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.

 

[TheELF]

Wow. Is Intel prioritizing MT over ST for Raptor Lake? I hope that's not true for the other SKUs, especially 13100,13400,13500 and 13600.

The 13900 will have another 8 e-cores so that's why it will be better in MT, ST will probably only change as much as clocks will increase which shouldn't be much.

Dunno, seems strange they are regressing in ST performance in this fixed comparison at static 3.8ghz
Maybe Raptor Lake took a large latency hit from doubling the L2 cache (?)
..Or its the extra stops on the ringbus that's hurting gaming performance ?

You can't keep everything else exactly the same no matter how much you try.
So this is probably just normal variance.

 

[igor_kavinski]

Maybe Raptor Lake took a large latency hit from doubling the L2 cache (?)

My bet is on that. If true, Intel seems to have run out of tricks for this generation. Things are pointing to Zen 4 embarrassing Intel pretty bad, especially in gaming workloads.

 

[pakotlar]

Ya’ll:

“””
Do note that the P-Cores for this engineering sample ran at a base clock of 1.4 GHz and a max boost clock of 3.8 GHz (P-Cores) while the E-Cores were split with 2 clusters running at 2.8 GHz and the other 2 clusters running at 1.0 GHz so we can already see some inconsistency in the clock speeds.
“””

its base clocks are <50% retail, which will also mean >2x latency for the caches. From this raptor lake is very promising.

 

[CakeMonster]

I don't count on this to be final at all, but with pretty much every generation you have to lay down some groundwork for the future. The additional E-cores is certainly part of that, maybe along with logic and design for handling more threads properly, like added cache. Its easy to just complain and say they could have done thing differently, but that may not have been an option.

 

[nicalandia]

Do note that the P-Cores for this engineering sample ran at a base clock of 1.4 GHz and a max boost clock of 3.8 GHz (P-Cores) while the E-Cores were split with 2 clusters running at 2.8 GHz and the other 2 clusters running at 1.0 GHz so we can already see some inconsistency in the clock speeds.

The P Cores were set at 3.8 Ghz so it's a valid test for single core IPC performance in CBR23 and at gaming since none of those use the e-cores..

 

[igor_kavinski]

Raptor Lake might also be the last architecture they had finalized before Keller joined Intel. Maybe the future is brighter if Intel can execute properly.

 

[nicalandia]

Raptor Lake might also be the last architecture they had finalized before Keller joined Intel. Maybe the future is brighter if Intel can execute properly.

Alder Lake/Raptor use Golden/Raptor Cove uArch built on Intel 7, now what is next to that? a Die Shrink to Intel 4 and Redwood Cove which so far has look nearly identical to it's larger process brother.

 

[pakotlar]

The P Cores were set at 3.8 Ghz so it's a valid test for single core IPC performance in CBR23 and at gaming since none of those use the e-cores..

They were not
Their max boost was capped to 3.8ghs, 1.4ghz below alder lake.
Read the articles.

 

[nicalandia]

They were not
Their max boost was capped to 3.8ghs, 1.4ghz below alder lake.
Read the articles.

Both Chips the 13900 ES and The 12900K were clocked at ISO speed of 3.8 Ghz Fixed frequencies. Stop making things up to make Raptor Look better. The numbers don't lie

 

[dullard]

They were not
Their max boost was capped to 3.8ghs, 1.4ghz below alder lake.
Read the articles.

This is supposedly a 13900 chip compared to an underclocked 12900K chip. The underclocking is an attempt to put them at roughly the same frequency to get an idea of how well each chip performs independent of frequency. That is, do (A) the changes to the core fundamentally impact performance or (B) the changes in performance come all from number of cores and ultimate final production frequency of the cores. Underclocking isn't a perfect comparison, but it does show that option (A) seems to be the correct answer. Meaning the added cache to the Raptor Cove core does have a significant impact on performance.

And note: be careful when you say 1.4 GHz below Alder Lake, because then you would be comparing a 65 W chip (13900) to a 150 W chip (12900KS). A better comparison would be to the 65 W 12900 which is only 1.2 GHz above this engineering sample 13900. Not 1.4 GHz.

