Intel "Alder Lake-S" Due for September 2021

[Makaveli]

I thought Alder Lake was 7nm and due in 2023?
What happened to that?
Or are these the larger "Extreme Edition" chips??

No

Intel just said they fixed 7nm.
Alder Lake is 10nm.

 

[L'Eliminateur]

You will also need a need motherboard and ram for Alderlake so buying into Rocket Lake S kind of a dead end.

Buying ANYTHING this year is a dead end, DDR5 is coming by the end of the year and whilst it's going to take time to ramp up anything DDR4 is EOL by now, same thing when DDR4 came out, the year before it's a dead year for the old tech.

I do wonder if AMD will launch a transitional CPU, as with chiplets they could "theoretically" release a Zen 4 with DDR4 support(by using the existing zen2-3 cIOD) -unless zen4 chiplets use a totally different infinity fabric-(maybe wider, surely it will be much faster to take advatage of DDR5 BW) without much hassle. Or they'll simply go all in on zen4/am5 with a new clean design free of AM4 legacy(i think it will be this option)

 

[Makaveli]

I do wonder if AMD will launch a transitional CPU, as with chiplets they could "theoretically" release a Zen 4 with DDR4 support(by using the existing zen2-3 cIOD) -unless zen4 chiplets use a totally different infinity fabric-(maybe wider, surely it will be much faster to take advatage of DDR5 BW) without much hassle. Or they'll simply go all in on zen4/am5 with a new clean design free of AM4 legacy(i think it will be this option)

Instead of Zen 4 with with DD4 I wondering if they will go the other way around. So to test the waters on the AM5 platform and do Zen 3+ on AM5 with DDR5, So by the time Zen 4 is launching in 2022 they already have the boards and memory on the platform which has seen 6 months on the market already.

 

[SL2]

Maybe he's a martian. Who knows. The argument he presented only discusses power usage, nothing else. Speculating on other reasons with no information to back them up is pointless.

If you are upgrading to a new device to "save energy" you are a fool who has bought into green marketing that ONLY looks at energy used and not energy to manufacture and transport.

He only said what "would be sweet", and that he cares, he never said anything about upgrading at that point.
You started speculating about buying, no one else.

 

[L'Eliminateur]

Instead of Zen 4 with with DD4 I wondering if they will go the other way around. So to test the waters on the AM5 platform and do Zen 3+ on AM5 with DDR5, So by the time Zen 4 is launching in 2022 they already have the boards and memory on the platform which has seen 6 months on the market already.

Well they've sais that Zen5 will be out this year(maybe Q1 of 2022), so i don't think there will be enough time to port zen3 to DDR5, considering the lack of stock and time it takes for a respin i don't think it would be wise to waste resources on a "+" minor improvement("minor" per se, it would probably require considerable engineering to optimize for ddr5) and further deviate fab capacity from Zen4 directly

 

[SL2]

Well they've sais that Zen5 will be out this year(maybe Q1 of 2022)...

Who are "they"? AMD? Like I said before, it sounds unlikely to me.
Though I'm strictly talking about Ryzen, not Epyc, etc.

AMD has been launching its Ryzen models with more than 12 months between. I just looked at review dates here, as they usually are published when NDA ends (I'd assume).

1800X - 2 March 2017

2700X - 19 April 2018, 413 days after the launch before

3700X - 7 July 2018, 445 days after the launch before

5600X - 5 November 2020, 487 days after the launch before


Given that the latest Ryzen models are the most competitive for its time, and that AMD never had such a hard time meeting the demands as they do now, I'd be surprised if we'll see any major launch from AMD this year. Boring -XT models doesn't count.

7 nm production shortage would be one reason to move faster, but I'm not sure about the 5 nm production capacity anyway.

Edit: Or, launching Zen 4 on desktop this year is the very reason for why Ryzen 5000 has so few models. (not only low supply)
Let's say AMD launched those four 5000 models just to stay competitive, but they never intended to replace the whole 3000 lineup, that will happen with Zen 4.
This would also work as a plan B. If they'd expect delays with Zen 4, they'd still have the quite recent Ryzen 5000, and could in turn launch more budget models for instance.

