Sunday, October 30th 2022

Intel's Next-Gen Desktop Platform Intros Socket LGA1851, "Meteor Lake-S" to Feature 6P+16E Core Counts

Keeping up with the cadence of two generations of desktop processors per socket, Intel will turn the page of the current LGA1700, with the introduction of the new Socket LGA1851. The processor package will likely have the same dimensions as LGA1700, and the two sockets may share cooler compatibility. The first processor microarchitecture to debut on LGA1851 will be the 14th Gen Core "Meteor Lake-S." These chips will feature a generationally lower CPU core-count compared to "Raptor Lake," but significantly bump the IPC on both the P-cores and E-cores.

"Raptor Lake" is Intel's final monolithic silicon client processor before the company pivots to chiplets built on various foundry nodes, as part of its IDM 2.0 strategy. The client-desktop version of "Meteor Lake," dubbed "Meteor Lake-S," will have a maximum CPU core configuration of 6P+16E (that's 6 performance cores with 16 efficiency cores). The chip has 6 "Redwood Cove" P-cores, and 16 "Crestmont" E-cores. Both of these are expected to receive IPC uplifts, such that the processor will end up faster (and hopefully more efficient) than the top "Raptor Lake-S" part. Particularly, it should be able to overcome the deficit of 2 P-cores.
Intel could find itself with a similar product differentiation problem it faced with the 11th Gen Core "Rocket Lake-S" desktop processors, where the physically low CPU core-count compared to the previous-generation (8-core vs. 10-core for "Comet Lake-S"); meant that both the Core i7-11700K and i9-11900K ended up being 8-core/16-thread processors. Here, we could see 6P+16E being the core-config of nearly all top SKUs, segmented by clock-speeds; while the mid-tier SKUs end up being 6P+8E.

Besides the CPU, "Meteor Lake-S" is expected to debut the new Xe-LPG graphics architecture for the iGPU, which could meet DirectX 12 Ultimate logo requirements. The iGPU on the "Meteor Lake-S" processor is expected to feature 4 Xe Cores, which works out to 64 EUs, and 512 unified shaders. This would still be a significant uplift from the iGPU of "Raptor Lake-S" with 32 EUs.

Intel is expected to restore CPU core-counts back to current levels with the 15th Gen "Arrow Lake-S" (2024-25). These chips are expected to come with core-configurations of up to 8P+16E. While the E-cores are expected to remain the same, the P-cores get a performance uplift, besides the addition of more cores. The "Compute Tile" (the die with the CPU cores) of "Meteor Lake-S" is built on the Intel 4 node (isopower characteristics comparable to TSMC 5 nm); while those of "Arrow Lake-S" will be built on the Intel 20A node (Intel is hyping 20A to be a pathbreaking node competitive with TSMC's sub 2 nm nodes).
Source: Wccftech
Add your own comment

113 Comments on Intel's Next-Gen Desktop Platform Intros Socket LGA1851, "Meteor Lake-S" to Feature 6P+16E Core Counts

#101
TheoneandonlyMrK
Dr. DroBeing fair, the AM4 platform was sold on a promise of forwards compatibility that was barely kept (and only kept because of external factors i. e. Alder Lake supremacy at the budget end), and users were highly encouraged to upgrade their processors. It was the exception rather than the rule.

I don't think it's only 0.5% of people, but I do believe the vast majority of customers purchase an OEM pre-built and do not change a single thing in the device through its service life. Not so much in the DIY channel.
Oh I agree fully with your Stance, I just think it's worth merit to exactly the kind of people who advised others what to buy.
The Enthusiasts are allll also small to large pool of people influencer's.

Was it perfect f no, was it worth it, a fair few think it was.

Over four years I did two motherboards and three CPU, always gaining performance and compatibility while only buying one part, oh and 5 memory kits.

All of which live on in other PC.
Posted on Reply
#102
mplayerMuPDF
Dr. DroThe USB problem was not on ASMedia, it was on AMD. It was fixed with the AGESA v2 1.2.0.2 update and in fact - it also crashed external USB controllers, in addition to the USB dropouts my EVGA sound card (which uses a custom USB bridge) also stopped functioning. It was both hilarious and sad.

Now if they fix the EDC bug... sounding like a broken record about it already, I think AM4 can finally be called "done and polished". Only took 'em more than half a decade, when and if they fix it...
Except it has never been fixed on my system or those of some others (including on TPU), even with the very latest AGESA (1.2.0.7) and if you do a quick web search you will find out that ASMedia simply has an atrocious reputation and that issues with their implementation were known even during the development of Ryzen. I have never had any USB issues with my (older) AMD powered laptops because they do not (seem to) rely on ASMedia garbage. At least for the Llano and Trinity/Richland chipset it is confirmed that AMD partnered with Renesas for their USB 3.0 implementation. I will never buy a product in which ASMedia has been involved in any way again in my life.
Posted on Reply
#103
Dr. Dro
mplayerMuPDFExcept it has never been fixed on my system or those of some others (including on TPU), even with the very latest AGESA (1.2.0.7) and if you do a quick web search you will find out that ASMedia simply has an atrocious reputation and that issues with their implementation were known even during the development of Ryzen. I have never had any USB issues with my (older) AMD powered laptops because they do not (seem to) rely on ASMedia garbage. At least for the Llano and Trinity/Richland chipset it is confirmed that AMD partnered with Renesas for their USB 3.0 implementation. I will never buy a product in which ASMedia has been involved in anyway again in my life.
I dunno if we're talking about the exact same problem, anyway, but once the problem was brought to AMD's attention on reddit I made a post there and was contacted by an AMD engineer who requested some data from my machine, the problems I've had were completely fixed two AGESA updates later.

