Question Raptor Lake - Official Thread

[Hulk]

Since we already have the first Raptor Lake leak I'm thinking it should have it's own thread.
What do we know so far?
From Anandtech's Intel Process Roadmap articles from July:

Built on Intel 7 with upgraded FinFET
10-15% PPW (performance-per-watt)
Last non-tiled consumer CPU as Meteor Lake will be tiled

I'm guessing this will be a minor update to ADL with just a few microarchitecture changes to the cores. The larger change will be the new process refinement allowing 8+16 at the top of the stack.

Will it work with current z690 motherboards? If yes then that could be a major selling point for people to move to ADL rather than wait.

 

[poke01]

Your physics is incorrect here, in your example above the i9 will output 250w per second vs 230w for Zen4. Power(W) = heat output per second (Jules per second). The temperature of the heat source doesn't matter. The total heat output is what matter to increasing the room temperature. And secondly, the 95C is the hot-spot temperature of the die. The temperature of the heat-sink will be much lower.

okay thanks for correcting me.

 

[poke01]

Because he use an AIO, , with an air cooler it would be at 130°C at this power...

Edit : At 250W it will be hotter than a stock 7950X when using the same cooler...

Apparently a poster here gets 67C at 250 watts with an AIO. (i9 13900k)

 

[poke01]

Nobody have asked but my PBO enabled Zen4 is getting these Cinebench R23 MT numbers at the following packet power limits: (PPT)

38.7k points @ 122w
33.5k points @ 80w
24.8k points @ 50w
18.5k points @ 35w

yeah thats impressive. Zen 4 is efficient for sure. Can you please tell temps? 85C?

 

[Det0x]

Are these watts real power drawn by the CPU? If yes, these numbers would be very impressive. Is the CPU tweaked somehow - undervolted?

Just PBO CO enabled, i shared results here in the "PES-assessing-power-and-performance-efficiency-of-x86-cpu-architectures" thread.

yeah thats impressive. Zen 4 is efficient for sure. Can you please tell temps? 85C?

At these lower power figures with my large custom watercooling i think i was well below 50 degrees.

 

[PJVol]

I don't think we really understand the true percentage of the input ppwer is dumped as waste heat for either product.

Hmm... There's some form of output energy other than heat? I thought power loss of CPUs is ~ 99.9999....%

 

[Kocicak]

Actually there may be a direct physical link between change of information happening in the processor (being linked to change of entropy) and energy, but it is safe to presume that almost all the energy consumed by the CPU gets converted to heat.

 

[leoneazzurro]

Another part is electromagnetic emissions, but compared to the heat produced it is normally negligible.

 

[moinmoin]

They are using the die space of 10P's for 8+16 but getting a more competitive chip.

The die space of 12P's actually, 10P's was for 8+8.

 

[LightningZ71]

Hmm... There's some form of output energy other than heat? I thought power loss of CPUs is ~ 99.9999....%

A non-zero percentage of the input power onto the processor "system" is converted to useful work.the rest is dissipated primarily as heat but is also emitted in various forms of electromagnetic radiation.

 

[nicalandia]

Inter Thread Lateness have improved indeed.

Details for Result ID 13th Gen Intel(R) Core(TM) i9-13900K (8C + 16c 32T 5.49千兆赫（GHz） + 4.3千兆赫（GHz）, 4.9千兆赫（GHz）/5千兆赫（GHz） IMC, 8x 2兆字节（MB） + 4x 4兆字节（MB） L2, 36兆字节（MB） L3)

ranker.sisoftware.co.uk

13900K


For comparison a 12900K

Details for Result ID 12th Gen Intel(R) Core(TM) i9-12900K (8C + 8c 24T 5.08GHz + 3.88GHz, 3.6GHz/4.7GHz IMC, 8x 1.25MB + 2x 2MB L2, 30MB L3)

ranker.sisoftware.co.uk

 

[Asterox]

Not good at all, if this is i9+Unlimited power option.

https://twitter.com/x/status/1580903890324574209

 

[deasd]

Ok..... this is the most shocking news to me

13400/F are not fully rebranded, only part of it belongs to ADL(C0). L3 cache size are the same but L2 are likely different.
Share same name but different performance??? Or else, perf are the same dispite L2 difference?

https://twitter.com/x/status/1580916952909770753

 

[mikk]

Not good at all, if this is i9+Unlimited power option.