 

[blu3dragon]

oh yea Alderlake Announcement Date "September 2021"

Laptop SKUs = December 2021
Desktop AlderLake S 10nm = March 2022
CPU in Stock and at MSRP = 2H 2022

This seems the more likely scenario. I need to google a little harder, but wasn't Rocket lake coming in 2020, 6 months after comet lake at one point?

 

[efikkan]

I'm not convinced Alder Lake-S will be available in volumes in September 2021.

This big.LITTLE architecture is so stupid to me in desktop. There will be tons of bugs on both sides (CPU and operating system) in this first generation.

And in desktop this is not needed at all. Who cares, that light-load tasks are consuming 1-5W instead of 5-20W in the desktop?

Yes, this will be very challenging for the OS, and perhaps even a few applications will behave sub-optimally.
But unfortunately, core count sells.

Buying ANYTHING this year is a dead end, DDR5 is coming by the end of the year

There is always the next big thing.
DDR5 will be offering more bandwidth, not lower latency. So I don't expect it to be a big deal for the desktop. Servers on the other hand…

and whilst it's going to take time to ramp up anything DDR4 is EOL by now, same thing when DDR4 came out, the year before it's a dead year for the old tech.

I think you should do a search for what EOL means.
DDR4 will be made for years to come.

 

[L'Eliminateur]

Who are "they"? AMD? Like I said before, it sounds unlikely to me.
Though I'm strictly talking about Ryzen, not Epyc, etc.



Edit: Or, launching Zen 4 on desktop this year is the very reason for why Ryzen 5000 has so few models. (not only low supply)
Let's say AMD launched those four 5000 models just to stay competitive, but they never intended to replace the whole 3000 lineup, that will happen with Zen 4.
This would also work as a plan B. If they'd expect delays with Zen 4, they'd still have the quite recent Ryzen 5000, and could in turn launch more budget models for instance.

"they" is AMD itself, but i've been rechecking the news sources just now and it appears that indeed Zen 4 has slipped to 2022 and there will be a zen3 refresh in Q3 of this year

DDR5 will be offering more bandwidth, not lower latency. So I don't expect it to be a big deal for the desktop. Servers on the other hand…

every DDR generation has not offered lower latency, so that's a moot point, the goal has always been "more BW, same latency", but DDR5 has big improvements on BW, density and parallelismo(being 2x32b channels changes a lot) , so it will be a big deal on desktop as well if only because all the platforms move to DDR5 so you'll have to upgrade just as well as when DDR4 came out.

 

[efikkan]

every DDR generation has not offered lower latency, so that's a moot point, the goal has always been "more BW, same latency", but DDR5 has big improvements on BW, density and parallelismo(being 2x32b channels changes a lot) , so it will be a big deal on desktop as well if only because all the platforms move to DDR5 so you'll have to upgrade just as well as when DDR4 came out.

Well, that's easy to estimate. Just look at benchmarks of HEDT vs. mainstream CPUs, and you'll se which workloads may benefit significantly from additional bandwidth (most wouldn't).

 

[RandallFlagg]

I'm not convinced Alder Lake-S will be available in volumes in September 2021.

I think they are saying the chips will ship in Sep 2021. That would mean that availability would begin about 4-6 weeks later in shipping OEM rigs and wide availability in about 3 months. So, holiday season, 2021.

I think the biggest obstacle to getting these out there will be availability and cost of DDR5. It's kind of a chicken and egg scenario. There is no reason to make DDR5 DIMMs until there's something that can use it. Shifting to something like DDR5 involves a lot more than just getting the CPU out there.

For that reason I imagine we will see this combination of Alder Lake + DDR5 from OEMs first, Dell HP Lenovo etc. Those OEMs will seed the market for DDR5. Then the OEM users who want to upgrade RAM would create demand for DDR5 from retailers like Newegg etc.

I wouldn't expect a lot of supply in DIY channel for several months after the OEMs get theirs. It's always like that with Intel to some degree anyway, but will be even more pronounced with Alder Lake needing DDR5 seeding the market so that retailers will buy and be confident that they won't have product sitting on the shelf for many months.

This also has to happen for AMD to launch DDR5 platforms. They don't (yet) have enough of the market to effectively create the demand needed for DDR5.

 

[B-Real]

The only CPU that’s worth buying this year

Sorry, what?

