News Intel Announces 10nm Tiger Lake Processors to Arrive in 2020

[PaulAlcorn]

Intel's new 10nm Tiger Lake processors will come to market in 2020.

Intel Announces 10nm Tiger Lake Processors to Arrive in 2020 : Read more

 

[alextheblue]

Go get em, Tiger!

...I'll see myself out.

 

[mdd1963]

more laptop CPUs....

That will soon show R9-3700X who's boss!

 

[Armani316]

Too bad Intel didn't discuss whether we'll see an H-series for laptops within the Tiger Lake CPU family. No word on Ice Lake-H series either. Strange.

 

[SgtScream]

It will still be 22nm. Trust.
Coffee Lake +++++++++++++++++++++++++++++++ anyone?

 

[JamesSneed]

They're grrrrrreat!

 

[jimmysmitty]

more laptop CPUs....

That will soon show R9-3700X who's boss!

The mobile market sells quite a bit. Most businesses now buy employees laptops/2-in-1s instead of desktops unless they have a need for it and the majority of consumers buy laptops over desktops. Its a pretty important market.

We don't really count. We are the small group in this world, the minority.

However I am sure they will have something to compete with Zen2 at some point. Its just that again the enthusiast market is not large enough to just cater to.

 

[joeblowsmynose]

more laptop CPUs....

That will soon show R9-3700X who's boss!

A couple possible reasons why Intel seems to be removing focus from the high end at this point:

They've had to do some things that I am sure they did not want to do (or got caught up in doing) just to try to stay ahead of Zen and Zen+ ... have you seen 8th and 9th gen power consumption? Almost double that of Ryzen in some workloads: https://www.tomshardware.com/reviews/intel-core-i9-9900k-9th-gen-cpu,5847-11.html - this is still almost a secret as if you asked almost anyone, they would tell you that AMD is hot and consumes more power than equivalent Intel processors, but in this one area Intel has had to sacrifice their very long lived efficiency trophy for clock speeds in an attempt to maintain the perception of staying ahead of AMD ... what will be next on that chopping block? Then next? It might be a lesser loss to brand image to give up the winner's trophy in this one area, and focus on gaining new or ensuring you keep the other trophies in the other areas ... one of those other areas seems to be mobile for Intel right now - there is nor doubt there is money to be had there, and AMD is putting more focus in the high end and less on mobile so competition at this end is much lighter.

The other aspect is that they are working on new designs and new nodes, and things haven't been going great. It would be much easier to work out any issues by focusing on the low power and low core parts first, then you can scale up the improvements to the larger and faster chips which would take much more time to get the improvements out of (thus more loss) than focusing on the mobile first.

By judging the way things look currently, I think Intel might just be strategically (finally) willing to not try to beat AMD in every area, but pick and choose the areas where there is least risk and most gain to be had - it appears this is mobile. Once they have that worked out, they will have invested a fair bit of time and research into power efficiency into these new designs, which they can then use to improve the higher end chips without having put up with the embarrassingly "fake" TDP number on their processors (9900k - yeah I'm looking at you!).

 

[Dark Lord of Tech]

Intel will remain King.

 

[Saadish]

i want to buy a powerful PC in winter 2020. i want intel Tiger Lake cpu with minimum 10 core and minimum 5.0Ghz base performance. if intel make late or bad again i have to buy AMD.

 

[mdd1963]

'minimum 5 GHz base'?

Certainly base clock will be much lower, likely at 3.7 GHz or so to meet reasonable TDP limits.

You might get your wish for one or two cores to turbo as high as 5 or 5.1 GHz or so... as to how substantially improved this will be over the 8c/16t 9900k in gaming, this remains to be seen. (Certainly CInebench should see a roughly 20-25% hypothetical improvement with a 25% core count increase)

 

[SgtScream]

That would be so dope if 5ghz was base. I'd definitely buy intel if they did that.

 

[InvalidError]

However I am sure they will have something to compete with Zen2 at some point. Its just that again the enthusiast market is not large enough to just cater to.

The desktop market is still large enough. Intel's problem here is that with Ryzen in the picture, the lower-end isn't anywhere near as profitable as it used to be anymore so Intel is focusing on higher-value products in the laptop and server space now that it no longer has the process advantage required to produce huge profit margins on competitively priced mainstream parts.

