Discussion Intel current and future Lakes & Rapids thread

[uzzi38]

EXCLUSIVE ⋮ HARDWARE ⋮ LEAK
Intel’s Entire 10th Gen Comet Lake Desktop CPU Lineup Leaked – Up To 10 Cores, 20 Thread Core i9-10900, 80W TDP, W480 Chipset With New LGA 1200 Socket.

Intel's Entire 10th Gen Comet Lake Desktop CPU Lineup Leaked - Up To 10 Cores, 20 Thread Core i9-10900, 80W TDP, W480 Chipset With New LGA 1200 Socket

Intel's entire 10th Generation Comet Lake desktop CPU family along with the respective 400-series (LGA 1200) platform has been leaked.

wccftech.com

and now the Intel machine is in high gear,.

I know that the initial response to anything that WCCFTech spouts is unlikely at best, I will point out that Komachi does think it to be accurate, and he himself is an incredibly accurate source of rumours/leaks.

Anyway, I'mma sit here and laugh for the next week about the 80W TDP (yes, I know it's because of the 3GHz base clock, but still)

 

[uzzi38]

We don't need the big and bold font. Also, when are these launching ? nothing I could find in that article.

There's been leaked roadmaps saying either late Q1 or early Q2 2020.

 

[Markfw]

There's been leaked roadmaps saying either late Q1 or early Q2 2020.

leak,leak,leak,rumor,rumor... When is Intel going to tell us something we can reply on ?

 

[uzzi38]

leak,leak,leak,rumor,rumor... When is Intel going to tell us something we can reply on ?

You're expecting way too much there

Anyway, if they did, they'd cut very heavily into 9th Gen sales, so they won't for a while.

 

[Bouowmx]

There is a Comet Lake thread. But it got buried under page 1 already.

The documents look like really well produced stuff. But why do the SKUs already have detailed specifications?

 

[jpiniero]

The same rumor mill that called 10nm completely off 1 year ago? There is so much nonsense going on.

Charlie was right that Intel did cut most the fabs they had sitting on 10 nm back to 14 nm, but well he's Semi-Accurate. I could see them needing some additional 10 nm capacity if they are committed to actually selling Icelake-SP. But the AT post only talks about adding one additional fab to the two they have now.

 

[jpiniero]

There is a Comet Lake thread. But it got buried under page 1 already.

The documents look like really well produced stuff. But why do the SKUs already have detailed specifications?

I assume it's fake since the i5 and i3 specs have much lower base and turbo than their Coffee Lake equivalent parts. The i3 10100 I believe actually has a base of 3.6 (like the 9100) or 3.7 with a turbo of maybe 4.4.

Plus it's WTFTech. Some joker probally made up slides and sent it to them and they ran with it.

 

[mikk]

Charlie was right that Intel did cut most the fabs they had sitting on 10 nm back to 14 nm, but well he's Semi-Accurate.

Charlie claimed 10nm is killed off - he was dead wrong. You are claiming Alder Lake is definitely cancelled based on a "rumor mill"....big lol. You should have something more concrete when you say definitely cancelled.

And by trusting the rumor mill you are ignoring this rumor mill for some reason:

However, we heard from a reliable source that Intel's 10nm desktop plans absolutely include standalone CPUs, not just ones bound for NUCs, and "pretty cool ones" at that.

Intel denies rumor it's skipping 10nm desktop CPUs to focus on 7nm (Updated)

There are still plans to launch 10nm desktop processors.

www.pcgamer.com

Standalone CPUs, no NUCs.

 

[Ajay]

Then shut up and don't post such nonsense next time! You are shocked because you are clueless and ignorant more than once.

Then why not just post what the actual facts are instead of just whining. And grow up while your at it, this is just an enthusiast tech forum. I come here for fun, if you have other objectives, just put me on ignore.

 

[OriAr]

So many conflicting rumors.... At this stage I wonder if Intel is leaking false stuff to the media to put people off their plans and let their engineers work in peace.
Intel's growing confidence in 10nm makes me think that we will definitely see some desktop CPUs in 10nm, question will be the volume of them, I could see it in relatively low volume until 7nm comes in 2022, although who knows really, at this stage I think the media haven't got a clue either.

 

[jpiniero]

Standalone CPUs, no NUCs.

