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.

 

[jpiniero]

Roman calls this launch boring. He thinks they are missing a trick by not sampling the i7.

It is boring.

 

[ondma]

Roman calls this launch boring. He thinks they are missing a trick by not sampling the i7.

Yea, but they do not want to take attention away from the high end part. 14900K is probably maxed out in terms of thermals and power usage, but they could have made the launch more interesting if they had added another E core cluster to the 14600k as well.

 

[Hulk]

My 14900K preview...

As we all suspected the Raptor Lake Refresh 14900K is basically a better binned/stepped 13900K.
Benefits as follows:

P and E core base frequencies have increased by 200MHz indicating a (minor) efficiency bump at ISO frequencies.
All-core P specified as 5600MHz vs 5400MHz for 13900K/KS. Since the 13900K/KS have been shown to do 5.5/5.6 with adequate cooling via Intel Thermal Velocity Boost it may be possible for the 14900KS to achieve 5.8GHz all-core with sufficient cooling stock w/o O/C. We should know tomorrow...

E core boost to 4.4GHz.

Per core thermal throttling control.
This is actually interesting and may be useful. If you have 6 cores that are at 75CC while two are hovering near 85C by using this function you can "hold back" the hot cores while allowing the cooler ones to run faster. As above we should know how useful this is tomorrow. Perhaps it will mitigate hotspots?

Intel Application Optimization
Me thinks Intel is tired of the complaints regarding E's being engaged when they should not be when gaming and will be actively "hard programming" thread allocation for specific games. Just a grab to win the frame rate race or a legit workaround for a struggling Thread Director? I was hoping for a smarter Thread Director but I think what we're looking at here is simply a lookup table for specific applications (games).

Intel Extreme Overclocking Utility with AI
Sounds interesting but I've found these utilities to generally be more trouble than they're worth.

Prices are held the same as 13th Generation so now you can buy something faster than the 13900KS for the price of a 13900K. 13700K adds 200MHz and an additional E core cluster while 13600K simply adds 200MHz all core frequency bump.

This is obviously a minor evolution of Raptor Lake but better than nothing if you are looking for an easy upgrade like me.

That's my take anyway. I'm looking for the following out of the reviews tomorrow:

Power comparison vs 13900K at various power levels?
Usefulness of per core thermal throttling control?
Intel Application Optimization - Does it work and on what apps?
New O/C utility useful?
Higher overclocks?

 

[Shmee]

https://www.anandtech.com/show/2109...e-i9-14900k-core-i7-14700k-and-core-i5-14600k AT has a mini preview out. Seems a bit disappointing to me as far as an upgrade from Raptor Lake, but I guess if someone is building a new LGA 1700 system, they may as well go with one of these, save great discounts on 12th and 13th gen. Hopefully all the bugs are worked out by now with the 14th gen, as far as task scheduling and such, but I fear the caveat to that is Windows 11 or a recent Linux kernel is required.

 

[JoeRambo]

https://www.anandtech.com/show/2109...e-i9-14900k-core-i7-14700k-and-core-i5-14600k AT has a mini preview out.

It contains a glaring mistake with 14700K L3 cache, they have 11 slices of 3MB, not 10, as obviuosly additional cluster of E cores is enabled there.

 

[Shmee]

It contains a glaring mistake with 14700K L3 cache, they have 11 slices of 3MB, not 10, as obviuosly additional cluster of E cores is enabled there.

Hmm, good catch, I wasn't looking really at the L3 cache count. You should let Gavin know on the front page

 

[Hitman928]

HWUB review is up. Just skimmed through so far but from what I saw, 14900k = 13900k in multi core loads and it equals 13900ks in single or lightly threaded stuff. Seems the power limit holds it to the same performance as the 13900k when things get heavy. No improvement in gaming over 13900ks. In other words, 14900k = 13900ks with 13900k power limit.

 

[igor_kavinski]

New all-core temperature record achievement!



Congrats, Intel! Love your dedication to helping people fry their food better!

 

[Hitman928]

New all-core temperature record achievement!

View attachment 87295

Congrats, Intel! Love your dedication to helping people fry their food better!

