Tuesday, January 14th 2025

AMD Implements New CCD Connection in "Strix Halo" Ryzen AI Max Processors

Thanks to the informative breakdown by Chips and Cheese, we are learning that AMD's latest Ryzen AI processors for laptops, codenamed "Strix Halo," utilize a parallel "sea of wires" interconnect system between their chiplets, replacing the SERDES (serializer/deserializer) approach found in desktop Ryzen models. The processor's physical implementation consists of two Core Complex Dies (CCDs), each manufactured on TSMC's N4 (4 nm) process and containing up to eight Zen 5 cores with full 512-bit floating point units. Notably, the I/O die (IOD) is also produced using the N4 process, marking an advancement from the N6 (6 nm) process used in standard Ryzen IODs on desktops. The key change lies in the inter-chiplet communication system. While the Ryzen 9000 series (Granite Ridge) employs SERDES to convert parallel data to serial for transmission between chiplets, Strix Halo implements direct parallel data transmission through multiple physical connections.

This design achieves 32 bytes per clock cycle throughput and eliminates the latency overhead associated with serialization/deserialization processes. The parallel interconnect architecture also removes the need for connection retraining during power state transitions, a limitation present in SERDES implementations. However, this design choice necessitates additional substrate complexity due to increased connection density and requires more pins for external connections, suggesting possible modifications to the CCD design compared to desktop variants. AMD's implementation required more complex substrate manufacturing processes to accommodate the dense parallel connections between chiplets. The decision to prioritize this more challenging design approach was driven by requirements for lower latency and power consumption in data-intensive workloads, where consistent high-bandwidth communication between chiplets is crucial.
Sources: Chips and Cheese, via HardwareLuxx
Add your own comment

24 Comments on AMD Implements New CCD Connection in "Strix Halo" Ryzen AI Max Processors

#2
claster17
Hopefully this will find its way into whatever Zen6 desktop CPUs will be called and fix the awful IOD idle power consumption we had ever since Zen2.
Posted on Reply
#3
rv8000
claster17Hopefully this will find its way into whatever Zen6 desktop CPUs will be called and fix the awful IOD idle power consumption we had ever since Zen2.
Yeah, 20w of idle power is really breaking the bank! :rolleyes:

I’m more concerned about the implications of requiring more socket pins, and if that’s the case and zen6 uses a new IO die we get:

1) New socket

or

2) Design compromises to meet current AM5 socket pinout.
Posted on Reply
#4
LittleBro
rv8000Yeah, 20w of idle power is really breaking the bank! :rolleyes:

I’m more concerned about the implications of requiring more socket pins, and if that’s the case and zen6 uses a new IO die we get:

1) New socket

or

2) Design compromises to meet current AM5 socket pinout.
Why would new IOD require new socket? Why would it require more pins? Explain, please.

Anyway, Zen 6 will have new IOD, finally.
Posted on Reply
#5
rv8000
LittleBroWhy would new IOD require new socket? Why would it require more pins? Explain, please.

Anyway, Zen 6 will have new IOD, finally.
Read the news post.
Posted on Reply
#6
Zach_01
rv8000I’m more concerned about the implications of requiring more socket pins, and if that’s the case and zen6 uses a new IO die we get:

1) New socket

or

2) Design compromises to meet current AM5 socket pinout.
I hope not, and in OP talks about more connections per CCD with the substrate. Not the CPU with socket.
Posted on Reply
#7
rv8000
Zach_01I hope not, and in OP talks about more connections per CCD with the substrate. Not the CPU with socket.
OP talks about “more pins for external connections”, overall unclear but I would assume socket pins over internal the way it’s worded. It could also mean both.

*Socket FP11 has 2077 pins and AM5 has 1718 pins, obviously the new AI Max processors support quad channel memory, but unsure how that alone directly effects socket pin count differences.
Posted on Reply
#8
usiname
rv8000OP talks about “more pins for external connections”, overall unclear but I would assume socket pins over internal the way it’s worded. It could also mean both.

*Socket FP11 has 2077 pins and AM5 has 1718 pins, obviously the new AI Max processors support quad channel memory, but unsure how that alone directly effects socket pin count differences.
More memory channels require more pins. Also more PCI line increase the required pins, but this is not the case here, Its the connection between the CCDs and the IOD which should not affect the pins count.
Posted on Reply
#9
Zach_01
From OP:
"However, this design choice necessitates additional substrate complexity due to increased connection density and requires more pins for external connections, suggesting possible modifications to the CCD design compared to desktop variants"

All I read was more dense and complex substrate for more parallel connections.
This will require changes to CCDs and IOD.

No word for socket pins count

Probably the word pins is what throws you out.
But chiplets can have pins(connections) to the substrate.

Hopefully...
Posted on Reply
#10
rv8000
Zach_01From OP:
"However, this design choice necessitates additional substrate complexity due to increased connection density and requires more pins for external connections, suggesting possible modifications to the CCD design compared to desktop variants"

All I read was more dense and complex substrate for more parallel connections.
This will require changes to CCDs and IOD.

No word for socket pins count
Posted on Reply
#11
dragontamer5788
The pattern of Parallel > Serial > Parallel > Serial > Parallel continues ehhh?

