Intel "Tiger Lake" Architecture Combines Willow Cove CPU Cores and Xe iGPU

[btarunr]

Even as Intel banks on 10 nm "Ice Lake" to pull it out of the 14 nm dark ages, the company is designing a fascinating new monolithic processor SoC die that succeeds it. Codenamed "Tiger Lake," and slated to debut in 2020, this die packs "Willow Cove" CPU cores and an iGPU based on Intel's Xe architecture, not Gen11. "Willow Cove" CPU cores are more advanced than the "Sunny Cove" cores "Ice Lake" packs, featuring a redesigned on-die cache, additional security features, and transistor optimization yielded from the newer 10 nm+ silicon fabrication process.

Intel is already boasting of 1 TFLOP/s compute power of the Gen11 iGPU on "Ice Lake," so it's logical to predict that the Xe based iGPU will be significantly faster. It will also support the latest display standards. The "next-gen I/O" referenced by Intel could be faster NVMe, Thunderbolt, and USB standards that leverage the bandwidth doubling brought about by PCI-Express gen 4.0. Here's the catch: much like "Ice Lake," the new "Tiger Lake" chip will get a mobile debut as Tiger Lake-Y or Tiger Lake-U, and desktop processors could follow later, possibly even 2021, depending on how much pressure it faces from AMD.



View at TechPowerUp Main Site

 

[Caring1]

Investor meeting slide says it all, keep pumping them up or they will deflate and blow away from all the letdowns.

 

[AceKingSuited]

So the first Xe is just vaporware at this point and they already have specs for the second gen Xe? What a pile of crap, lol. Oh my where the mighty has fallen?? Intel is definitely learning from Musk on how to make broken promises.

 

[londiste]

So the first Xe is just vaporware at this point and they already have specs for the second gen Xe? What a pile of crap, lol. Oh my where the mighty has fallen?? Intel is definitely learning from Musk on how to make broken promises.

Roadmap usually includes things on the roadmap that are planned. Gen11 is an intermediary product between current Gen9/9.5 and Xe. The changes are probably not that large architecturally when it comes to iGPUs and naming is just marketing.

We know AMD is planning Navi and Arcturus after it. Having multiple things on roadmap does not automatically make things vapourware. I would also bring example from Nvidia but we only know Ampere from them at this point.

 

[tfdsaf]

So what is new here? Their 10nm process is still slated for 2020 as we've known for two years now, and it's only coming to mobile because its too unrefined to do any sort of bigger dies than for small mobile chips.

Nothing new and certainly nothing that is going to make any difference.

 

[AceKingSuited]

Roadmap usually includes things on the roadmap that are planned. Gen11 is an intermediary product between current Gen9/9.5 and Xe. The changes are probably not that large architecturally when it comes to iGPUs and naming is just marketing.

We know AMD is planning Navi and Arcturus after it. Having multiple things on roadmap does not automatically make things vapourware. I would also bring example from Nvidia but we only know Ampere from them at this point.

They will never hit this road map if we are judging by their recent developments. Also, I think their GPU is going to fail spectacularly. It took NVDA and AMD decades to get their GPUs where they are now and they hold all the patents for GPUs. I just don't see how a intel GPU can make headway. It's like if NVDA had a x86 license and they decide to go into x86.

 

[londiste]

Also, I think their GPU is going to fail spectacularly. It took NVDA and AMD decades to get their GPUs where they are now and they hold all the patents for GPUs. I just don't see how a intel GPU can make headway.

Intel has had a fairly decent iGPU architecture for a while now, at least since Sandy Bridge. Whether and how well they can improve it and scale it up remains to be seen but they are not new to GPUs.
Patents should not be a problem for Intel. They have extensive cross-licensing agreements with both AMD and Nvidia and they have been doing R&D on GPUs throughout years including some of their own patents.

 

[Midland Dog]

They will never hit this road map if we are judging by their recent developments. Also, I think their GPU is going to fail spectacularly. It took NVDA and AMD decades to get their GPUs where they are now and they hold all the patents for GPUs. I just don't see how a intel GPU can make headway. It's like if NVDA had a x86 license and they decide to go into x86.

except nvidia would probably do a good job lol

 

[SoNic67]

Lots of Intel haters here...

 

[Vayra86]

except nvidia would probably do a good job lol

Tegra disagrees with that... I have fond memories of my LG Optimus 2X., the slowest dualcore phone e-ver.