 

[pakotlar]

“This alleged i9-13900 processor has preliminary clock speeds that will almost certainly be higher as a retail version. In this case the P-cores are running at 1.4 GHz to 3.8 GHz range, while one cluster of E-Cores runs at 1.0 GHz and the other at 2.0 GHz. There is therefore some inconsistency in frequencies, as noted by Wccftech.”
- https://videocardz.com/newz/intel-r...aster-than-alder-lake-in-multi-threaded-tests


“The 65W TDP indicates that we are looking at a non-K part, most likely Core i9-13900. The clock speeds are not high though. Because this is an engineering sample, the maximum boost clock is set to 3.8 GHz, significantly lower than i9-12900 (with max Turbo of 5.1 GHz).”
- https://videocardz.com/newz/alleged...engineering-sample-spotted-with-3-8-ghz-clock


“In terms of clock speeds, the Intel Core i9-13900 Raptor Lake CPU has a clock speed that's rated at 3.8 GHz. Do note that the P-Cores for this engineering sample ran at a base clock of 1.4 GHz and a max boost clock of 3.8 GHz (P-Cores) while the E-Cores were split with 2 clusters running at 2.8 GHz and the other 2 clusters running at 1.0 GHz so we can already see some inconsistency in the clock speeds. This is a Non-K 65W part so it will carry a much lower clock speed plus it is also an ES chip (engineering sample) so clocks are expected to be lower. The final chips are expected to yield clock speeds of up to 6 GHz.”
- https://wccftech.com/intel-raptor-l...aster-than-core-i9-12900k-in-multi-threading/

We’re all tempted to skip reading the articles, it is the internet after all. You’re wrong on this one, no big deal. They’re comparing a 65W severely underclocked engineering part to retail 12900K. The fact that it keeps up is amazing.

 

[igor_kavinski]

The fact that it keeps up is amazing.

OK, I'll give you that.

 

[nicalandia]

We’re all tempted to skip reading the articles, it is the internet after all. You’re wrong on this one, no big deal. They’re comparing a 65W severely underclocked engineering part to retail 12900K. The fact that it keeps up is amazing.

You are Delusional if you believe that this is nothing but an attempt to measure IPC at ISO Speed. yeah sure a 3.8 Ghz 13900 65W ES CPU is nearly 30% fasteR in MT than a stock 12900K

 

[dullard]

We’re all tempted to skip reading the articles, it is the internet after all. You’re wrong on this one, no big deal. They’re comparing a 65W severely underclocked engineering part to retail 12900K. The fact that it keeps up is amazing.

Psst: both were running lower than final clocks. The Raptor Lake chip was running at lower frequencies since Raptor Lake isn't finished yet. The Alder Lake was also running at lower frequencies in order to be a better comparison. Yes, Raptor Lake will be at higher frequencies when released, we know that. But we also know that we can easily do the math to scale from 3.8 GHz to the final Raptor Lake speed.

To help you understand the article, I circled a few things on this photo. Notice how the Alder Lake is running at lower voltages and thus lower frequencies. The test was not at 5.2 GHz.

 

[inf64]

They seem to have locked both chips to 3.8Ghz to get the apples-to-apples comparison. This is expected, it's like Ice Lake -> Tiger Lake upgrade. The max clock speeds should be higher with Raptor Lake but per clock it's basically performing almost the same (P cores), at least in this "review".

 

[pakotlar]

Psst: both were running lower than final clocks. The Raptor Lake chip was running at lower frequencies since Raptor Lake isn't finished yet. The Alder Lake was also running at lower frequencies in order to be a better comparison. Yes, Raptor Lake will be at higher frequencies when released, we know that. But we also know that we can easily do the math to scale from 3.8 GHz to the final Raptor Lake speed.

To help you understand the article, I circled a few things on this photo. Notice how the Alder Lake is running at lower voltages and thus lower frequencies. The test was not at 5.2 GHz.
View attachment 63520

Ah you’re right that they locked the alder lake to 3.8. Note that the TDP on the alder lake remains at 125W, while raptor lake is 65W.

