• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

Discussion Intel current and future Lakes & Rapids thread

Page 217 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Ajay

Diamond Member
Jan 8, 2001
9,204
3,768
136
It's based upon Intel's typical release schedule. Comet Lake this year, Rocket Lake a year later, Alder Lake a year after that, Meteor Lake a year later.
That, makes some sense. I'd be surprised if Adler Lake is able to be ramped up enough to displace all desktop processor needs from 14nm parts.
 

Exist50

Senior member
Aug 18, 2016
319
341
136
The real question about Alder Lake seems to be whether they can get a mostly-functional product out the door in 2021. They may not have fab issues to deal with, but with presumably two new cores (Golden Cove, Gracemont), hybrid, DDR5, PCIe 5.0 (?), and who knows what else, seems like there's a lot of room for error.

Clearly Alder Lake is more about mobile and per core performance than just straight MT. After all, it only has 8 big cores max.
I interpreted this the other way around. Assuming the rumor actually is an S-series chip, Atom cores are probably a good way to boost throughput without spending the extra silicon big cores would need. Also probably a useful pipecleaner for mobile. Hell, if something goes catastrophically wrong, they could probably disable all of the Atom cores and still have a viable desktop product.
 

Exist50

Senior member
Aug 18, 2016
319
341
136
Is Alderlake 2021 or 2022?..do we have any reliable info to what cores it is using?
The most detailed leak so far seems to be this one.

Golden Cove seems like a given for the Alder Lake, and if this is right about a hybrid configuration, then Gracemont seems like the logical companion, given Intel's roadmap.

As for 2021 vs 2022, well those cores are shown right on top of the 2021 mark, but who knows with Intel these days.
 

Dave2150

Senior member
Jan 20, 2015
638
178
116
Clocking sky high ? Huge IPC gain ? What drugs are you on ? I need some...
IPC gain will be huge compared to what Intel have now. I'm still using my 6700k, from August 2015. This is the exact same architecture as used in all the subsequent generation of Intel CPU, even the new Comet Lake 10 core has the same architecture.

It will be a very exciting time for Intel to get a modern architecture out for desktop :)
 

coercitiv

Diamond Member
Jan 24, 2014
4,514
6,137
136
10nm should be clocking sky high by then
I imagine 10nm will be clocking much higher (I used the adjective "skyhigh") when Alderlake-S releases in ~2 years
Hi, you seem to have been disconnected from the current state of affairs in Intel silicon development, so here's a quick refresh from Intel themselves:
Intel CFO George Davis warned that 10nm is not going to be this big savior for the company’s chip operations. This is despite the fact that Intel even has 10nm server CPUs SKUs on the horizon.

Davis explained that the company’s 10nm/10nm+ process node "just isn't going to be the best node that Intel has ever had." He even went on to add, "It's going to be less productive than 14nm, less productive than 22nm, but we're excited about the improvements that we're seeing and we expect to start the 7nm period and at a with a much better profile of performance over that starting at the end of [2021].”
If you still want to push some buttons around here you should start predicting the meteoric rise of their 7nm node, anything else will just make people confuse you with some lost time traveler from the past decade. Here's some more reading material to further improve your game.
 

Dave2150

Senior member
Jan 20, 2015
638
178
116
Hi, you seem to have been disconnected from the current state of affairs in Intel silicon development, so here's a quick refresh from Intel themselves:


If you still want to push some buttons around here you should start predicting the meteoric rise of their 7nm node, anything else will just make people confuse you with some lost time traveler from the past decade. Here's some more reading material to further improve your game.
Times change, 10nm or 7nm may be in a much better place sooner than we think. Either way I think it's great that Intel's 5 year old architecture is still best in class in single threaded workloads, versus the competition's brand new architecture. I suspect you're frustrated by this, hence the condescending tone in your post. We'll see who has the last laugh :)
 

uzzi38

Golden Member
Oct 16, 2019
1,753
3,491
106
Hi, you seem to have been disconnected from the current state of affairs in Intel silicon development, so here's a quick refresh from Intel themselves:


If you still want to push some buttons around here you should start predicting the meteoric rise of their 7nm node, anything else will just make people confuse you with some lost time traveler from the past decade. Here's some more reading material to further improve your game.
They've fixed something with Tiger Lake, and yields have significantly improved as well as per their latest statements (iirc they've said something that sounds kind of like roughly 2x ICL-U yields on launch).

Intel might actually be onto something in the next couple of years. I don't see 5GHz happening again, but high-4GHz (4.7-4.8GHz) by ADL-S should be doable.

As for the other guy who said 4.7GHz on TGL-U though, now THAT's a meme if I ever saw one. Nah, 4.5GHz is probably the top and to be completely honest, it's a pretty darn good place to be after Ice Lake, to say the least. Maybe they'll bin or TVB their way to 4.7GHz, but lets be real, that's not feasible for most devices outside fo microbursts. It's as much of an achievement as the 4.7GHz on the 3950X for the most part.