Significantly cooler at stock than the 13900k at least. Though a sample size of more than 1 is really needed for this kind of comparison.

 

[igor_kavinski]

No improvement in gaming over 13900ks.

The humble 7800X3D still a serious thorn in Intel's side.

 

[igor_kavinski]

Significantly cooler at stock than the 13900k at least.

They either made their transistors less leakier or using a better IHS.

 

[Hitman928]

They either made their transistors less leakier or using a better IHS.

IHS is the same. HWUB’s sample hit 100C out of the box so maybe sample quality variance or motherboard default settings variance. Most likely something in the testing conditions because there isn’t really any evidence in the performance or efficiency of much difference between the generations.

 

[igor_kavinski]

The IHS IS different: https://www.igorslab.de/en/raptor-l...dragonrbattle-in-the-energetic-final-stage/2/

The IHS surface is smoother.

 

[inf64]

Computerbase numbers :https://www.computerbase.de/2023-10/intel-core-i9-14900k-i7-14700k-i5-14600k-test/3/
This is basically 13900KS all over again. Slower in games than 7800X3D, slower in MT vs 7950X (vanilla) and only having ST advantage (mostly due to poor benchmark selection that inflates the difference and also 6Ghz vs 5.7Ghz max boost clocks). Arrow Lake with its 5%/15% improvement will only be a minor upgrade versus even Zen 4 parts, so I wonder how it will compete versus Zen 5.

 

[Hitman928]

The IHS IS different: https://www.igorslab.de/en/raptor-l...dragonrbattle-in-the-energetic-final-stage/2/

The IHS surface is smoother.

Nice find. So, “same” IHS but better finish to make it flatter and smaller pits.

 

[igor_kavinski]

Nice power consumption improvement over the 13900K. I have to say, this makes the 14900K bit attractive to even current 13900K users.


7950X dethroned in this test by a hair. Intel engineers must be jumping with joy.

 

[inf64]

View attachment 87296
Nice power consumption improvement over the 13900K. I have to say, this makes the 14900K bit attractive to even current 13900K users.

Look at 7950X3D's power draw, it's basically having HALF the power draw of 13900K... Crazy stuff.

 

[igor_kavinski]

Look at 7950X3D's power draw, it's basically having HALF the power draw of 13900K... Crazy stuff.

Yeah and only loses 40 secs in rendering time against the 14900K/7950X. That is pretty awesome!

 

[Hulk]

Has anybody found a review which compares 14900K performance at various power levels to the 13900K/KS? As I wrote yesterday I think that is really the only unknown metric we need to know about these parts and the only one not reported on.

All of the Raptor Lake and Raptor Lake Refresh parts are architecturally identical (except for the 14700K but you know what I mean). What the Refresh really brings is supposedly higher quality silicon. Without performance testing of this gen vs. prior at the same power levels we don't know if the silicon is really better, or by how much?

Let's face it with the 13900K and now especially the 14900K we have a chip that can reach ridiculous frequencies at the expense of ridiculous power and heat. Even 360 AIO's can't contain it, meaning exotic cooling is required to extract full performance.

So, given a cooling/power level the given user can "tolerate" how much better is the 14900K vs the 13900K? That's what I want to know. I have a feeling I won't until I buy one and test it.

So far the only good thing I see about this release it that it already drove down the 13900KS price!

 

[biostud]

Ah good old intel again, leaving AMD in the rearview mirror...

 

[inf64]

Ah good old intel again, leaving AMD in the rearview mirror...

Yes, power draw wise they are having moar powa

 

[inf64]

Found this in the comments section from one "news" website:


Insanity.

 

[Saylick]

Gahhh dayum! Those power figures. Is this thing a CPU or is it a GPU? I know perf/W doesn't matter too much in the desktop space but everything has a limit.

 

[biostud]

Gahhh dayum! Those power figures. Is this thing a CPU or is it a GPU? I know perf/W doesn't matter too much in the desktop space but everything has a limit

Uops, nm

 

[Glo.]

Gahhh dayum! Those power figures. Is this thing a CPU or is it a GPU? I know perf/W doesn't matter too much in the desktop space but everything has a limit.

It does very much matter.