This is a 45% speedup over 12900k: https://www.techpowerup.com/review/amd-ryzen-9-7950x/7.html

It's a good result for Raptor Lake.

Ok..... this is the most shocking news to me

13400/F are not fully rebranded, only part of it belongs to ADL(C0). L3 cache size is the same but L2 is likely different.
Share same name but different performance??? Or else, perf are the same dispite L2 difference?

It comes from a support list. To me it makes no sense that the 13400 partially gets (the better?) B0 whereas 13600/13500 won't get.

 

[nicalandia]

Posting picture for reference

 

[nicalandia]

@deasd @mikk
@Exist50


The 13600 and below are Alder Lake in built with Golden Cove cores(1.25MiB L2$) and OG Gracemont Cores(with 2MiB L2$). You can call them Rebranded Alder Lake or Alder Lake+ Or Golden Cove+ cores.


In the Red Square GB5 is reading the Gracemong System$ subsystem

13600


vs

13600K

Z690 Phantom Gaming 4/D5 - Geekbench

Benchmark results for a Z690 Phantom Gaming 4/D5 with an Intel Core i5-13600K processor.

browser.geekbench.com

If we compare the 13600 and the 13900K Memory Subsystems reading on GB5 You will clearly notice that the 13600K is using Raptormont cores(4MiB L2$) and the 13600 is using OG Gracemont cores 2MiB L2$

 

[Shivansps]

In 3DMark? Sure. We all know how Intel likes to churn out meaningless redults in synthetics 😅

It would depend on the game, yet at 2100mhz OC the RPL igp has about 95% of the perf in W3 1080P and about 75% of 1080 Shadow of the Tomb Raider compared to a stock 5700G.

Ryzen 7000 IGP from what i saw is just a little better than a vega 3. Problem being the minimal L2 cache as well, not only the 2 CUs.

 

[jpiniero]

13400/F are not fully rebranded, only part of it belongs to ADL(C0). L3 cache size are the same but L2 are likely different.
Share same name but different performance??? Or else, perf are the same dispite L2 difference?

I imagine they will cut the L2.

 

[nicalandia]

I imagine they will cut the L2.

They are NOT Cutting Raptor Cove L2, they are using Golden Cove cores on them.

See post https://forums.anandtech.com/threads/raptor-lake-official-thread.2599551/post-40866355

 

[jpiniero]

They are NOT Cutting Raptor Cove L2, they are using Golden Cove cores on them.

See post https://forums.anandtech.com/threads/raptor-lake-official-thread.2599551/post-40866355

I'm talking about the Raptor Lake version of the 13400.

 

[nicalandia]

I'm talking about the Raptor Lake version of the 13400.

There is NO Raptor Cove version of the 13400, All CPUs below 13600K use Golden Cove Cores with 1.25 L2$.

What would be easier, Mask, Print Raptor Cove along with Raptormont so later they will just laser off 0.75MiB on each P core and then laser off 2MiB on the e cores...OR... Use existing Golden Cove/Gracemont masks??

 

[jpiniero]

There is NO Raptor Cove version of the 13400, All CPUs below 13600K use Golden Cove Cores with 1.25 L2$.

That's what the chart says. They are selling the 13400/F with both Raptor Cove and Golden Cove. I imagine the extra L2 will be disabled.

Probably not even a fuse disable. Just in microcode I bet.

 

[Kaluan]

Ugh... temperature ≠ heat
This is a similar misunderstanding of physics that got carried into the "common sense", just like how people misatribute weight to what mass is.

Not good at all, if this is i9+Unlimited power option.

https://twitter.com/x/status/1580903890324574209

If the clock data means the average between the 8 P cores and 16 E cores than it might actually be overclocked too (> (5.5x8+4,3x16)/24), the 7950X one seems about right, 5,1x GHz is the typical all core average for a stock setup.

 

[nicalandia]

That's what the chart says.

That's just a made up chart that could be inaccurate.

This is the info we get from Geekbench. Look at the OG Gracemont System Cache reading on GB

 

[jpiniero]

That's just a made up chart that could be inaccurate.

This is the info we get from Geekbench. Look at the OG Gracemont System Cache reading on GB

That's the Gracemont Cluster it's reading.

 

[nicalandia]

That's the Gracemont Cluster it's reading.

I know, I've been saying that all day long.. OG Gracemont on Raptor Lake CPU means a rebranded Alder Lake..

Raptormont Cores have 4MiB of L2