(Slightly related side note, I'm really hoping AMD define boost clocks for Vermeer like they did with Renoir - all the R7s and up on Renoir can peak at 100mhz over their boost clocks, and do a good job at sustaining max boost - or in some cases, higher - in SSE loads.)
 
  • Like
Reactions: coercitiv

NostaSeronx

Diamond Member
Sep 18, 2011
3,277
816
136
For sure by now pentium iv users are already running their cpus at 10Ghz.
Why are they on a processor that can only retire at most 3/4ths a AMD64 instruction or 1.5 of two IA32 instructions? In reference to the microarchitecture at 4.5 GHz the core at 9 GHz provides; 4x 64-bit ALU, 1x 64-bit L-AGU, 1x 64-bit S-AGU per clock. It is also the part that consumes the least power, the microarchitecture w/o the core is 15x as power hungrier than just the core.
(Tulsa @ 4.5 GHz would probably be greater than >310W but the cores would actually only be pushing out >20.72 watts)

Modern processors like the 14nm/10nm core from Intel and 7nm core from AMD, can retire several times the above at lower clocks and lower power.
 
Last edited:

lobz

Golden Member
Feb 10, 2017
1,767
2,284
136
Times change, 10nm or 7nm may be in a much better place sooner than we think. Either way I think it's great that Intel's 5 year old architecture is still best in class in single threaded workloads, versus the competition's brand new architecture. I suspect you're frustrated by this, hence the condescending tone in your post. We'll see who has the last laugh :)
Suggesting that a relatively power hungry, higher clocked CPU gets a couple of percentages lead in some single-threaded workloads and software with glacier slow developers (hello Adobe!), all the while being severely outclassed in basically every other area despite costing more than its competitors makes people here frustrated, well, that painful disconnection from reality only speaks volumes of your true frustrations, although I guess that wasn't your intention 😉
 
Last edited:

coercitiv

Diamond Member
Jan 24, 2014
4,514
6,137
136
They've fixed something with Tiger Lake, and yields have significantly improved as well as per their latest statements (iirc they've said something that sounds kind of like roughly 2x ICL-U yields on launch).
I don't have a problem with Intel's progress with TGL, at least as far as we know based on base clocks & those inventory stocks which are presumably twice the volume ICL had prior to launch. (I still have a problem with the lack of performance demos - especially in the context of Renoir, but that is just my opinion).
 

Zucker2k

Golden Member
Feb 15, 2006
1,459
801
136
Davis explained that the company’s 10nm/10nm+ process node "just isn't going to be the best node that Intel has ever had." He even went on to add, "It's going to be less productive than 14nm, less productive than 22nm, but we're excited about the improvements that we're seeing and we expect to start the 7nm period and at a with a much better profile of performance over that starting at the end of [2021].”
I believe he's simply saying that because 7nm is just around the corner, 10nm's impact window is closing so it won't be as productive as their previous nodes. This makes a lot of sense, but I expect 10nm to continue improving, ala 14nm, and help carry Intel's future chip loads.
Even though 10nm has proven to be such a pain for Intel, I'm happy they didn't scrap it because I think it's the node that's positioned to combine the highest density/frequency, since it's expected that 7nm and below should not clock as high due to overall heat and hotspots arising from the higher densities. I think the positive news on 10nm in recent weeks is going to continue as Intel continues to tweak the node.
 

jpiniero

Diamond Member
Oct 1, 2010
9,923
2,269
136
That, makes some sense. I'd be surprised if Adler Lake is able to be ramped up enough to displace all desktop processor needs from 14nm parts.
It's certainly an issue that Intel is still going to be selling 14 nm desktop parts when AMD goes to TSMC 5 nm.
 

coercitiv

Diamond Member
Jan 24, 2014
4,514
6,137
136
I believe he's simply saying that because 7nm is just around the corner, 10nm's impact window is closing so it won't be as productive as their previous nodes. This makes a lot of sense, but I expect 10nm to continue improving, ala 14nm, and help carry Intel's future chip loads.
I partially agree. Yes, it's about financial impact, window is closing with 7nm around the corner, but his message is also about 7nm being their prime candidate for getting back into the game. It signals trust in early 7nm results. The great 14nm saw multiple improvements as it was their only node to use for everything. This will not be the case now, if 7nm is done well then server & mobile will need to jump there immediately. Even desktop might see a net gain in ST performance if clock regression is matched by IPC jump.

10nm simply won't get the beauty treatment 14nm received. Not with a healthy 7nm transition anyway.
 

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
21,494
9,551
136
I partially agree. Yes, it's about financial impact, window is closing with 7nm around the corner, but his message is also about 7nm being their prime candidate for getting back into the game. It signals trust in early 7nm results. The great 14nm saw multiple improvements as it was their only node to use for everything. This will not be the case now, if 7nm is done well then server & mobile will need to jump there immediately. Even desktop might see a net gain in ST performance if clock regression is matched by IPC jump.