Mind you socket AM5 is all AsMedia again, X570 is the only exception, so if you're adhering to that boycott you'll have to grab Intel.
Posted on Reply
#104
ThrashZone
Hi,
One more socket after this and they'll be close to 2066 again :laugh:
Posted on Reply
#105
mplayerMuPDF
Dr. DroI dunno if we're talking about the exact same problem, anyway, but once the problem was brought to AMD's attention on reddit I made a post there and was contacted by an AMD engineer who requested some data from my machine, the problems I've had were completely fixed two AGESA updates later.

Mind you socket AM5 is all AsMedia again, X570 is the only exception, so if you're adhering to that boycott you'll have to grab Intel.
Well, I was not planning to get AM5 anyway since it is way beyond my budget, it runs at insane temperatures and it is Pluton infested, so it is their loss honestly. They have made their bed in many different ways and now they will have to lie in it. I am not loyal to AMD; I am only loyal to some of their (old) products! To be honest, I am waiting for RISC-V to become a real option and until then I will simply stay with my pre-Ryzen hardware (I will be getting rid of my X470 setup before 2023!). I do not need the raw CPU power anyway. I have a fast and reliable SSD, good WiFi card and enough, reliable USB 3.0 ports, a decent iGPU with hardware H.264 decoding and that is enough for me. If I have to compile something it will take a bit longer but so be it. I have an EliteBook 725 G3 with damaged power jack (but it still works for now so if I keep it plugged in permanently it should be fine) and an EliteBook 725 G4 with damaged internal keyboard; both have been impossible to sell so I will be keeping at least one of them and use it as a replacement for my X470 based desktop. Even my 1.5 core Richland based ProBook 645 G1 is still very usable for web browsing with the right software setup (in fact I am typing this on it) and the Excavator based quad core in the 725s are monsters compared to the "dual" core (shared instruction decoder) in my ProBook. Once I install an M.2 MT7921K/L (you can get them for $10-20 on eBay from people who are ditching them for Intel AX cards because the Windows drivers supposedly suck) in the EliteBook it may actually be faster for web browsing than my Ryzen desktop with its 2.4-GHz-only 11n USB WiFi adapter. So basically I will be going back to where I was before I built my desktop in 2020 (and after I sold my Lenovo M91p mid tower) except it will have been a small upgrade from Kaveri to Carrizo/Bristol Ridge. I must say that I enjoy modding old prebuilts more actually than building my own desktop from scratch.

EDIT:
Literally minutes later as I boot my desktop today I experience the USB issue again (for some reason it always occurs with my USB WiFi adapter these days and I always discover it when I cannot connect to a WiFi network with wpa_supplicant; replugging it into the same port fixed it yet again):
"
[ 6.840389] usb 3-2: device descriptor read/64, error -110
[ 22.424350] usb 3-2: device descriptor read/64, error -110
[ 22.660196] usb 3-2: new high-speed USB device number 3 using xhci_hcd
[ 27.800210] usb 3-2: device descriptor read/64, error -110
[ 43.416361] usb 3-2: device descriptor read/64, error -110
[ 43.524389] usb usb3-port2: attempt power cycle
[ 43.936206] usb 3-2: new high-speed USB device number 4 using xhci_hcd
[ 54.760190] usb 3-2: device not accepting address 4, error -62
[ 54.888200] usb 3-2: new high-speed USB device number 5 using xhci_hcd
[ 65.512190] usb 3-2: device not accepting address 5, error -62
[ 65.512886] usb usb3-port2: unable to enumerate USB device
"
and also the usual: "[ 1.346225] usb: port power management may be unreliable" F ASMedia.
Posted on Reply
#106
Tech Ninja
can’t use Geekbench to estimate sales, can’t use steam, can’t use Userbenchmark. You can only use this one website in the UK or Germany that only sells to DIYers - when 90% of sales are OEM full PCs.

AMD is very popular in DIY and increasingly so in servers. It’s share of fully built desktops and laptops is terrible.

mans it’s dGPU share is down to 8%
Posted on Reply
#107
Mussels
Freshwater Moderator
Somehow we're down to 6 cores with higher wattages than ever before, and I hate this trend
Tech Ninjacan’t use Geekbench to estimate sales, can’t use steam, can’t use Userbenchmark. You can only use this one website in the UK or Germany that only sells to DIYers - when 90% of sales are OEM full PCs.

