Raptor Lake Leaks + Intel 4 developments

[Perfect_Chaos]

I just wish there were more Z690I boards DDR4 available for the ITX market right now. It's so stale with the only options being Gigabyte Aorus Z690I which is currently borked due to that PCIE issue or the Asrock Z690I AX which is so anaemic, it shouldn't be called a Z board.

Never had ITX so can't say much on it, looks a bit limiting though.

One problem is gonna be the BIOS flashing to support the new CPUs, which i guess you can pay some retailers to do, but not all will do that.

 

[Grim5]

SiSoft has their hands on an engineering sample 13900k and they are impressed, it's showing up to 50% performance improvement over the 12900k in workloads that benefit from the extra cache or cores.

There is more gains to be had, SiSoft says their 13900 has a clock speed limit of 3.7ghz

 

[Dave2150]

SiSoft has their hands on an engineering sample 13900k and they are impressed, it's showing up to 50% performance improvement over the 12900k in workloads that benefit from the extra cache or cores.

There is more gains to be had, SiSoft says their 13900 has a clock speed limit of 3.7ghz

New architecture & cache structure of Raptor Lake is looking good, can't see how Zen4 will match a 13900k in ST or MT at this point, yet alone the 14900k that'll soon follow in Meteor Lake.

 

[cliffy]

New architecture & cache structure of Raptor Lake is looking good, can't see how Zen4 will match a 13900k in ST or MT at this point, yet alone the 14900k that'll soon follow in Meteor Lake.

I think 13900k will edge out Zen 4 in ST but Zen 4 wull win in MT workloadss. Meteor Lake on desktop will be competing with Zen 5 not Zen 4 and is unlikely to arrive untill end of 2023. I do however think Zen 4 will win in majority of games (mainly down to increased clock speed) especially when Zen 4 3D is launched

I'll say again raptor lake is not new architecture, by your definition Zen 3d would be also.

 

[uscool]

New architecture & cache structure of Raptor Lake is looking good, can't see how Zen4 will match a 13900k in ST or MT at this point, yet alone the 14900k that'll soon follow in Meteor Lake.

but would need another board for 14900k ?? I think Zen4 3d and Zen4 will be on AM5 ? maybe even Zen5 ? I would have my money on AMD supporting the socket longer

 

[humbug]

SiSoft has their hands on an engineering sample 13900k and they are impressed, it's showing up to 50% performance improvement over the 12900k in workloads that benefit from the extra cache or cores.

There is more gains to be had, SiSoft says their 13900 has a clock speed limit of 3.7ghz

SiSoft are Intel.

PS: its also the software Intel used to prove the 15% performance improvements they claimed for every quad core of the decade long refresh, all of them turned out to be 3%.

 

[Dave2150]

SiSoft are Intel.

PS: its also the software Intel used to prove the 15% performance improvements they claimed for every quad core of the decade long refresh, all of them turned out to be 3%.

How do you think a 16 core 5950X, or a 16 core Zen4 CPU, will fare vs a 24 core 13900k in MT applications?

 

[Troezar]

How do you think a 16 core 5950X, or a 16 core Zen4 CPU, will fare vs a 24 core 13900k in MT applications?

Aren't there rumours of a 24C Zen4 CPU?

 

[Armageus]

How do you think a 16 core 5950X, or a 16 core Zen4 CPU, will fare vs a 24 core 13900k in MT applications?

I think AMD's 16 Core Zen4 will use less power than a 8+16 core 13900k, whilst performing fairly similarly

 

[Dg834man]

Aren't there rumours of a 24C Zen4 CPU?

Unlikely robert hallock said in an interview 16 cores.

 

[LePhuronn]

How do you think a 16 core 5950X, or a 16 core Zen4 CPU, will fare vs a 24 core 13900k in MT applications?

The 5950X will get crushed
The 6950X will do the crushing

 

[humbug]

I think AMD's 16 Core Zen4 will use less power than a 8+16 core 13900k, whilst performing fairly similarly

^/

 

[Troezar]

Ah I had missed it was 8+16. That really does tell you that Intel can't do 16 full power cores with any kind of sensible power consumption. Feels like a stop gap measure.