 

[pjl321]

My opinion on Alder Lake is don't be fooled, this is yet another 8 core CPU from Intel. Granted the low power cores will help run simple background tasks leaving the powerful cores to tackle the hard stuff more freely, so absolutely it will be faster than simply 8 Cyprus Cove cores just on their own but by how much is a very tough call. I guess if Alder Lake has a big performance lead in single thread over Zen 3 then an 8+8 CPU would make for a great gaming chip, but how it will do against Zen 3+ and Zen 4 is another story.

But it's the other use case scenarios that are very thread heavy where I see Intel losing very badly. I don't buy into the need for low power / low heat cores when talking about desktop CPUs, I'm sure the big.LITTLE design will work well for laptops and other highly power sensitive devices but not mains powered, high performance desktop machines. Modern CPUs can already down-clock and under-volt significantly when not in use or not being pushed hard so I just don't see how these little cores are going to be all that useful for anything really.

My reasoning for why I believe big.LITTLE designs won't compete performance wise much past the number of big cores is seen in Apple's M1 benchmarks. This amazing chip has fantastic single thread performance, up there with the very best chips on the market today and easily beating one of AMD's current top mobile chips the 4900HS (I know 5000's are almost here) but when it comes to multi threading the 4900HS totally destroys the M1 chip because the M1 is only a 4+4 chip and the 4900HS is a real 8 core chip. So Intel can bring 8+8 but I am pretty sure 5900X will be faster in work loads that use 12 threads or more, then we have the 6000 refresh which will only increase this gap and then Zen 4 will like make a 8 core CPU look very much entry level.

I think we all know the real reason Intel can't produce much more than 8 big cores in the mainstream and that is because of the aging ringbus design they are still using. Clearly AMD caught them by surprise with how quickly they caught up and then how far they have pulled away. Intel simply hasn't had the time nor FAB performance to react yet. I dare say if everything was running smoothly for Intel we would already be well into 7nm Meteor Lake chips right now and would be wondering if Zen 4 might close the gap to Intel in 2022. But that is not the situation and in my opinion Intel still needs to realise this, they need to price their products better to reflect they are only mid-range now and they desperately need to move to or licence TSMC's nodes if at all possible. I know the rumours are that Intel is meant to be using TSMC for 5nm CPUs this year then 3nm CPU next year but I will believe it when I see it! I just don't see how TSMC has the capacity to help Intel when it can't produce enough wafers for it's current long term (more important) customers.

Anyway, at the end of the day this 8 core CPU (with an extra few little cores) is meant to see Intel through 2022 and I believe into 2023, at the same time AMD will be giving us 24 high performance cores in the top end Zen 4 chips. I just don't see how Intel can put up any kind of fight again anything more than an 8 core Zen 4 chip, meanwhile AMD will be offering 12 core, 16 core, 20 core and 24 core options, plus Threadripper which will be up to 96 cores!

I just really hope Alder Lake has a massive single thread performance improvement over Zen 3 to give it the edge in many sparsely threaded allocations like Photoshop, Lightroom, etc and probably more importantly in gaming too. This should, if priced right, mean we have attractive options depending on what your usage needs are and crucially some competition so prices don't get out of hand in a monopoly again.

 

[Arc1t3ct]

What a mess...

in a period of 12 months we've had 3 CPU generations, 3 different chipsets and 2 cpu sockets!!! All for the same market!! Why are we putting up with this???

 

[pjl321]

What a mess...

in a period of 12 months we've had 3 CPU generations, 3 different chipsets and 2 cpu sockets!!! All for the same market!! Why are we putting up with this???

What are you purposing, maybe a military coup of Intel?

 

[AusWolf]

orrr maybe I was in the market for a new cpu anyway and I like the idea that it can turn basically part of itself off for lighter tasks to reduce power consumption?

Basically any modern CPU can do that, it has nothing to do with the big.LITTLE configuration.

Besides, buying a several hundred $ part to save pennies on your monthly electricity bill is kind of pointless imo.

This big.LITTLE architecture is so stupid to me in desktop. There will be tons of bugs on both sides (CPU and operating system) in this first generation.

And in desktop this is not needed at all. Who cares, that light-load tasks are consuming 1-5W instead of 5-20W in the desktop?

I wouldn't use the word "stupid", but it's weird to say the least. With games starting to benefit from 6+ cores, I can't see this being Intel's new gaming flagship.