 

[jimmysmitty]

The desktop market is still large enough. Intel's problem here is that with Ryzen in the picture, the lower-end isn't anywhere near as profitable as it used to be anymore so Intel is focusing on higher-value products in the laptop and server space now that it no longer has the process advantage required to produce huge profit margins on competitively priced mainstream parts.

Intel has always focused on servers and the mobile market. Ultrabooks are an Intel design after all.

My point is that the enthusiast, us people of TH, are not a large enough market and never have been to cater directly to us. They will focus on the markets they still have more control in like mobile and HPC.

 

[InvalidError]

My point is that the enthusiast, us people of TH, are not a large enough market and never have been to cater directly to us.

Depends on the definition of enthusiast. Last I heard, gaming PCs were still a growth market and most DIY PC parts manufacturers are slapping "gaming" in their branding because of it even on parts that technically don't give a damn what the system will ultimately be used for. Smells like catering to self-proclaimed enthusiasts (and even some non-enhusiasts) to me, just not coming from Intel.

Intel's problem is that it got used to absurdly large gross profit margins and with Ryzen in the picture, it can't get those from most of the DIY crowd anymore. Unless show-stopper issues get found in EPYC now that more companies are adopting it, it is only a matter of time before Intel gets forced to trim profit margins in the server space too.

 

[jimmysmitty]

Depends on the definition of enthusiast. Last I heard, gaming PCs were still a growth market and most DIY PC parts manufacturers are slapping "gaming" in their branding because of it even on parts that technically don't give a damn what the system will ultimately be used for. Smells like catering to self-proclaimed enthusiasts (and even some non-enhusiasts) to me, just not coming from Intel.

Intel's problem is that it got used to absurdly large gross profit margins and with Ryzen in the picture, it can't get those from most of the DIY crowd anymore. Unless show-stopper issues get found in EPYC now that more companies are adopting it, it is only a matter of time before Intel gets forced to trim profit margins in the server space too.

The DIY market has never been the profit margin leader even with growth in the market. They have always made their money on server products. Its why HPC, server and extreme desktop markets get the newest tech typically first.

That said, EPYC will gain some market but I don't know if it will be enough to squeeze Intel that much. They had nearly 99% of the HPC market and I have talked to a few IT professionals that have no plans in moving back to AMD. There are multiple reasons from not wanting to spend the money to upgrade all the servers to avoid possible VM issues to being burnt from the last time AMD left the server market.

That said, Intel looks to be planning 10nm in the server market very soon plus they have their new chip stacking which might help to make them more competitive on a per core level.

 

[InvalidError]

Its why HPC, server and extreme desktop markets get the newest tech typically first.

Hm, how about no? Historically, server chips were one generation behind mainstream because Intel used mainstream to test new architectures in the wild before bringing them to server. Intel's current server CPUs are based on Cascade Lake which is the successor to Skylake-X, which puts it about one generation behind Coffee on the generational scale, though the evolutionary split after Skylake makes it unclear how much of Coffee and other improvements may have been rolled into it.

 

[jimmysmitty]

Hm, how about no? Historically, server chips were one generation behind mainstream because Intel used mainstream to test new architectures in the wild before bringing them to server. Intel's current server CPUs are based on Cascade Lake which is the successor to Skylake-X, which puts it about one generation behind Coffee on the generational scale, though the evolutionary split after Skylake makes it unclear how much of Coffee and other improvements may have been rolled into it.

Except Optane DIMMs, The Mesh interconnect (or the ring bus before that), Nehalem, Intels first IMC, etc. uArch, yes. Technologies? No.

 

[JayNor]

Six months later ... We still don't know much about the integrated Xe gpu in the Tiger Lake chip, except that it is 96 EU vs the 64 EU gen11 gpu on Ice Lake.

There is info confirming the cache sizes growing in Tiger Lake.

I haven't seen any info about going Foveros 3D on Tiger Lake, but if Lakefield is already ramping up with that 3D manufacturing, why wouldn't they use it in Tiger Lake?

 

[InvalidError]

why wouldn't they use it in Tiger Lake?

At the lower end of the scale, I'd imagine the cost overheads of mixing multiple packaging technologies together to be too high. When you go multi-die packaging, each die needs to allocate extra area and power on chip-to-chip IO on top of the more complex substrate to tie them together. Multi-chip does not make sense until the design is too large to be economically viable as a single die or involves components that have to be made on a significantly different process such as DRAM.