Gigantic LGA 9324 Socket Test Interposer For Intel's Future "Diamond Rapids-AP" Xeons Spotted

[Fouquin]

Intel has begun sampling the test tools for their "Oak Stream" platform which will house the "Diamond Rapids" generation of processors sometime in late 2025 or early 2026. Previously rumored to continue using the "Birch Stream" platform LGA 7529 socket that will soon be shipping with the 288-core flavor of the "Sierra Forest" efficiency core Xeons as well as 120-core "Granite Rapids" performance core Xeons, "Diamond Rapids" appears to instead be moving up to a substantially larger LGA 9324 socket. This is Intel's next-next generation of Xeon from what is shipping today, following up on the next-gen Intel 18A based "Clearwater Forest" which was only just reported to be powering on earlier this month. Other than the codename there is almost nothing currently known about "Diamond Rapids" but the rumor mill is already fired up and mentioning things such as increased core counts, 16 DRAM channels (similar to what AMD is expected to introduce with EPYC "Venice") and PCI-E 6.0 support.

The LGA 9324 test interposer for use with Intel's Gen 5 VR Test Tool that appeared on their Design-in Tools storefront before the page went to a 404 error carried a price tag of $900 USD and stipulated that this was a pre-order with an expected shipment date in Q4 2024.



View at TechPowerUp Main Site | Source

 

[Vincero]

Seeing as LGA 7529 was almost as wide as a DIMM (pictured), would love to see how big that LGA 9324 is...

 

[P4-630]

And then accidently dropping a hammer on the socket....

 

[Wirko]

Obviously this is the combined result of many departments' hard work, each of them requesting several hundred more pins for various new features, and no one holding them back...

 

[Vincero]

Obviously this is the combined result of many departments' hard work, each of them requesting several hundred more pins for various new features, and no one holding them back...

If it's anything like the desktop socketed parts, a large percentage are power delivery.
That said, 16-DRAM channels is 8x the average desktop so there could be maybe 1000 pins just for that, as well as still leaving scope for dozens of extra PCIe lanes, and QPI style links to other CPUs for multi socket systems.

 

[ZoneDymo]

And then accidently dropping a hammer on the socket....

too definitive, dropping a screwdriver seems more painful, just a handful of pins messed up....but now completely useless

 

[LabRat 891]

I'd poke fun at over 9000
but that's nightmarish...

 

[Ruru]

PCIe 6.0 support? We just have fully adapted PCIe 4.0 into the mainstream, 5.0 for the fastest enthusiast-level SSDs and they're already pushing a new gen out? Damn.

 

[LabRat 891]

PCIe 6.0 support? We just have fully adapted PCIe 4.0 into the mainstream, 5.0 for the fastest enthusiast-level SSDs and they're already pushing a new gen out? Damn.

AFAIK this is about Intel's Industry-only Halo-tier Server/HPC/AI market.
They demand bleeding edge+1.

 

[Ruru]

AFAIK this is about Intel's Industry-only Halo-tier Server/HPC/AI market.
They demand bleeding edge+1.

Makes sense. I just didn't know that gen6 is already ready

 

[Vincero]

Also, this product isn't actually out yet so still time to work on implementing it.
But yes, at this level PCIe 6 would be expected in the data centre for high bandwidth links to storage and data connections.

 

[Lawrencetate2024]

Intel LGA 9324 Socket​

This is intel latest "Monster Xeon" Socket with also 1000 pins.

With Data Centers worldwide run currently at their peak,

This CPU that fits into this socket will keep all data centers happy!

This CPU will be power hungry, not to mention all that memory &
Data I/O will create some real heat!

This will do some business with this one, that if Intel can get it to
work, it would leave others in the dust for sure.

If this huge CPU runs right; it is going to do allot of high end
technical applications perform some unbelievable tasks!

They will need to come up with a real relievable cooling freezer just to keep it cool.

I would love to see some of the new customers come up with any application
that would slow this one down!

I do not think so!

This is the game changer for sure!!!!!

This one will be very expensive!!!!!!!

Larry

 

[ssateneth]

Soon, CPU's will need a screw through the center of the CPU in order to give balanced mounting pressure, since all the pins will press the CPU away from the socket and the only mounting pressure is from the edges of the CPU IHS.

Maybe this can be mitigated by having the heatsink act as mounting pressure and having a convex surface on the CPU IHS or heatsink surface, but it would need both mounting pressure from the socket mounting mechanism edges AND the heatsink for mounting pressure on the center of the CPU

 

[LabRat 891]

Maybe this can be mitigated by having the heatsink act as mounting pressure and having a convex surface on the CPU IHS or heatsink surface, but it would need both mounting pressure from the socket mounting mechanism edges AND the heatsink for mounting pressure on the center of the CPU

To my knowledge, at least AMD already does this. IIRC, Intel factors 'flex' into their IHS tolerances too. It is indeed a "mitigation", not a solution.