Lots of Intel haters here...

What's there to love about Intel, in your mind? They have nothing truly promising in the pipeline and we've been seeing the same slides reshuffled for the last 6 months. Oh, and they've just produced a new slide saying they will push data center first; so for consumer, that means we'll be looking that much more and longer at sub-optimal CPUs.

By the way, 'nothing there to admire' is not the same as 'hate'. Hating happens in Youtube emo videos.

 

[kings]

Also, I think their GPU is going to fail spectacularly.

I sincerely hope not, the market needs a strong Intel in GPUs.

If we only count on AMD to shadow Nvidia, we'll have dark years ahead of us in that chapter.

 

[prtskg]

Roadmap usually includes things on the roadmap that are planned. Gen11 is an intermediary product between current Gen9/9.5 and Xe. The changes are probably not that large architecturally when it comes to iGPUs and naming is just marketing.

We know AMD is planning Navi and Arcturus after it. Having multiple things on roadmap does not automatically make things vapourware. I would also bring example from Nvidia but we only know Ampere from them at this point.

Your post about roadmap is right. Just like to add that Arcturus is a chip name not architecture name. An AMD employee has already told so at Phoronix.

I sincerely hope not, the market needs a strong Intel in GPUs.

If we only count on AMD to shadow Nvidia, we'll have dark years ahead of us in that chapter.

I doubt 1st gen consumer dgpu from Intel will touch Nvidia products. Nvidia is quite good IMO.

 

[Midland Dog]

Tegra disagrees with that... I have fond memories of my LG Optimus 2X., the slowest dualcore phone e-ver.



What's there to love about Intel, in your mind? They have nothing truly promising in the pipeline and we've been seeing the same slides reshuffled for the last 6 months. Oh, and they've just produced a new slide saying they will push data center first; so for consumer, that means we'll be looking that much more and longer at sub-optimal CPUs.

By the way, 'nothing there to admire' is not the same as 'hate'. Hating happens in Youtube emo videos.

tegra was arm not x86, and tbh was a half arsed effort, the gpu side of tegra was pretty good tho, look at the custom soc the switch uses, pretty sure thats tegra based and it performs quite good for the form factor

 

[Vayra86]

tegra was arm not x86, and tbh was a half arsed effort, the gpu side of tegra was pretty good tho, look at the custom soc the switch uses, pretty sure thats tegra based and it performs quite good for the form factor

That's all fine, but it was mostly to underline that not everything Nvidia touches turns into green gold. And their CPU business is the perfect example. Their ARM design wasn't great. Tegra was slow crap in its first iterations and the versions that followed later (with Denver cores/custom soc) were power hungry. Fast, but hungry. They are selling Tegra but they barely recoup the expenses up to this point.

 

[londiste]

This gets offtopic but Tegra is an interesting part-success and part-failure. Recent Tegra iterations are pretty good technically but struggle to find a place in the market as they are too powerful, power-hungry and expensive for widespread use and for specific devices manufacturers who need a SoC this powerful tend to create their own. Switch is pretty much the only widespread success that a Tegra-based device has had.

GPU part of Tegras has always been at least good enough, usually very good. Matches what you would expect from Nvidia who are GPU company after all.

CPU part is the problem. History of Tegra is closely related to Project Denver which was a failure and took years to get somewhere. Tegra was intended to have Denver for CPU (which is an interesting bit in CPU history) but the design for that took more and more time so Nvidia hastily replaced the CPU with ARM cores and not always good and suitable ones. There is a lot unknown in this but where Denver started was Transmeta-like instruction set translation from x86 into its own internal instructions. There is suspicion that current iterations simply do the same for ARM. This was a part of Nvidia's attempt in going for x86 which failed not so much for technical reasons but largely because Nvidia was unable to secure x86 licensing from Intel.

pretty sure thats tegra based and it performs quite good for the form factor

It does perform quite good. The problem is that this form factor is niche. It is a tablet SoC and a big, power-hungry one at that. There is no market for this. Tablets are in decline and market is in smaller, lighter tablets that do not need that much power. Nvidia is struggling to find a use for Tegra - it goes into automotive, AI and other specific use cases but not with much success. On the other hand, they have not given up on Tegra yet.

Now that I think of it, this might not be as offtopic as it seems. Lakefield on that slide is Intel's next attempt at SoC. What market it is aimed at, I am not sure but low-power laptops and big tablets might fit the bill.