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.

 

[nicalandia]

Only the 13900K and 13900KS will offer the highest single thread burst frequencies.

Those and the 7950X will reach nearly 6 Ghz at stock.

 

[igor_kavinski]

Only the 13900K and 13900KS will offer the highest single thread burst frequencies.

Currently I'm running a Gskill DDR4 kit at 5866Mhz CL21 daily and my work apps really LOVE high-bandwidth mem kits.

Are those custom apps or anything we would know about?

 

[AlltheWay LeeWhy]

Are those custom apps or anything we would know about?

Two statistical scientific apps, Scrivener writing, Final Draft 12 screenplay, Microsoft Office 2021 - Word and Excel, common applications...

My PC is basically a Typewriter Build. A water-cooled Typewriter Build.

 

[igor_kavinski]

Microsoft Office 2021 - Word and Excel, common applications...

Yeah, some calculations in Excel need serious horsepower. Do check out Librecalc 7.4 Beta 1. If your Excel workloads have countifs() and related formulas, you may experience a speedup of 10x with that version of Librecalc. At least, I did on my Haswell work PC.

 

[AlltheWay LeeWhy]

And you can only imagine which keyboard design works best with a Typewriter Build.

It's got to offer very comfortable switches able to use over long hours with very slight actuating feedback points.

And those keyboards are never cheap.

 

[Markfw]

The 13900KS will be my 1st 8core processor and I've already made the decision not to overclock it. It's a work computer that doesn't rely on sustained CPU loading, only need quick bursts of ST performance while working. Coupled with a new PCIe 5.0 NVMe SSD and the fastest Gskill Hynix DDR5 kit available mounted to the ROG Z790 Apex 2dimm board, I should do ok.

Only the 13900K and 13900KS will offer the highest single thread burst frequencies.

Currently I'm running a Gskill DDR4 kit at 5866Mhz CL21 daily and my work apps really LOVE high-bandwidth mem kits.

The 13900k and ks do not exist yet. By the time they do Zen4 may be out. I would not bet the farm yet,

 

[Hulk]

After using my 12700K for 9 months I understand the Big.Little rationale a little better. I have come to think of it as foreground.background. Even if you have an extraordinary amount of compute cores available you may have a "ridiculously parallel" application running, which could slow the progress of the foreground application. While of course it is useful for all compute work to be accomplished as quickly as possible, the priority is for the foreground application to respond to the user immediately.

The Big.Little concept allocates most of the die space to the P cores, which operate at high speeds and high power to keep the user experience "real time" with the foreground app. In order to keep things moving along with the background processes Intel decided to use the remaining die space for Gracemont cores, which of course were designed for area efficiency, first and foremost.

Intel made the starting point assumption that 8 Big cores is the number required to keep the foreground application responsive. It's a pretty good assumption in my opinion as 8 Golden Coves do a pretty good job at keeping Photoshop, DAW's, games, and video editing apps moving along. Yes, I realize GPU is important for video work but many NLE's still require quite a bit of CPU compute.

Next they utilized the rest of the die area with the most area efficient multiprocessing cores they could design... Gracemont. With Raptor Lake the major improvement is to provide a boost to those background processes with an additional 8 Gracemont cores.

Obviously not the ideal compute distribution for everyone but a good fit for some.

 

[ZGR]

Alder Lake is quite efficient once underclocked and undervolted; so I expect Raptor Lake to be as well.

There is a good Der8auer video of undervolted 5800X3D vs undervolted Alder Lake and both performed fantastically.

Once retail samples are out, I wanna see if voltages for clock speeds under 5Ghz are lower overall vs AL.

The other thing I’m interested in is Raptor Lake’s IMC vs Zen 4’s for high speed, low latency DDR5.

Skylake had some lackluster reviews in certain
games until it was paired with decent DDR4.

 

[Hulk]

I'm really curious about how many Gracemont cores will be present as you move up the Raptor Lake stack.
Will the 13700K have 8 Gracemont cores (probably) or 12 (doubtful but would be great)?

 

[Markfw]

Alder Lake is quite efficient once underclocked and undervolted; so I expect Raptor Lake to be as well.

There is a good Der8auer video of undervolted 5800X3D vs undervolted Alder Lake and both performed fantastically.

Once retail samples are out, I wanna see if voltages for clock speeds under 5Ghz are lower overall vs AL.

The other thing I’m interested in is Raptor Lake’s IMC vs Zen 4’s for high speed, low latency DDR5.

Skylake had some lackluster reviews in certain
games until it was paired with decent DDR4.

First, why did Intel not do this ? why must users (and OEM can not) Fix this ?

 

[ZGR]

First, why did Intel not do this ? why must users (and OEM can not) Fix this ?

The same thing sorta happens on Ryzen, but far less severe. I can give my CPU a -25mV per core undervolt and temperatures drop significantly. I think Intel is pushing their silicon harder to remain competitive.

 

[DrMrLordX]

You guys do know the ES Raptorlake is 20% faster in MT despite being 10% slower in ST?

It ought to be, considering how many more cores it has. As far as the voltage planes go, I was personally hoping Intel would give Raptor Lake-s separate voltage planes for Gracemont clusters so it would get that extra efficiency. But others have told me, "oh no, it will only have that on mobile", so we will see.