I would consider any of the three that I mentioned as fulfilling as what Intel said about releasing 10 nm desktop parts. You have to be realistic about what Intel can actually produce on 10 nm, and if it makes sense to actually release compared to what they can do with 14 nm and what AMD is doing.

 

[mikk]

Then why not just post what the actual facts are instead of just whining. And grow up while your at it, this is just an enthusiast tech forum. I come here for fun, if you have other objectives, just put me on ignore.

The facts: You were dead wrong and started personal attacks because you run out of arguments, this is such a disgrace. Your posts are full of shit and nonsense and you are unable to accept it. Acceppt your falsehood or shut up.

I would consider any of the three that I mentioned as fulfilling as what Intel said about releasing 10 nm desktop parts. You have to be realistic about what Intel can actually produce on 10 nm, and if it makes sense to actually release compared to what they can do with 14 nm and what AMD is doing.

You have to be realistic about 7nm as well. In H2 2021 and even more in 2022 they will slowly start shifting the first product lines to the 7nm process like GPUs. There is no room for the desktop with other more important products in the waiting queue. By shifting the first products to the 7nm node they will get free capacity which they can use for the desktop. Also it sounds like you don't expect improvements on 10nm, be it frequency or volume improvements, do you think this is realistic? With Alder Lake they probably won't even need 5 Ghz, if they get another healthy IPC increase with Golden Cove they should be fine with 4-4.5 Ghz. There is no guarantee 7nm has better clock abilities in the first 1-2 years, I wouldn't count on it. I doubt we will see 7nm in desktop before 2023. I would prefer 7nm desktop parts in 2022 but I don't think this is realistic.

 

[Zucker2k]

The facts: You were dead wrong and started personal attacks because you run out of arguments, this is such a disgrace. Your posts are full of shit and nonsense and you are unable to accept it. Acceppt your falsehood or shut up.




You have to be realistic about 7nm as well. In H2 2021 and even more in 2022 they will slowly start shifting the first product lines to the 7nm process like GPUs. There is no room for the desktop with other more important products in the waiting queue. By shifting the first products to the 7nm node they will get free capacity which they can use for the desktop. Also it sounds like you don't expect improvements on 10nm, be it frequency or volume improvements, do you think this is realistic? With Alder Lake they probably won't even need 5 Ghz, if they get another healthy IPC increase with Golden Cove they should be fine with 4-4.5 Ghz. There is no guarantee 7nm has better clock abilities in the first 1-2 years, I wouldn't count on it. I doubt we will see 7nm in desktop before 2023. I would prefer 7nm desktop parts in 2022 but I don't think this is realistic.

You have to give it to him (and one other individual in this thread), he's been very consistent in his pessimism towards Intel's 10nm yields. I personally can't wait to see him eat some crow when the full-blown 10nm desktop chips drop in the near future.

 

[ondma]

The question is, even if 10 nm comes to the desktop, will it bring any significant improvement. In mobile perhaps, better igpu, higher IPC, and maybe power savings. On the desktop, any ipc gains will probably be eaten up by lower clocks. And igpu is not a major factor. And with the yield issues, I dont really see them increasing core counts. So I dont really hold much optimism for 10 nm on the desktop, even if it shows up.

 

[Ajay]

The facts: You were dead wrong and started personal attacks because you run out of arguments, this is such a disgrace. Your posts are full of shit and nonsense and you are unable to accept it. Acceppt your falsehood or shut up.

What falsehood? Tell me, really, Im not a mind reader. Apparently, you can’t endure someone having a bit of fun trying to guess what something means or doesn’t mean. So, in your opinion, I am not posting up to your standards. Sorry that I can’t meet you standard of perfection. Why can’t you give me a reasoned response to me like you just did with JP. We're all guessing here because of the lack of information from Intel.

 

[Zucker2k]

The question is, even if 10 nm comes to the desktop, will it bring any significant improvement. In mobile perhaps, better igpu, higher IPC, and maybe power savings. On the desktop, any ipc gains will probably be eaten up by lower clocks. And igpu is not a major factor. And with the yield issues, I dont really see them increasing core counts. So I dont really hold much optimism for 10 nm on the desktop, even if it shows up.

I'd leave the performance decision to Intel. They'll keep holding off till 10nm is ready, so if they decide to release it then it'll mean the performance is there over the previous generation.

