Friday, February 18th 2022

Intel "Meteor Lake" and "Arrow Lake" Use GPU Chiplets

Intel's upcoming "Meteor Lake" and "Arrow Lake" client mobile processors introduce an interesting twist to the chiplet concept. Earlier represented in vague-looking IP blocks, new artistic impressions of the chip put out by Intel shed light on a 3-die approach not unlike the Ryzen "Vermeer" MCM that has up to two CPU core dies (CCDs) talking to a cIOD (client IO die), which handles all the SoC connectivity; Intel's design has one major difference, and that's integrated graphics. Apparently, Intel's MCM uses a GPU die sitting next to the CPU core die, and the I/O (SoC) die. Intel likes to call its chiplets "tiles," and so we'll go with that.

The Graphics tile, CPU tile, and the SoC or I/O tile, are built on three different silicon fabrication process nodes based on the degree of need for the newer process node. The nodes used are Intel 4 (optically 7 nm EUV, but with characteristics of a 5 nm-class node); Intel 20A (characteristics of 2 nm), and external TSMC N3 (3 nm) node. At this point we don't know which tile gets what. From the looks of it, the CPU tile has a hybrid CPU core architecture made up of "Redwood Cove" P-cores, and "Crestmont" E-core clusters.
The Graphics tile packs an iGPU based on the Xe LP graphics architecture, but leverages an advanced node to significantly increase the execution unit (EU) count to 352, and possible increase graphics clocks. The SoC and I/O tile packs the platform security processor, integrated northbridge, memory controllers, PCI-Express root-complex, and the various platform I/O.

Intel is preparing "Meteor Lake" for a 2023 launch, with development completing within 2022, although mass-production might still commence next year.
Add your own comment

36 Comments on Intel "Meteor Lake" and "Arrow Lake" Use GPU Chiplets

#26
Ruru
S.T.A.R.S.
sam_86314EDIT: Apparently the old Core 2 Quads had multiple dies.



Picture from @Ruslan

I knew that the older Pentium D chips also had multiple dies.

Both dies were processor dies. This was back in the days of having the northbridge on the motherboard.
Yeah the Pentium D 900 series had multiple dies, D 800 had one big die (put basicly a dual Pentium 4 500 series)
Posted on Reply
#27
Wirko
TiggerSo Intel had the great idea of chiplets/tiles as early as clarkdale in 2010, did they go back to a single die after this? If so I wonder why. As long as the interconnect between them is fast enough it's a great setup as seen by Ryzen.

What is the interface between the tiles on these?
Nephilim666At what point does the cost and yeild benefit disappear with all this complex alignment and 'gluing' of chipsets tiles?
It's not some fixed point. Chip manufacturing is advancing, albeit slowly, price per transistor is heading upwards instead of downwards, and there's a size limit of 26x33 mm. Meanwhile, packaging tech is advancing at an amazing rate - see Intel, TSMC, Cerebras - but in my opinion, exotic tech comes at an exotic price. So it's all dictated by economics... like ever before.

Edit: Hey, can anyone identify this price-no-object processor called Itanium? It has four dies on a single substrate. They can't be anything else but cache - it's from a period when cache was separate from the CPU, like in Pentium II and III. Source: chipscapes.com/products/intel-generations
Posted on Reply
#28
thestryker6
Wirko...

Edit: Hey, can anyone identify this price-no-object processor called Itanium? It has four dies on a single substrate. They can't be anything else but cache - it's from a period when cache was separate from the CPU, like in Pentium II and III. Source: chipscapes.com/products/intel-generations
That would be either an Itanium 733 or 800 with 4 MB L3 cache. I believe the first generation were the only ones to feature the separated cache.

I've often wondered what the landscape would look like if Intel had been able to shift to IA-64 and leave x86 behind. AMD was smart to design the extension to x86 with AMD64, because of the state of the software market, but a clean break could have changed everything.
Posted on Reply
#29
Wirko
thestryker6That would be either an Itanium 733 or 800 with 4 MB L3 cache. I believe the first generation were the only ones to feature the separated cache.
You're right. So Intel (and HP) did glue several chips on a common substrate as early as 2001. IBM and others did it before (early examples I found on a Russian site). Maybe IBM had EMIB but didn't tell anyone, and those who cut IBM modules apart didn't notice anything because it was too advanced.
thestryker6I've often wondered what the landscape would look like if Intel had been able to shift to IA-64 and leave x86 behind. AMD was smart to design the extension to x86 with AMD64, because of the state of the software market, but a clean break could have changed everything.
In 2026, thanks to Itanium E-cores, notebooks would finally become as thin as they used to be in 2000.
Posted on Reply
#30
thestryker6
WirkoYou're right. So Intel (and HP) did glue several chips on a common substrate as early as 2001. IBM and others did it before (early examples I found on a Russian site). Maybe IBM had EMIB but didn't tell anyone, and those who cut IBM modules apart didn't notice anything because it was too advanced.
During that era everything would have been connected by a bus (most likely dedicated between cache and core) as the process node technology wasn't good enough to even imagine something like EMIB.
WirkoIn 2026, thanks to Itanium E-cores, notebooks would finally become as thin as they used to be in 2000.
:D You're not wrong with how terribly they started out.
Posted on Reply
#31
Cutechri
SteevoMy 90nm identifies as 1nm and failure to respect that is hate crime.
Ugh.
Posted on Reply
#32
MikeMurphy
TheoneandonlyMrKI can't believe they're spinning this as new , I mean a reasonably useful GPU in an Intel CPU would be new, but Intel doing an McM with a CPU and GPU is so last decade.

And a few weeks ago they were buying their competitors GPU to place on their McM.(sarcasm :p)

In fact the only new bit is Intel has decided to nick AMD'S McM IO die concept, no?!.
Integrated performance demands are much higher now than in 2010, and with big appetites for many cores it makes sense to move the the large GPU die off of the CPU die.
Posted on Reply
#33
TheoneandonlyMrK
MikeMurphyIntegrated performance demands are much higher now than in 2010, and with big appetites for many cores it makes sense to move the the large GPU die off of the CPU die.
I wasn't against it ?

Just it's as new as my bad knee.
Posted on Reply
#34
Richards
This is like alien technology from intel
Posted on Reply
#35
Wirko
MikeMurphyIntegrated performance demands are much higher now than in 2010, and with big appetites for many cores it makes sense to move the the large GPU die off of the CPU die.
With MCM, Intel can also fracture the supply chain in ways unimaginable in 2010, and have parts of a single product partly made in India, Germany, China, Ireland, Mexico, Italy, Japan, Malaysia, US, South Korea, Israel, Taiwan and Jamaica. It's cheaper this way.
Posted on Reply
#36
TheoneandonlyMrK
RichardsThis is like alien technology from intel
Say what now?!.

Which time around ,this time or last time?!
Posted on Reply
Add your own comment
Nov 25th, 2024 23:51 EST change timezone

New Forum Posts

Popular Reviews

Controversial News Posts