Regardless losing ST performance moving from Golden Cove to Raptor Cove does not seem like a good sign. Again it's just an ES chip. We can't draw too many conclusions from it.

 

[TheELF]

First, why did Intel not do this ? why must users (and OEM can not) Fix this ?

I think Intel is pushing their silicon harder to remain competitive.

Yes, intel should have enforced a max TDP of 160W, maybe up to 190W and punish any reviewer that used unlimited TDP MCE or anything other that would increase power draw.

Messungen: Leistungsaufnahme, Temperatur und Energieeffizienz - Seite 9 - Hardwareluxx

Core i9-12900K und Core i5-12600K: Hybrid-Desktop-CPUs Alder Lake im Test.

www.hardwareluxx.de

 

[Markfw]

The same thing sorta happens on Ryzen, but far less severe. I can give my CPU a -25mV per core undervolt and temperatures drop significantly. I think Intel is pushing their silicon harder to remain competitive.

Mine are all stock and work just fine. And take 142 watts for a LOT of computing power, way more than my 12700F at less wattage.

 

[Markfw]

Yes, intel should have enforced a max TDP of 160W, maybe up to 190W and punish any reviewer that used unlimited TDP MCE or anything other that would increase power draw.

Messungen: Leistungsaufnahme, Temperatur und Energieeffizienz - Seite 9 - Hardwareluxx

Core i9-12900K und Core i5-12600K: Hybrid-Desktop-CPUs Alder Lake im Test.

www.hardwareluxx.de

Believe it or not, I agree with you. If Intel had maxed at 160 watt, its more than 5950x, but WAY less power/heat.

But in the total benchmark suite it could have hurt them, but it would have been the sensible thing to do.

 

[nicalandia]

First, why did Intel not do this ? why must users (and OEM can not) Fix this ?

Undervolted 12900K will not be competitive with 5950X..

 

[Markfw]

Undervolted 12900K will not be competitive with 5950X..

My point exactly. They had to jack the power to infinity to beat AMD. It was stupid. I want to see some sanity in Raptor lake.

 

[nicalandia]

I want to see some sanity in Raptor lake.

Well, they gone and added 8 more cores on the same process node and.. Jacked up the speed to Nearly 6 Ghz..

 

[JasonLD]

My point exactly. They had to jack the power to infinity to beat AMD. It was stupid. I want to see some sanity in Raptor lake.

As long as I can control the voltage and max TDP, I don't see why that should be a big issue though.

 

[Markfw]

As long as I can control the voltage and max TDP, I don't see why that should be a big issue though.

Well, maybe for you. What about the millions that will buy an OEM system, and have no idea about any of this. If the product is not "out of the box great", then it has a problem. Tweaking on AMD can achieve nice results, but out of the box they are great. For Intel to suceed, they need an "out of the box" success.

 

[Exist50]

All this complaining about high TDPs for the enthusiast products seems rather silly. Intel used to enforce caps, but many motherboard vendors just went ahead and removed them anyway (i.e. Multi Core Enhancement). Clearly the market is fine with that. If for whatever reason you want a lower limit, it's trivial to apply.

 

[JasonLD]

Well, maybe for you. What about the millions that will buy an OEM system, and have no idea about any of this. If the product is not "out of the box great", then it has a problem. Tweaking on AMD can achieve nice results, but out of the box they are great. For Intel to suceed, they need an "out of the box" success.

Aren't they more limited on OEM systems anyways?

 

[Exist50]

Aren't they more limited on OEM systems anyways?

They are. The locked chips don't have unlimited boosts by default, and the PL2 is typically lower as well. But that's clearly inconvenient for some to acknowledge...

 

[Markfw]

They are. The locked chips don't have unlimited boosts by default, and the PL2 is typically lower as well. But that's clearly inconvenient for some to acknowledge...

I have a 12700F. Without modifying the bios, it draws 200-300 watts, CPU only from the wall. If you have proof that MOST OEMS lower some of these power specifications, please provide them. All the reviews said they had the high watt limits. I will believe them before you.

Also, is all you have to do is apologize and make excuses for Intel ?

 

[Exist50]

I have a 12700F. Without modifying the bios, it draws 200-300 watts, CPU only from the wall. If you have proof that MOST OEMS lower some of these power specifications, please provide them. All the reviews said they had the high watt limits. I will believe them before you.

Also, is all you have to do is apologize and make excuses for Intel ?

The 12700 has a PL2 duration of 28 seconds and a PL2 of 180W. These details are available with an absolutely trivial amount of research.

Intel Core i7-12700F Specs

Alder Lake-S, 12 Cores, 20 Threads, 2.1 GHz, 65 W

www.techpowerup.com

Intel® Core™ i7-12700F Processor (25M Cache, up to 4.90 GHz) Product Specifications

Intel® Core™ i7-12700F Processor (25M Cache, up to 4.90 GHz) quick reference guide including specifications, features, pricing, compatibility, design documentation, ordering codes, spec codes and more.

ark.intel.com

And why should anyone believe your claims when you make ridiculous assertions like Ryzen being 2x the performance at half the power? Why don't you provide an independent source for once?