Question Raptor Lake - Official Thread

Page 60 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Hulk

Diamond Member
Oct 9, 1999
4,191
1,975
136
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.
 
  • Like
Reactions: vstar

Zucker2k

Golden Member
Feb 15, 2006
1,810
1,159
136
The question, as framed, is a bit disingenuous. From a processing point of view, the e-cores are not different from the p-cores in that they don't sit idle doing nothing all day. The 12600k (because it has e-cores) is a good example to show how e-cores contribute to make what by implication would be said to be a hexacore core, to be quite competitive in an area where previous hexacores haven't done too well. It's a valid example.

The question seeks to understand the relevance of e-cores and received an appropriate answer. The e-cores have a wide variety of uses, just as any other core does. With RPL-S this is going to be even more obvious.

Yeah my bad. But the tests he showed didn't really showcase their performance.
How did you arrive at that conclusion?
 
  • Like
Reactions: Carfax83

maddie

Diamond Member
Jul 18, 2010
4,723
4,628
136
The question, as framed, is a bit disingenuous. From a processing point of view, the e-cores are not different from the p-cores in that they don't sit idle doing nothing all day. The 12600k (because it has e-cores) is a good example to show how e-cores contribute to make what by implication would be said to be a hexacore core, to be quite competitive in an area where previous hexacores haven't done too well. It's a valid example.

The question seeks to understand the relevance of e-cores and received an appropriate answer. The e-cores have a wide variety of uses, just as any other core does. With RPL-S this is going to be even more obvious.


How did you arrive at that conclusion?

I don't need to see anything beyond that part of your response because it's a false misdirection, below is the statement I was replying to.

Here I'll bold it for you so you don't miss it again.




If you look at how something like the 12600K performs in average desktop user use cases (which is to say, superbly, crushing much higher tier Zen 3 chips), what anyone 'thinks' about the e-cores kind of pales next to that reality.

Since we're going down this road again, I'll once again show what it is for a top line chip to get smacked down in typical day to day use cases :

12600K 13.5% faster than 5950X in Octane / Web browser benchmark :

View attachment 66386
Selective deceptive presentation of information tells us a lot about any presenter's intentions to mislead.

In this case we have info that can clarify this deception. TPU just happened to also test the i5-12600 CPU. Remember this, the i5 version without e-cores. The other difference is a 4.8 GHz vs a 4.9 GHz frequency.

The performance difference in the test exactly corresponds to the frequency difference, with no effect from the e-cores. We are all smart enough to see what is happening here.


12600 octane.png
 

coercitiv

Diamond Member
Jan 24, 2014
6,151
11,685
136
The 12600k (because it has e-cores) is a good example to show how e-cores contribute to make what by implication would be said to be a hexacore core, to be quite competitive in an area where previous hexacores haven't done too well. It's a valid example.
The word hexacore shouldn't have a place next to 12600K, it only reinforces the idea that E cores don't really matter (in terms of both benefits and cost). Ever since we started debating on the usefulness of E cores, their benchmark was the equivalent in P cores (area, power, performance). If the 13500 is a 6+8 chip then the benchmark for E cores should be one 8+0 chip with the same power limits, same max clocks and roughly similar die area - in other words an equivalent chip that can be obtained for the same cost, with the same IP.

@biostud 's question is easy to translate using this criteria: is the average desktop user going to get a meaningful benefit from a 6+8 config versus 8+0 config? Now we have no framing and/or personal opinion, we're looking solely at what Intel can achieve for the consumer using the same die area.
 

Furious_Styles

Senior member
Jan 17, 2019
492
228
116
The word hexacore shouldn't have a place next to 12600K, it only reinforces the idea that E cores don't really matter (in terms of both benefits and cost). Ever since we started debating on the usefulness of E cores, their benchmark was the equivalent in P cores (area, power, performance). If the 13500 is a 6+8 chip then the benchmark for E cores should be one 8+0 chip with the same power limits, same max clocks and roughly similar die area - in other words an equivalent chip that can be obtained for the same cost, with the same IP.