 

[TheGiant]

So many conflicting rumors.... At this stage I wonder if Intel is leaking false stuff to the media to put people off their plans and let their engineers work in peace.
Intel's growing confidence in 10nm makes me think that we will definitely see some desktop CPUs in 10nm, question will be the volume of them, I could see it in relatively low volume until 7nm comes in 2022, although who knows really, at this stage I think the media haven't got a clue either.

if they fake the 10nm news, that will be the end of Intel as we know it today
IMO they found something how to fix the 10nm, it is not as good as it should be but its working
you can buy Icelake everywhere so the mobile parts are not paper launchs

 

[jpiniero]

You have to be realistic about 7nm as well. In H2 2021 and even more in 2022 they will slowly start shifting the first product lines to the 7nm process like GPUs. There is no room for the desktop with other more important products in the waiting queue.

Which is why I'm inclined to believe it'll be Rocket Refreshed but I'm a little optimistic that between EUV and chiplets they will be able to move the entire client lineup to 7 nm in 2022.

 

[TheGiant]

I'd leave the performance decision to Intel. They'll keep holding off till 10nm is ready, so if they decide to release it then it'll mean the performance is there over the previous generation.

IMO they caught themselves with more+ 14nm and skylake performance at 5ghz
They can maybe clock icelake to 4.2ghz but that means a tie not a win

 

[mikk]

On the desktop, any ipc gains will probably be eaten up by lower clocks.

Alder Lake comes with Golden Cove archictecture, the IPC Gap could be really big.

And with the yield issues, I dont really see them increasing core counts. So I dont really hold much optimism for 10 nm on the desktop, even if it shows up.

Don't you expect yield improvements towards 2022? You guys are really thinking Intel won't improve the 10nm node anymore despite Intel saying the yields are improving. Improving the 10nm is a must have for dGPU and Sapphire Rapids. Desktop 10nm isn't coming anytime soon, it's a generation next after Rocket Lake which comes in 2021, there is so much time.....the current state of 10nm isn't meaningful for Alder Lake-S. We haven't seen 10nm++ yet, who knows maybe with Tigerlake it suddenly looks much better, I mean they have confirmed transistor optimization for Willow Cove. With Icelake we already have seen a big improvement over the half broken Cannonlake.

 

[Ajay]

Just figured I'd repost this from the main site, since it something concrete :

From the same article, Intel has said that yield improvement on 10nm is ahead of schedule.

 

[OriAr]

Intel can probably put out a 4.5 GHz ICL desktop CPU out but it's not that big of a performance jump so they might not want to release it.
By the time Golden Cove hits the IPC gap over SKL could be nearing NetBurst > Core levels so at that stage even if there is a bit of regress in clock speeds the performance jump will more than justify it (Like it did in 2006) so Intel will decide to put it out.
At this stage I think they'll want any 10nm desktop CPU to beat CFL at 5.3 GHz by at least 10% or else it is not worth it for them. (Which means any ICL desktop parts will have to hit 4.9 GHz reliably, not an easy thing to do right now)
And regarding the shape of 10nm, Intel visibly has much more confidence in 10nm than they did this time last year and that ICL-SP and the dGPU are coming out next summer, not to mention ICL actually shipped in a decent volume (You can actually get an ICL laptop fairly easily), so I do think yields have improved considerably.
If 10nm desktop is indeed the generation after RCL then we are probably looking at a late 2021 release for it, plenty of time for yields to improve into a level Intel will consider acceptable, especially if it's Golden Cove which means some clock speed regression might be acceptable.

 

[OriAr]

if they fake the 10nm news, that will be the end of Intel as we know it today
IMO they found something how to fix the 10nm, it is not as good as it should be but its working
you can buy Icelake everywhere so the mobile parts are not paper launchs

I actually meant that Intel might leak that 10nm is dead on desktop to give their engineers the chance to work at some peace, as well as maybe to flush out any leaks.
It's obvious Intel are now running a very tight ship regarding leaks and they could very well work on 10nm desktop parts and the media will have no clue.
At this stage Intel probably found out how bring 10nm up to an acceptable level, even if it's not quite at the place of 14nm yet (It's probably better than how 14nm was at 2015 though).

 

[Markfw]

Just figured I'd repost this from the main site, since it something concrete :

From the same article, Intel has said that yield improvement on 10nm is ahead of schedule.

The way that chart is made, 10+ may not be out until the end of 2020.

 

[OriAr]

The way that chart is made, 10+ may not be out until the end of 2020.

I think ICL is 10nm at this chart.
10+ is TGL.