AMD is very popular in DIY and increasingly so in servers. It’s share of fully built desktops and laptops is terrible.

mans it’s dGPU share is down to 8%
Well, duh?

Only one of those shows sales, the others do not.
Userbench is proven to be utterly trash and should never be used, period - they have lost all credibility and have been proven to falsify all their results.
Steam shows every man and his dog, and their dogs cousins server - steam stats do not show "gamers" they show gaming PC's, they show shitty boxes used for in-home streaming, they show the laptop used to play peggle. Steam is inherently going to show older outdated systems, not new hardware.

Geekbench is of course, absolutely a retailer of CPU's and should be used to judge sales. MMmmmmhmmmm.

What the heck are you going on about here man?
Posted on Reply
#108
mplayerMuPDF
MusselsSomehow we're down to 6 cores with higher wattages than ever before, and I hate this trend
I am sorry but this is utter nonsense. There are additionally literally 16 E cores that, I repeat, do absolutely count. We are talking about more performance than FOUR Haswell i5s combined or viewed from another perspective more than a Broadwell Xeon E5-2680 v4 that these E cores together constitute. Now I know that some people are going to say "but muh games and single thread perf" and I respond with "how many damn P cores do your shitty games need?". Seriously maybe it is time to people stop playing crap games that are incredibly bloated and start changing their expectations. The PS4 was literally using Jaguar cores, which probably have about half the single threaded performance of these E cores if not less and then you have 4 or 6 P cores or whatever on top of those.

Chip designers simply cannot keep making their cores more complex and run at increasing frequencies forever. I don't think that people grasp how insanely large these new Sunny Cove derivative cores actually are. I mean, Sandy Bridge and Haswell/Broadwell were already pretty large cores. That is how Intel got away with mainstream and even "high end" dual cores in laptops for so long. Skylake is even bigger and Sunny Cove is massively bigger still. The increase in amount of required transistors was disproportional to the increase in single threaded performance of Sunny Cove. Go read Anandtech's deep dive. To increase multithreaded performance in an efficient way more modestly sized cores is the only way.
Posted on Reply
#109
Mussels
Freshwater Moderator
mplayerMuPDFThere are additionally literally 16 E cores that, I repeat, do absolutely count.
They dont for most tasks, and they're less efficient than anything else out there
This is from TPU's 12900K testing, but despite being named "efficiency cores" the secret sauce seems to be the missing word "low" before that

12900K's 8 E-cores is about equal to a 4 core ryzen 3300x in performance, and a 3600x in efficiency

E-cores are utter trash, and I will stand by that firmly because the evidence backs it up.

Posted on Reply
#110
mplayerMuPDF
MusselsThey dont for most tasks, and they're less efficient than anything else out there
This is from TPU's 12900K testing, but despite being named "efficiency cores" the secret sauce seems to be the missing word "low" before that

12900K's 8 E-cores is about equal to a 4 core ryzen 3300x in performance, and a 3600x in efficiency

E-cores are utter trash, and I will stand by that firmly because the evidence backs it up.

I recommend reading chipsandcheese.com/2022/01/28/alder-lakes-power-efficiency-a-complicated-picture/ The Gracemont core may not be the most amazing design ever but when run at a sane frequency (i.e. not stock) that suits it they are more efficient. "Gracemont is very efficient with integer workloads in the low 3 GHz range." The biggest problem here is that Intel is essentially doing an overclock out of the box to one up AMD in the benchmark game.
Posted on Reply
#111
Mussels
Freshwater Moderator
mplayerMuPDFI recommend reading chipsandcheese.com/2022/01/28/alder-lakes-power-efficiency-a-complicated-picture/ The Gracemont core may not be the most amazing design ever but when run at a sane frequency (i.e. not stock) that suits it they are more efficient. "Gracemont is very efficient with integer workloads in the low 3 GHz range." The biggest problem here is that Intel is essentially doing an overclock out of the box to one up AMD in the benchmark game.
You mean, just like their P cores as well?
(And AMD with Zen4)
Posted on Reply
#112
mplayerMuPDF
MusselsYou mean, just like their P cores as well?
(And AMD with Zen4)
Yes, but if I am not mistaken when run in their sweet spot frequency range they use less energy to complete an integer-heavy workload than a Golden Cove core could at any frequency, according to that Chips & Cheese article.
Posted on Reply
#113
Mussels
Freshwater Moderator
mplayerMuPDFYes, but if I am not mistaken when run in their sweet spot frequency range they use less energy to complete an integer-heavy workload than a Golden Cove core could at any frequency, according to that Chips & Cheese article.
Yeah, i'm salty that basically nothing runs in those sweet spots. No innovation happening right now, just throw voltage at the problem to beat the other guy in useless benchmarks at the detriment to every day use
Posted on Reply
Add your own comment
May 7th, 2024 20:26 EDT change timezone

New Forum Posts

Popular Reviews

Controversial News Posts