 

[Bencher]

Ah I had missed it was 8+16. That really does tell you that Intel can't do 16 full power cores with any kind of sensible power consumption. Feels like a stop gap measure.

That's a misconception based on flawed logic. Power consumption is a total non issue. They can put 32 full cores and it wouldn't change a thing. I have no idea why people keep propagating this idea, it's a bunch of horse manure. You do realize cpu's have power limits? You can dictate exactly how much power you want them to consume. So they can have a 32 full power cored CPU drawing 100 watts if they want to. I mean, it's the exact same thing AMD is doing. The 5950x has more cores than the 5800x, but these cores run at a lower frequency to stay within the TDP. It's not magic, it's common sense. I mean the existence of a 12900 non k @ 65w base TDP is proof of that.

They are not going full P cores cause they are worse for performance per die space. Basically, for the same size of a P core, they can extract more performance by using 4 E cores. It's as simple as that. In the future, when for example games start utilizing 12 cores, Intel will increase the P core count. But for now, there is no application out there that would run faster on a fully P cored CPU compared to a P + E core one.

 

[Troezar]

That's a misconception based on flawed logic. Power consumption is a total non issue. They can put 32 full cores and it wouldn't change a thing. I have no idea why people keep propagating this idea, it's a bunch of horse manure. You do realize cpu's have power limits? You can dictate exactly how much power you want them to consume. So they can have a 32 full power cored CPU drawing 100 watts if they want to. I mean, it's the exact same thing AMD is doing. The 5950x has more cores than the 5800x, but these cores run at a lower frequency to stay within the TDP. It's not magic, it's common sense. I mean the existence of a 12900 non k @ 65w base TDP is proof of that.

They are not going full P cores cause they are worse for performance per die space. Basically, for the same size of a P core, they can extract more performance by using 4 E cores. It's as simple as that. In the future, when for example games start utilizing 12 cores, Intel will increase the P core count. But for now, there is no application out there that would run faster on a fully P cored CPU compared to a P + E core one.

So what you're saying is the P cores and E cores are exactly the same just the E cores are clocked lower as per the AMD example? As I understood it the AMD cores were all the same just the high core count chips were lower clocked but still "full fat" cores?

 

[Bencher]

So what you're saying is the P cores and E cores are exactly the same just the E cores are clocked lower as per the AMD example? As I understood it the AMD cores were all the same just the high core count chips were lower clocked but still "full fat" cores?

No, that's not what I said, i've no idea where that came from


What I'm saying is, if they went for 16 P cores, they don't have to run them at 5ghz like they run on the 12900k, exactly like AMD is doing. Higher core parts run at lower frequencies to stay within the TDP. So power consumption is not what's stopping Intel from going with 16 / 32/ 48 or whatever number of P cores. That's the least of their concern

 

[Journey]

But for now, there is no application out there that would run faster on a fully P cored CPU compared to a P + E core one.

Why make stuff up? That sentence is a load of codswallop, you have literally no empirical data to support that conclusion, just pure guessing. You like to complain about other people, but you do the same.

 

[Bencher]

Why make stuff up? That sentence is a load of codswallop, you have literally no empirical data to support that conclusion, just pure guessing. You like to complain about other people, but you do the same.

Can you name an application?

 

[Troezar]

Interesting, that there are no applications that can take advantage of more than 8 full cores as that would exceed the power budget of the current technology. May as well stick with 8 cores then

 

[Armageus]

What I'm saying is, if they went for 16 P cores, they don't have to run them at 5ghz like they run on the 12900k, exactly like AMD is doing. Higher core parts run at lower frequencies to stay within the TDP. So power consumption is not what's stopping Intel from going with 16 / 32/ 48 or whatever number of P cores. That's the least of their concern

Eh? So what is stopping them from making "real" 16/24/32/48/64 core parts?

Also if E cores are "just as good", then why aren't there any E-core only parts? You can get 4 E cores in the same space as a P Core, so surely we could have had a 0P 40E core part by now?
Instead of 38 or 40 Core Xeon parts, we could have had 152 or 160 E Core Xeon's which the server market would be lapping up?