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.

 

[DrMrLordX]

Imagine 4 13900K's on one substrate

That would be one messy interconnect. Plus the scheduler would have fits.

Yes, 7950x is faster at the same wattage, it just won't go over 230 for any reason (all-core)

It will with PBO. But not stock.

 

[Kocicak]

View attachment 69285

Well, the X axis datapoints should not be the same distance apart, have you used XY type of graph?

 

[Det0x]

Yes, 7950x is faster at the same wattage, it just won't go over 230 for any reason (all-core)

Not with PBO no, but with some automatic overclocking features in the asus bios, or/or manual tweaking i'm hitting ~42k @ ~260w

 

[coercitiv]

not as bad a performance from Raptor as I had expected given all of the doom and gloom I've been reading

Pick a performance target for Raptor, let's say the 230W you consider still ok for daily use. Then use your charts to calculate equivalent Zen 4 power for ISO performance, which falls around 130W. That's a +100W delta to do the job Raptor was built to excel at.

 

[Kocicak]

I am afraid that the efficiency is a secondary goal for Intel, the primary is to SOMEHOW get to the level of performance of the AMD counterpart, and they managed it. I am getting 35500 score at 195W well manageable by a small Freezer 34 air cooler.

I would say they got in the same performance region at an acceptable power draw, and they have an advantage of these CPUs being drop-in replacements in existing motherboards, and they work with DDR4 as well.

One huge problem is that these CPUs can be power draw unrestricted on some motherboards as a default setting, and CPU drawing 350W out of the box without warning in simply INSANE.

If I needed that performance, it is a pretty reasonable option. I think I will end up with 13600K, I am not sure yet. I hope the reviewers are going to test these new CPUs with DDR4 too!

These lower end models will be very competitive against AMD due to the large number of cores.

 

[Det0x]

Don't like this character but apparently he have a sneakpeek of the 13900k out. Haven't watched the video myself

 

[DrMrLordX]

Don't like this character but apparently he have a sneakpeek of the 13900k out. Haven't watched the video myself

He disables the e-cores and runs some arbitrary gaming benchmarks after overclocking both CPUs and tuning DDR5 for both.

 

[Kocicak]

.... Haven't watched the video myself

He disables the e-cores and runs some arbitrary gaming benchmarks after overclocking both CPUs and tuning DDR5 for both.

Perhaps it is better to watch the video before you recommend it to other people??? I would not promote such ****. Disabling one and changing other of the main defining characteristics which differ one product from the other product and then claiming that the products are similar? Really?

 

[nicalandia]

Intel 13400 Review.. Alder Lake Refresh..

Intel 13th Gen Core AlderLake-Refresh (i5 13400) Review & Benchmarks – Value Hybrid Efficiency – SiSoftware

www.sisoftware.co.uk

 

[poke01]

Intel 13400 Review.. Alder Lake Refresh..

Intel 13th Gen Core AlderLake-Refresh (i5 13400) Review & Benchmarks – Value Hybrid Efficiency – SiSoftware

www.sisoftware.co.uk

View attachment 69307

View attachment 69308

At least we get 4 more e cores and a bit of extra cache

 

[igor_kavinski]

View attachment 69308

Why is the inter-thread latency so much higher between the cores in the E-core cluster?

 

[DrMrLordX]

Perhaps it is better to watch the video before you recommend it to other people???

I didn't recommend it to others. Perhaps you were talking to the other guy . . .

Disabling one and changing other of the main defining characteristics which differ one product from the other product and then claiming that the products are similar? Really?

Some gamers run their Alder Lake-S with e-cores disabled. The idea behind the video was showing whether a tuned Raptor Lake will be any better than a tuned Alder Lake in games in such a setup. That being said, a few games (Spiderman) can make use of extra cores, so it's not always wise to disable the e-cores.

 

[igor_kavinski]

That being said, a few games (Spiderman) can make use of extra cores, so it's not always wise to disable the e-cores.

But is there evidence that allowing the E-cores to host gaming threads does not result in hitching or stuttering?

 

[inf64]

Hmm not much if a difference in games according to the video sneak preview (12900K vs 13900K). Only 3 more days until the official NDA liftoff.

 

[Gideon]

EDIT:

Sorry reycling the very content of this thread

 

[inf64]

13900K(F) Sisoft sandra (pre)view: https://www.sisoftware.co.uk/2022/10/15/intel-13th-gen-core-raptorlake-i9-13900ks/

Edit: one odd thing is they didn't show 7950X in the chart, only the 7900X. I guess they went for similar price range but 24T vs 32T is not really comparable (even then, in Sandra test suit, 7900X is basically neck even overall).