It always amazes me that communications swaps back and forth between these two methodologies so much.
Posted on Reply
#12
LittleBro
rv8000Read the news post.
IOD itself should not need more pins or changes in socket whatsoever.
As Zach pointed out, those extra Strix Halo pins are required for additional memory channels, NPU and maybe GPU, or not?
There's no need to increase pin count in order to implement new IOD, because that's CCD interconnection thing (substrate thing).
If new IOD required more pins, then that would literally mean end of AM5. However, AMD commited to hold it alive till at least 2027.

I think we will see modified version of this Strix Halo's IOD in Zen 6.
Posted on Reply
#13
dragontamer5788
LittleBroIf new IOD required more pins, then that would literally mean end of AM5. However, AMD commited to hold it alive till at least 2027.
Strix Halo is obviously not-AM5. 256-bit wide LPDDR5x is enough to prove that.
Posted on Reply
#14
LittleBro
dragontamer5788Strix Halo is obviously not-AM5. 256-bit wide LPDDR5x is enough to prove that.
No one said that it is. I'm talking about AMD reusing/modifying this Halo's IOD for Zen 6 purposes would not require new pins or socket.
Posted on Reply
#15
ymdhis
Give me this for desktop.
Posted on Reply
#16
dragontamer5788
ymdhisGive me this for desktop.
Strix Halo is intended for 45W laptops to 120W Desktops btw.

45W is "desktop replacement laptop", very high power usage for laptops but technically doable.
Posted on Reply
#17
Zach_01
dragontamer5788Strix Halo is intended for 45W laptops to 120W Desktops btw.

45W is "desktop replacement laptop", very high power usage for laptops but technically doable.
45W sounds enough but this thing can have 8-16 and 32-40CUs
50W is for breakfast...
40CU x64 = 2560 shaders



2560 shaders is between 7600XT and 7700XT

Thats why it needs to be the whole thing on 4nm node
Wont be cheap either.
Posted on Reply
#18
Daven
dragontamer5788Strix Halo is intended for 45W laptops to 120W Desktops btw.

45W is "desktop replacement laptop", very high power usage for laptops but technically doable.
A laptop Geforce 4090 alone goes up to 175W so I'm not sure what you are talking about.
Posted on Reply
#19
rv8000
LittleBroIOD itself should not need more pins or changes in socket whatsoever.
As Zach pointed out, those extra Strix Halo pins are required for additional memory channels, NPU and maybe GPU, or not?
There's no need to increase pin count in order to implement new IOD, because that's CCD interconnection thing (substrate thing).
If new IOD required more pins, then that would literally mean end of AM5. However, AMD commited to hold it alive till at least 2027.

I think we will see modified version of this Strix Halo's IOD in Zen 6.
The OP clearly states “external pin connections” as highlighted a few posts up, also implying pin count differences between FP11 and AM5 are not exclusively due to dual vs quad channel differences. Whether or not the OP is incorrect, vague, or otherwise summarized the chips and cheese info improperly, it may or may not require pin out changes.

That’s why I listed two scenarios for zen 6, one being a different socket, the other a modified design which could end up with a less efficient (data/speed) wise to accommodate socket limitations.

I think the wording of the OP explicitly implies pin count changes, why would you mention “external” otherwise. Another question would be how many pins are currently not in use on AM5 that can be repurposed to accommodate potential changes.
Posted on Reply
#20
LittleBro
I'd vote for the latter scenario you mentioned. Otherwise, many people (incl. myself) would get angry at AMD for breaking commitments and changing the socket too early.
Posted on Reply
#21
rv8000
LittleBroI'd vote for the latter scenario you mentioned. Otherwise, many people (incl. myself) would get angry at AMD for breaking commitments and changing the socket too early.
While it seems more likely at this point Zen 6 will come on AM5, socket support till 20XX doesn’t necessarily mean new designs as seen with parts like the 5700X3D, 5900XT, etc… It can mean several different things, and certainly doesn’t guarantee a new design on the socket.

I also wonder if they plan to make use of cu-dimms at any point, as this route is infinitely better than the disaster/failure that is camm; it’s currently unclear if cu-dimm support will be dependent on just cpu arch design updates or also require motherboard/socket changes. I don’t know enough about this to really say.
Posted on Reply
#22
JustBenching
claster17Hopefully this will find its way into whatever Zen6 desktop CPUs will be called and fix the awful IOD idle power consumption we had ever since Zen2.
Amen, browsing the web at 60w will be a thing of the past. Hopefully.
Posted on Reply
#23
Wirko
dragontamer5788The pattern of Parallel > Serial > Parallel > Serial > Parallel continues ehhh?

It always amazes me that communications swaps back and forth between these two methodologies so much.
It's not just back and forth, multi-lane interfaces such as PCI Express are serial and parallel at the same time.

AMD's IFOP ... not that we know many details about it but it seems to be a set of 40+32+1 serial interfaces. 40 bits from IOD to CCD, 32 bits from CCD to IOD, and clock, working at 8 GT/s with a 2 GHz FCLK clock.

chipsandcheese.com/p/pushing-amds-infinity-fabric-to-its
Posted on Reply
#24
Visible Noise
That interview is unreadable. These guys need to hire an editor.
Posted on Reply
Add your own comment
Jan 14th, 2025 21:06 EST change timezone

New Forum Posts

Popular Reviews

Controversial News Posts