currently an unknown. First AMD has to match ADL IPC and gaming efficency. RPL is rumoured to be another 10-15% IPC over ADL.

Zen 4 could end up being 10% higher IPC than RPL, if it is I'll get one :)
 
Zen 4 IPC is currently an unknown. First AMD has to match ADL IPC and gaming efficency. RPL is rumoured to be another 10-15% IPC over ADL.

Zen 4 could end up being 10% higher IPC than RPL, if it is I'll get one :)
ADL isn't problem, Intel is using much higher power to achive that lead, Zen 3 with 3d-cache and little higher power limit will be enough. RPL will be small upgrade over ADL, you could call it polished ADL. And someone mention Zen 4 won't support ddr4, well that means it will be full ddr5 controller without compromises, it will support much higher speed, and by that time DDR5 availability will be better.
 
'Raptor Lake' is probably just Golden Cove cores in a different configuration imo.

I don't see much point in another desktop series in 2022. Most of the 'Golden Cove' desktop products are still to launch in 2022, and HEDT and server CPUs. Have Intel really got enough viable silicon to do all that, and produce another desktop series too on the same 'Intel 7' Process?

It's a lot of effort for not much gain, in reality I think their plans are focused on Meteor Lake and successive CPUs.

Another important thing to note, is that Intel has said nothing about a new core technology in between 'Golden Cove' and 'Redmond Cove' (the presumed name of the Meteor Lake architecture). As some have pointed out though, there might be some changes made to Golden Cove in some Intel CPUs. I find it unlikely there would be an IPC improvement, we didn't see this with Skylake iterations.
 
Last edited:
'Raptor Lake' is probably just Golden Cove cores in a different configuration imo.

I don't see much point in another desktop series in 2022. Most of the 'Golden Cove' desktop products are still to launch in 2022, and HEDT and server CPUs. Have Intel really got enough viable silicon to do all that, and produce another desktop series too on the same 'Intel 7' Process?

It's a lot of effort for not much gain, in reality I think their plans are focused on Meteor Lake and successive CPUs.

Another important thing to note, is that Intel has said nothing about a new core technology in between 'Golden Cove' and 'Redmond Cove' (the presumed name of the Meteor Lake architecture). As some have pointed out though, there might be some changes made to Golden Cove in some Intel CPUs. I find it unlikely there would be an IPC improvement, we didn't see this with Skylake iterations.

Raptor Lake is a new architecture, feating IPC improvements on the P cores, along with 8 extra E cores, making this a 24 core CPU.

It also has cache improvements, again according to leaks.
 
Intel hasn't mentioned continuing the use of the big.LITTLE architecture though. Sapphire Rapids and Meteor Lake aren't hybrid designs like Alder Lake.

Sorry, but I just don't buy that they have a new, improved architecture ready for 2022, besides 'Golden Cove', which is barely out the door. If Intel had one, it would be going into their server CPUs in the 2nd half of 2022.

AMD is prioritizing their most advanced architecture (Zen 4, on TSMC's 5nm fab. tech) for their servers and cloud chips in late 2022... If you look under the Wikipedia entry for the fifth Zen generation, that's all it says. Link here:
https://en.wikipedia.org/wiki/Zen_(microarchitecture)#Fifth_generation

Zen 3 with V-Cache will be AMD's next desktop CPU series in 2022, they will compete well enough with these, especially because it's an easier upgrade path for many (should be able to just fit a new CPU and boot up...).
 
Last edited:
Intel hasn't mentioned continuing the use of the big.LITTLE architecture though. Sapphire Rapids and Meteor Lake aren't hybrid designs like Alder Lake.

Sorry, but I just don't buy that they have a new, improved architecture ready for 2022, besides 'Golden Cove', which is barely out the door. If Intel had one, it would be going into their server CPUs in the 2nd half of 2022.

AMD is prioritizing their most advanced architecture (Zen 4, on TSMC's 5nm fab. tech) for their servers chips in late 2022...

Zen 3 with V-Cache will be AMD's next desktop CPU series in 2022, they will compete well enough with these, especially because it's an easier upgrade path for many (should be able to just fit the new CPU and boot up...).
Big/little is doomed, we all know that. 3D stacking cores is not far away, and Intel will soon have something similar to chiplets so they won't need those crap little cores. ADL was temporary solution so Intel could say they have same amount of coress.
 
You can see why it's in Intel's interests not to deny rumours about a desktop 'Raptor Lake' CPU generation in 2022 (or even confirm any plans for 2022).

They want people to upgrade to Alder Lake, but if people knew LGA1700 motherboards are likely to be a dead end after the 12th generation, they might not be so keen, and just wait for Zen 4 and Meteor Lake desktop CPUs in 2023.

Propaganda is a hell of a drug :)
 
You can see why it's in Intel's interests not to deny rumours about a desktop 'Raptor Lake' CPU generation in 2022 (or even confirm any plans for 2022).

They want people to upgrade to Alder Lake, but if people knew LGA1700 motherboards are likely to be a dead end after the 12th generation, they might not be so keen, and just wait for Zen 4 and Meteor Lake desktop CPUs in 2023.

Raptor Lake's in the official "leaked" roadmaps mate, the same as we've been explosed to for many years for each Intel generation. Will you accept it exists once it's aviailable to purchase etc?
 
Yes ofc I would, if they can build better CPUs (10% IPC improvement etc), who wouldn't want that?

They aren't official roadmaps. I don't understand why some people keep claiming they are, it confuses the hell out of people. No respected website (Such as Wikipedia or Wikichip) has said anything about 'Raptor Lake', because there is no official news from Intel.

Just taking MLID's word for is just gullible (sorry if that seems rude...), he's just one guy. It doesn't matter how many times he supposedly 'has been right' in the past (or not).

Notice how none of the 'leaks' give any significant detail about the 'Raptor Cove' core architecture itself? Odd isn't it? If I was leaking info, that's where I'd start.
 
Last edited:
Back
Top Bottom