 

[Abwx]

13900K(F) Sisoft sandra (pre)view: https://www.sisoftware.co.uk/2022/10/15/intel-13th-gen-core-raptorlake-i9-13900ks/

Edit: one odd thing is they didn't show 7950X in the chart, only the 7900X. I guess they went for similar price range but 24T vs 32T is not really comparable (even then, in Sandra test suit, 7900X is basically neck even overall).

View attachment 69310

Their efficency estimation is a joke, they use 125W as basis for the 13900K and rate their perf/watt benches according to this power, this way they inflate RPL perf/watt by a little 100%...

 

[nicalandia]

Why is the inter-thread latency so much higher between the cores in the E-core cluster?

Because each core on the e cluster has to do a round trip(go out to L3/Ring) to comunicate to each other. Ch&Ch explained that better.

 

[igor_kavinski]

Because each core on the e cluster has to do a round trip(go out to L3/Ring) to comunicate to each other.

That's dumb. Maybe in future, they will integrate a cluster of two E-cores with each P-core to reduce the latency impact. The current solution seems like it was cooked up in a hurry to compete with the increasing Zen core counts.

 

[DrMrLordX]

But is there evidence that allowing the E-cores to host gaming threads does not result in hitching or stuttering?

Depends on the game. Spiderman uses the cores for realtime shader compilation. Other games shouldn't be sending threads to the e-cores unless the p-cores are saturated. Shouldn't. There are probably enough instances of that happening that many gamers choose to disable them as a precaution.

 

[Kocicak]

I didn't recommend it to others. Perhaps you were talking to the other guy . . .

Yes, I was, not sure why you reacted when you did not recommend it...

Why are they (Sisoftware) only comparing to Ryzens 5900X and 7900X? Where are the .950X models?

Their efficency estimation is a joke, they use 125W as basis for the 13900K and rate their perf/watt benches according to this power, this way they inflate RPL perf/watt by a little 100%...

Well, if they used a number from Intel spec sheet and this number is the same for both Intel processors but it reflects real world CPU performance differently for each processor and thus the comparison makes little sense, it is really on Intel, not them...

Making some comparisons using numbers from spec sheets from two different companies with two different methodologies makes no sense at all, of course.

 

[Hulk]

Pick a performance target for Raptor, let's say the 230W you consider still ok for daily use. Then use your charts to calculate equivalent Zen 4 power for ISO performance, which falls around 130W. That's a +100W delta to do the job Raptor was built to excel at.

True but if you pick 170W then it's not nearly as bad. But I'm not going to argue. Zen 4 is much more efficient. There is no doubt.

 

[nicalandia]

Why is the inter-thread latency so much higher between the cores in the E-core cluster?

Also of Note.. That issue has been mostly addressed by True Raptor Lake(not the refreshed ADL), by giving the e cores Twice as much L2 Cache and faster Ring and Cores

This is the one for 13900K

 

[Abwx]

True but if you pick 170W then it's not nearly as bad. But I'm not going to argue. Zen 4 is much more efficient. There is no doubt.

It require 170-180W for the 13900K to perform like a 7950X@100W.

If we correct Sisoftware numbers by the actual TDP then the 7950X should be roughly 80% more efficient than the 13900K wich is itself 22% more efficient than the 12900K.

Well, if they used a number from Intel spec sheet and this number is the same for both Intel processors but it reflects real world CPU performance differently for each processor and thus the comparison makes little sense, it is really on Intel, not them...

Making some comparisons using numbers from spec sheets from two different companies with two different methodologies makes no sense at all, of course.

For this "estimation" to be right AMD should had stated something like 65W TDP on their specs sheet...

 

[nicalandia]

13600K Performance numbers just popping up.

Details for Result 13th Gen Intel Core i5-13600K 6C + 8c 20T 5.09Ghz + 3.89GHz, 4.5Ghz 6x 2MB + 2x 4MB L2, 24 MB L3

Details for Result ID 13th Gen Intel(R) Core(TM) i5-13600K (6C + 8c 20T 5.09千兆赫（GHz） + 3.89千兆赫（GHz）, 4.5千兆赫（GHz） IMC, 6x 2兆字节（MB） + 2x 4兆字节（MB） L2, 24兆字节（MB） L3)

ranker.sisoftware.co.uk

Compared to 7700X

Details for Result ID AMD Ryzen 7 7700X 8-Core Processor (8C 16T 5.54GHz, 1.33GHz IMC, 8x 1MB L2, 32MB L3)

ranker.sisoftware.co.uk