Not to mention the generational differences. 10-core 10th gen, then 8-core 11th gen, then 4+4-core 12th gen. What's next? The return of the dual core Core i3?

 

[qubit]

That's another year until I upgrade my ancient 2700K then. <sigh>

 

[InVasMani]

Basically any modern CPU can do that, it has nothing to do with the big.LITTLE configuration.

Besides, buying a several hundred $ part to save pennies on your monthly electricity bill is kind of pointless imo.


I wouldn't use the word "stupid", but it's weird to say the least. With games starting to benefit from 6+ cores, I can't see this being Intel's new gaming flagship.

Not to mention the generational differences. 10-core 10th gen, then 8-core 11th gen, then 4+4-core 12th gen. What's next? The return of the dual core Core i3?

It's more than that you can have different chips entirely with different fixed instruction sets on each chip big.LITTLE is actually a great design, but it'll take time to evolve much like Ryzen has taken time. It's a different approach to the same concept. Where it is different and better is like I said mixed instruction sets on mixed chiplets. You could assign specific tasks to specific chiplet designs. By doing that you could have certain chiplet designs bigger and more monolithic or smaller. If you can't see the implications of that and think outside the box you'll fail to grasp what Intel is doing and/or planning with big.LITTLE in the long term as opposed to near term. Atom architecture believe it or not are good at certain tasks in regard to silicone space, heat output, and efficiency over their x86-64 Core design and vice versa. Beyond that there is also Itanium and other designs Intel has had along the way. Intel has the ability and mix and match different IP core architecture designs together with bigLITTLE and offer perks and advantages of them all. By doing that they increase efficiency of tasks performed and offer a wider array of chip designs to consumers across various market segments. It's more complex than Nvidia with numerous SKU's of the same architecture and squeezing out AMD's GPU offers, but a similar concept with more complexity than with that and a chiplet approach. It's easy to look at it right now and feel like it's underwhelming, but people looked at the initial Ryzen architecture in much the same light and same with RNDA, but the follow ups were much more exciting. We have to see how things pan out with the design and also how well they work in practice in the right hands. If the weaker ATOM chiplet allows for better usage of the CORE chiplet especially in the right hands even if not perfect, but as a proof of concept Intel most assuredly take it steps forward in the future.

I think imageCFG.exe will be quite useful for big.LITTLE's early chip design potentially especially dependent upon how well Windows Scheduler works and interacts in tandem with it. That tool allows you to change .exe's process affinity mask (hexadecimal value) to the specified executable, so that it always runs using the specified CPU cores. So you could assign a program to permanently run off the Atom core like a music player or browser or whatever anything you want to be more of a background process than foreground process you could essentially permanently assign with it to the ATOM chiplet cores so it doesn't bog down the CORE chiplet cores. I think big.LITTLE will mature in good ways quite firmly. I think eventually Intel could have a perimeter of smaller scale sized ATOM chiplets around a larger CORE chiplet.

 

[AusWolf]

It's more than that you can have different chips entirely with different fixed instruction sets on each chip big.LITTLE is actually a great design, but it'll take time to evolve much like Ryzen has taken time. It's a different approach to the same concept. Where it is different and better is like I said mixed instruction sets on mixed chiplets. You could assign specific tasks to specific chiplet designs. By doing that you could have certain chiplet designs bigger and more monolithic or smaller. If you can't see the implications of that and think outside the box you'll fail to grasp what Intel is doing and/or planning with big.LITTLE in the long term as opposed to near term. Atom architecture believe it or not are good at certain tasks in regard to silicone space, heat output, and efficiency over their x86-64 Core design and vice versa. Beyond that there is also Itanium and other designs Intel has had along the way. Intel has the ability and mix and match different IP core architecture designs together with bigLITTLE and offer perks and advantages of them all. By doing that they increase efficiency of tasks performed and offer a wider array of chip designs to consumers across various market segments. It's more complex than Nvidia with numerous SKU's of the same architecture and squeezing out AMD's GPU offers, but a similar concept with more complexity than with that and a chiplet approach. It's easy to look at it right now and feel like it's underwhelming, but people looked at the initial Ryzen architecture in much the same light and same with RNDA, but the follow ups were much more exciting. We have to see how things pan out with the design and also how well they work in practice in the right hands. If the weaker ATOM chiplet allows for better usage of the CORE chiplet especially in the right hands even if not perfect, but as a proof of concept Intel most assuredly take it steps forward in the future.