I don’t see where they said that the Raptor Lake was locked to 3.8, important bit bolded:

“In terms of clock speeds, the Intel Core i9-13900 Raptor Lake CPU has a clock speed that's rated at 3.8 GHz. Do note that the P-Cores for this engineering sample ran at a base clock of 1.4 GHz and a max boost clock of 3.8 GHz (P-Cores) while the E-Cores were split with 2 clusters running at 2.8 GHz and the other 2 clusters running at 1.0 GHz so we can already see some inconsistency in the clock speeds. This is a Non-K 65W part…”

edit: the 2-cluster clock domains are encouraging, as e-cores power consumption scales very poorly with clock speed.

 

[pakotlar]

You are Delusional if you believe that this is nothing but an attempt to measure IPC at ISO Speed. yeah sure a 3.8 Ghz 13900 65W ES CPU is nearly 30% fasteR in MT than a stock 12900K

Watch it with the ad hominem.

 

[igor_kavinski]

the 2-cluster clock domains are encouraging, as e-cores power consumption scales very poorly with clock speed.

Should be great news on the power consumption front. Maybe now the P-cores can get the power they need to boost higher without the E-cores being hungry power parasites.

 

[IntelUser2000]

They seem to have locked both chips to 3.8Ghz to get the apples-to-apples comparison. This is expected, it's like Ice Lake -> Tiger Lake upgrade. The max clock speeds should be higher with Raptor Lake but per clock it's basically performing almost the same (P cores), at least in this "review".

It is an ES sample after all. Who knows maybe the memory performance is low. I'm tired of performance leaks. I want performance benchmarks, which means reviews on a production chip.

Should be great news on the power consumption front. Maybe now the P-cores can get the power they need to boost higher without the E-cores being hungry power parasites.

What? P cores can take up 250W all by itself while the E cores take 50W at most. I think you should slow down and give it a thought before you post sometimes. How is reducing 50W over 250W a logical thing to do?

 

[inf64]

It is an ES sample after all. Who knows maybe the memory performance is low. I'm tired of performance leaks. I want performance benchmarks, which means reviews on a production chip.

Yeah, I agree with you. We need to wait for a final product to be tested, these "leaks" are based on ES with unknown firmware/drivers.

 

[dullard]

It is an ES sample after all. Who knows maybe the memory performance is low.

It was using DDR5-5200. Raptor Lake is rumored to support DDR5-5600 at stock. Although at these lowered engineering sample clock speeds that might not matter.

 

[IntelUser2000]

Yeah, I agree with you. We need to wait for a final product to be tested, these "leaks" are based on ES with unknown firmware/drivers.

This goes same for Zen 4. Pretty much all leaks were wrong on that one.

I've seen it before. Some reliable leakers pop up so they eventually get misdirected or even hired altogether. Remember when Intel hired Ashraf?

It was using DDR5-5200. Raptor Lake is rumored support DDR5-5600 at stock. Although at these lowered engineering sample clock speeds that might not matter.

It doesn't matter. It's an ES sample with premature firmware(on the CPU) and BIOS/UEFI. Even the hardware may be immature with certain sections not performing optimally.

Regardless without the final chip we'll be arguing endlessly.

 

[dullard]

Regardless without the final chip we'll be arguing endlessly.

To me, without the final price we'll be nowhere. If this thing (13900) launches at or near Alder Lake (12900) prices it will be completely different than if it is a $999 chip.

New CPU lines launch every ~6 to ~18 months. This at least gives us something to talk about in the mean time. The final tweaks will gain a percent here or a percent there. That will make the gaming roughly even and the multi-threaded tasks in the lower 20% range better than Alder Lake. Depending on the rumor you believe, then add in 2% to 13% more gains for final clocks. That won't make Raptor Lake a worthy upgrade for gamers or average users. But, it will be a compelling argument for heavy multi-threaded users.

 

[nicalandia]

To me, without the final price we'll be nowhere. If this thing launches at or near Alder Lake prices it will be completely different than if it is a $999 chip.

The 13900K is a High End Tier CPU and it will be priced accordingly which is about the same as the 7950X

 

[dullard]

The 13900K is a High End Tier CPU and it will be priced accordingly which is about the same as the 7950X

This rumor was a 13900, not a 13900K.