@biostud 's question is easy to translate using this criteria: is the average desktop user going to get a meaningful benefit from a 6+8 config versus 8+0 config? Now we have no framing and/or personal opinion, we're looking solely at what Intel can achieve for the consumer using the same die area.
That answer is very likely a no. Because the average user probably would rarely even use all 8 cores.
 
  • Like
Reactions: Rigg and Zucker2k

Zucker2k

Golden Member
Feb 15, 2006
1,810
1,159
136
Now we have no framing and/or personal opinion, we're looking solely at what Intel can achieve for the consumer using the same die area.
Intel has 8p cores already, and most importantly, even the 6p cores trades blows wiith the competition's 16c processor in the non-embarrassingly parallel workloads where the "average" user spends most of their computing time.

The argument for the e-cores is that they produce more processing throughput than the area they occupy on the silicon when compared to the p-cores. They also arguably consume comparatively less power to do so. This is where the efficiency stems from. It is both space and power. Their presence also enables the OS to assign them to background tasks while freeing up all the p-cores to handle foreground tasks.
 

ondma

Platinum Member
Mar 18, 2018
2,718
1,278
136
Selective deceptive presentation of information tells us a lot about any presenter's intentions to mislead.

In this case we have info that can clarify this deception. TPU just happened to also test the i5-12600 CPU. Remember this, the i5 version without e-cores. The other difference is a 4.8 GHz vs a 4.9 GHz frequency.

The performance difference in the test exactly corresponds to the frequency difference, with no effect from the e-cores. We are all smart enough to see what is happening here.


View attachment 66408
Be careful of who you accuse of choosing selective benchmarks. Obviously web browsing is a type of benchmark that will not benefit from the E cores, so your conclusion that E cores dont matter is suspect at best.
 
  • Like
Reactions: Zucker2k

maddie

Diamond Member
Jul 18, 2010
4,723
4,628
136
Be careful of who you accuse of choosing selective benchmarks. Obviously web browsing is a type of benchmark that will not benefit from the E cores, so your conclusion that E cores dont matter is suspect at best.
Why do you never surprise me?

If you had read this thread correctly you will see that the post I was replying to made that argument. Jumping into the middle of a much longer argument by several here, shows the same behavior I was commenting on and only reinforces my argument.

They claimed, that the e-cores were responsible for the i5-12600K beating the AMD offerings. The results showed differently, not that I think it will matter. After all, to those with an agenda, what does a little fact have to do with what they proclaim.
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
How did you arrive at that conclusion?

There are two basic ways to showcase the e-cores:

1). Compare a CPU's performance with e-cores vs. the same CPU with e-cores disabled. Or,
2). Select a single MT benchmark, start the benchmark, and then shift context to a foreground application (such as an idle web browser) to shunt your other background application to only the e-cores. This method isn't often used, but it may remove interconnect penalties from the equation and give a better representation of e-core performance.

Comparing a 12600k vs. a 5950X does not accomplish either feat.
 
  • Like
Reactions: IEC and Elfear

nicalandia

Diamond Member
Jan 10, 2019
3,330
5,281
136
Can E-cores in the same complex not communicate directly between one another? It looks like a E-cores need to traverse the entire ring bus to communicate with their closest neighbors.
That is exactly what is going on.

1661295529459.png

1661295600269.png

"What’s however a bit perplexing is that the core-to-core latencies between Gracemont cores is extremely slow, and that’s quite unintuitive as one would have expected coherency between them to be isolated purely on their local L2 cluster. Instead, what seems to be happening is that even between two cores in a cluster, requests have to travel out to the L3 ring, and come back to the very same pathway."

 
Last edited:

jpiniero

Lifer
Oct 1, 2010
14,510
5,159
136

This one Canadian retailer seems to believe in the 20ish% higher prices for Raptor Lake talk.
 

tamz_msc

Diamond Member
Jan 5, 2017
3,710
3,554
136

This one Canadian retailer seems to believe in the 20ish% higher prices for Raptor Lake talk.
These are all placeholder prices.
 

shady28

Platinum Member
Apr 11, 2004
2,520
397
126

This one Canadian retailer seems to believe in the 20ish% higher prices for Raptor Lake talk.

That's not +20%. Gen 12 is discounted from MSRP, 13 won't be at launch of course.

MSRP on a 12600K is $320 USD. $355 is only +10%. I was actually expecting them to launch at $350ish.