I think imageCFG.exe will be quite useful for big.LITTLE's early chip design potentially especially dependent upon how well Windows Scheduler works and interacts in tandem with it. That tool allows you to change .exe's process affinity mask (hexadecimal value) to the specified executable, so that it always runs using the specified CPU cores. So you could assign a program to permanently run off the Atom core like a music player or browser or whatever anything you want to be more of a background process than foreground process you could essentially permanently assign with it to the ATOM chiplet cores so it doesn't bog down the CORE chiplet cores. I think big.LITTLE will mature in good ways quite firmly. I think eventually Intel could have a perimeter of smaller scale sized ATOM chiplets around a larger CORE chiplet.

I partially agree with that. Atom really has its use cases. I myself have a compute stick that needs 3 Amps through a micro-USB feed and runs Windows 10. It's great for browsing, or watching videos that are supported by the iGPU's decoder. It's not so great for anything else, though. If these new desktop CPUs get 4 Atom cores with 4 Core cores, there won't be enough big cores for heavy tasks, while the small cores will offer very little help. For gamers and content creators, Atom cores are just a waste of die space. Of course we'll have to see it in action, and I'm very curious, but I'm not expecting much, at least not from this first attempt.

 

[Tom Sunday]

You will also need a need Motherboard and RAM for Alder Lake...

Some will buy Rocket Lake because they do not know what Alder Lake has in store. They cannot wait. The Premium Basics: A new Intel Core i9-11900K will run about $550. A matching Asus ROG Maximus XIII will be about $850. A RTX 3080 will be around $1000.

So essentially in the end its all about the money. Surely the great many of the so-called enthusiasts here do not have the deep pockets for the above 'premium basics.' Trolls like me from under the rock most certainly do not have the money either. For that matter who today still has real spendable income anymore?

With Alder Lake on the near horizon the hardware costs will go up even further and exponentially. Then who is buying all that hardware? Less than 1% of all worldwide computer users or enthusiasts? Remember the times of well paying US corporate jobs right out of college with 401K, pensions, paid vacations and annual bonuses are long over. Now "Work From Home" is the new norm. Show me the money!

 

[Makaveli]

Some will buy Rocket Lake because they do not know what Alder Lake has in store. They cannot wait. The Premium Basics: A new Intel Core i9-11900K will run about $550. A matching Asus ROG Maximus XIII will be about $850. A RTX 3080 will be around $1000.

So essentially in the end its all about the money. Surely the great many of the so-called enthusiasts here do not have the deep pockets for the above 'premium basics.' Trolls like me from under the rock most certainly do not have the money either. For that matter who today still has real spendable income anymore?

With Alder Lake on the near horizon the hardware costs will go up even further and exponentially. Then who is buying all that hardware? Less than 1% of all worldwide computer users or enthusiasts? Remember the times of well paying US corporate jobs right out of college with 401K, pensions, paid vacations and annual bonuses are long over. Now "Work From Home" is the new norm. Show me the money!

I've been doing this for a pretty long time and I don't know anyone that spends $500+ on motherboards unless they are doing it for work. Then its a workstation board with ECC memory.

lol $850 on a board that will not see any cpu's past rocket Lake S

 

[Tom Sunday]

$500+ on motherboards unless they are doing it for work. Then its a workstation board with ECC memory...lol $850 on a board that will not see any cpu's past rocket Lake S

"$850 on a board that will not see any cpu's past rocket Lake S"...Great statement and 100% on the money. Thanks! As to the cost of a mobo...last year I went to the computer show with a guy from Fidelity Investments who always has to have the latest and the greatest. He bought a 'GODLIKE' just for gaming. I loved the board and had to handle it like a baby. Or a dream as I told myself. So there are people out there as an exception. He will never get all of the goodies out off the GODLIKE...but that's life.

 

[RandallFlagg]

Some will buy Rocket Lake because they do not know what Alder Lake has in store. They cannot wait. The Premium Basics: A new Intel Core i9-11900K will run about $550. A matching Asus ROG Maximus XIII will be about $850. A RTX 3080 will be around $1000.