10nm simply won't get the beauty treatment 14nm received. Not with a healthy 7nm transition anyway.
I don't agree or disagree with any of this. But until I see a chip on 10nm or 7 nm from Intel, that not 4 cores and 25 watts or so (their only 10 nm product I have seen), I just have no faith in them.

The day they have something to compete with Rome in the server space, or threadripper in the HEDT space or even the 3900x or 3950x in desktop, then I will say they have a good manufacturing ability. Until then, its all guesses.
 
  • Love
Reactions: spursindonesia

Exist50

Senior member
Aug 18, 2016
319
341
136
I'm not sure where all this clock speed pessimism is coming from, and I think Tiger Lake should help align expectations accordingly. My personal stake in the ground is 4.8GHz (ignoring TVB) if anyone wants a juicy quote for their signature.

No, I think the question isn't whether they'll get back to 5GHz, but rather if and when they'll go beyond that. 6GHz for 5nm, maybe?
 
  • Haha
Reactions: spursindonesia

uzzi38

Golden Member
Oct 16, 2019
1,753
3,491
106
I believe he's simply saying that because 7nm is just around the corner, 10nm's impact window is closing so it won't be as productive as their previous nodes. This makes a lot of sense, but I expect 10nm to continue improving, ala 14nm, and help carry Intel's future chip loads.
Even though 10nm has proven to be such a pain for Intel, I'm happy they didn't scrap it because I think it's the node that's positioned to combine the highest density/frequency, since it's expected that 7nm and below should not clock as high due to overall heat and hotspots arising from the higher densities. I think the positive news on 10nm in recent weeks is going to continue as Intel continues to tweak the node.
Just remember, real products don't necessarily sport the same densities as the node, and just because Intel gave a density figure higher than N7 doesn't mean their final products will be more dense than competing ones.

I can't state this enough - Tiger Lake needs to fix a LOT of serious problems with current 10nm. And thankfully, it looks like it is.

Still, I wouldn't go claiming highest density/clock even when it becomes known it clocks higher.
 

NostaSeronx

Diamond Member
Sep 18, 2011
3,277
816
136
Desktop CPUs on 7nm are still ~2 years away. That's assuming Intel hits one out of the park on Meteor Lake.
Technically, the renewed push roadmap started in 2019.

2019 = 10nm HVM
2021 = 7nm HVM
2023 = 5nm HVM
2025 = 3nm HVM
2027 = 2nm HVM
2029 = 1.4nm HVM

Two years away, we will be entering 5nm territory from Intel.
 

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
21,494
9,551
136
Technically, the renewed push roadmap started in 2019.

2019 = 10nm HVM
2021 = 7nm HVM
2023 = 5nm HVM
2025 = 3nm HVM
2027 = 2nm HVM
2029 = 1.4nm HVM

Two years away, we will be entering 5nm territory from Intel.
Its early 2020. So 2 years away is 2022. Thats 7 nm by your post, NOT 5.
 

NostaSeronx

Diamond Member
Sep 18, 2011
3,277
816
136
Its early 2020. So 2 years away is 2022. Thats 7 nm by your post, NOT 5.
That is one year after 7nm goes HVM and one year before 5nm goes HVM. Which means in two years we will be entering Intel's 5nm territory.

7nm risk production will start this year. <== Meaning 7nm IP is all done and the processors are in yield ramp.
5nm risk production will start in 2022. <== Meaning 5nm IP is all done and the processors are in yield ramp.
 

uzzi38

Golden Member
Oct 16, 2019
1,753
3,491
106
Technically, the renewed push roadmap started in 2019.

2019 = 10nm HVM
2021 = 7nm HVM
2023 = 5nm HVM
2025 = 3nm HVM
2027 = 2nm HVM
2029 = 1.4nm HVM

Two years away, we will be entering 5nm territory from Intel.
Maybe you misread what was said. 7nm desktop chips will not exist on the market until at least 2 years from now.

Stating when 7nm begins HVM is irrelevant if products on the node aren't of the aforementionned category of desktop CPUs.
 

NostaSeronx

Diamond Member
Sep 18, 2011
3,277
816
136
7nm desktop chips will not exist on the market until at least 2 years from now.
imho, the first consumer desktop 7nm CPU will be in a MCP config and will probably be on an element AIB. Which will probably launch sooner than later as a Xeon Element, with a micro-server scale element-compatible case. (2 top-left eles, 2 bottom-left eles, 2 top-right eles, 2 bottom-right eles replacing the traditional 8P motherboard // CXL crossbar per eight element + UPI 2.x per two-element + Gen-Z to interconnect multiple Xeon Element clusters in a rack.)

Aftermarket Element-compatible multi-wide case.
Aftermarket Element-compatible 3-wide >250W heatsink/cooler
Aftermarket Element-board with more than one ATX12/EPS12

New aged, simple PL2 tweaked overclocking.
 
Last edited:

ASK THE COMMUNITY

TRENDING THREADS