Core i9-13900K941 CAD / 726 USDCore i9-12900K818 CAD / 631 USD
Core i9-13900KF901 CAD / 695 USDCore i9-12900KF771 CAD / 595 USD
Core i7-13700K663 CAD / 511 USDCore i7-12700K566 CAD / 436 USD
Core i7-13700KF626 CAD / 483 USDCore i7-12700KF534 CAD / 411 USD
Core i5-13600K461 CAD / 355 USDCore i5-12600K399 CAD / 308 USD
Core i5-13600KF424 CAD / 327 USDCore i5-12600KF336 CAD / 259 USD
 

Vattila

Senior member
Oct 22, 2004
799
1,351
136
"What’s however a bit perplexing is that the core-to-core latencies between Gracemont cores is extremely slow, and that’s quite unintuitive as one would have expected coherency between them to be isolated purely on their local L2 cluster. Instead, what seems to be happening is that even between two cores in a cluster, requests have to travel out to the L3 ring, and come back to the very same pathway." (AnandTech)

The passage continues: "That’s quite weird, and we don’t have a good explanation as to why Intel would do this."

Isn't the explanation quite simply that cache coherence is resolved in the last-level cache (L3)? If my understanding is correct, we will see the need for smarter, faster and larger last-level caches as cache-coherence between CPU and GPU — and other accelerators — becomes ever more important.

1661373836652.png

PS. Remember that there are no instructions for communicating between cores. Cores communicate through shared memory — which requires a cache coherency protocol to maintain a consistent memory model.
 
Last edited:

shady28

Platinum Member
Apr 11, 2004
2,520
397
126
Nope. $289 is RCP i.e. for purchase of 1000 units in bulk. Retail prices are usually $10-20 higher.

Amazon shows it as $342 MSRP and Microcenter shows MSRP at $319.

Not that it matters, real price is whatever the market will bear. I still think these will be $350+ when they come out. Past that, the 12700K will be a viable option.
 
  • Like
Reactions: Kaluan

nicalandia

Diamond Member
Jan 10, 2019
3,330
5,281
136
If my understanding is correct, we will see the need for smarter, faster and larger last-level caches as cache-coherence between CPU and GPU — and other accelerators — become ever more important.

True. What takes me to my next question. How is Intel going to stitch Sierra Forest?

So far their Mesh(Mesh of Half Rings) have proven to be quite inferior to their Bidirectional Ring . Due to design power limits their Mesh can't grow to larger than 2MiB/L3$ per core.
 
Last edited:

Exist50

Platinum Member
Aug 18, 2016
2,445
3,043
136
Due to design power limits their Mesh can't grow to larger than 2MiB/L3$ per core.
What on earth gave you that idea? There's no such inherent limit to a mesh fabric (Intel's isn't anything special). Nor is L3 even something to focus on from a cloud perspective (Sierra Forest and Bergamo target market). Intel will surely reuse as much as they can from GNR, mesh included.

Though that said, I do wonder at what point having such a large unified L3 becomes more of a burden than it's worth. Surely that can't scale infinitely.
 

psolord

Golden Member
Sep 16, 2009
1,875
1,184
136
Not sure if anyone noticed but those price pics seem to confirm one thing about gen 13 - default max boost clocks.


View attachment 66569
View attachment 66570
View attachment 66571
My stupid ass though they were grams and I was like, hey look, the bigger the cpu, the heavier it gets, aka more transistors, yeepee, I'm so smart! xD

(yes I know about the fusing off of parts on same dies and related tiering-I wasn't born yesterday, I'm just stupid, it happens)
 

coercitiv

Diamond Member
Jan 24, 2014
6,151
11,685
136
I'm putting all my eggs in the 13600K basket as the all-round value champ from the Intel lineup. The 13500 will offer more raw value, but the inferior memory sub-system and locked SA voltage will put a lid on max performance. There's also decent overclock potential for the 13600K, as the delta in max clocks between i5 and i9 is growing this generation. We're looking at almost 10% OC for the i5 (assuming many dies will be able to hit a 5.x-5.6Ghz dynamic OC).
 
  • Like
Reactions: Elfear and shady28