So essentially in the end its all about the money. Surely the great many of the so-called enthusiasts here do not have the deep pockets for the above 'premium basics.' Trolls like me from under the rock most certainly do not have the money either. For that matter who today still has real spendable income anymore?

With Alder Lake on the near horizon the hardware costs will go up even further and exponentially. Then who is buying all that hardware? Less than 1% of all worldwide computer users or enthusiasts? Remember the times of well paying US corporate jobs right out of college with 401K, pensions, paid vacations and annual bonuses are long over. Now "Work From Home" is the new norm. Show me the money!

You are getting really hyperbolic man. For most enthusiasts it looks more like $200-$300 CPU (11400 up to 11600K) + $200ish motherboard.

GPU prices are what will kill the DIY market soon if they don't get under control. No way would I fork up $700 to $1900 for current GPUs.

It's almost like they intentionally want to kill the desktop, and more specifically the DIY market, with component prices so high. Gaming laptops are a way, way, waaayy better deal and new laptops with 3060 / 3070 are already available, with more releasing every week or so for the next month.

 

[InVasMani]

I partially agree with that. Atom really has its use cases. I myself have a compute stick that needs 3 Amps through a micro-USB feed and runs Windows 10. It's great for browsing, or watching videos that are supported by the iGPU's decoder. It's not so great for anything else, though. If these new desktop CPUs get 4 Atom cores with 4 Core cores, there won't be enough big cores for heavy tasks, while the small cores will offer very little help. For gamers and content creators, Atom cores are just a waste of die space. Of course we'll have to see it in action, and I'm very curious, but I'm not expecting much, at least not from this first attempt.

The Atom cores are fine for a lot of the OS services a bit of compression and decompression along with rapid storage and other odds and ends and light program tasks and usage. Basically you've got a chip to offload all of that while you have another chip that can have close to 100% CPU usage dedicated to the game itself. I don't see it being perfect initially, but I'll bet you Intel improves it and the second iteration will improve quite a bit further. I don't think content creators are who Intel's trying to sell them to they want to compete with AM4 Ryzen the best they can til a more proper node shrink puts them in a better position to really compete more closely evenly with them again.

"$850 on a board that will not see any cpu's past rocket Lake S"...Great statement and 100% on the money. Thanks! As to the cost of a mobo...last year I went to the computer show with a guy from Fidelity Investments who always has to have the latest and the greatest. He bought a 'GODLIKE' just for gaming. I loved the board and had to handle it like a baby. Or a dream as I told myself. So there are people out there as an exception. He will never get all of the goodies out off the GODLIKE...but that's life.

So a guy from Fidelity Investments makes lots of money and spends cash like a drunken sailor who would have thought? He can probably even write it off on his taxes potentially if it's multipurpose and used for the job especially now with work from home being more common.

 

[Tom Sunday]

It's almost like they intentionally want to kill the desktop, and more specifically the DIY market...

For me at the time of my post and reading the newest "leaks" on the 11th generation pricing...I was dumbfounded with the Intel CPU and Z590 mobo pricing. And Alder Lake pricing will apparently not be too kind either. I think we agree that whatever may come down the road, the DIY market will be challenged like never before. The very thought about the "intentionally desktop market kill" by the manufacturers is worrisome as those are only driven by the $$$ and grabbing the largest market share. Hello DELL, HP and Lenovo!

The market seems to becoming even more interested in selling completed or 'all-In-one' systems rather than piecemeal components. It's more profitable. As we all know the major disadvantage of buying a pre-made PC is the cost. Laptop Thoughts: I now see 'more and more' so called "White Boxes" or base laptop systems hit the open consumer market. These have the base components such as chassis, screen, and motherboard installed. Users can then select items such as memory, drives, processors, and graphics to finalize the laptop computer. These basic laptop chassis historically were only sold to PC companies to then badge as their own systems after finishing off the component installations. This type of laptop marketing is nothing really new, but the sheer numbers are increasing expodentially.

Sounds to me like another grab or invasion into the desktop DIY market? I had a hard day...so I be better off to play 'Metro Exodus' for a while to flush my mind. Perhaps Mom has a leftover piece of Cherry Pie upstairs in the kitchen for yours truly?