22.1 C

Intel’s Ponte Vecchio to be its first and foremost Xe-HPC based GPU

During the recent Intel unleashed webcast, Intel’s CEO, Pat Gelsinger, unveiled new details of the 7nm Xe-HPC-based Ponte Vecchio GPU. According to sources, the Intel Ponte Vecchio GPU is first and foremost based on the Xe-HPC graphics architecture which is the flagship product leveraging Intel’s 7nm EUV process node.

The chip also has a ton of other compute tiles that are based on different process nodes, and they merge into one singular exascale graphics processing unit known as Ponte Vecchio. From what we know so far, the GPU primarily makes use of Intel’s 7nm EUV process node but Intel will also be producing some Xe-HPC compute dies through external fabs (such as TSMC).

Raja Koduri teased that there are 7 advanced technologies here, and these would be:

- Advertisement -TechnoSports Ad
  • Intel 7nm
  • TSMC 7nm
  • Foveros 3D Packaging
  • EMIB
  • 10nm Enhanced Super Fin
  • Rambo Cache
  • HBM2

Following is how Intel gets to 47 tiles on the Ponte Vecchio chip:

  • 16 Xe HPC (internal/external)
  • 8 Rambo (internal)
  • 2 Xe Base (internal)
  • 11 EMIB (internal)
  • 2 Xe Link (external)
  • 8 HBM (external)

According to reports, the Ponte Vecchio chip is composed of two separate GPU dies, and each die consists of six Xe-HPC Compute units. Each GPU block is attached to four HBM2 stacks which could be either 4-hi or 8-hi. There are eight HBM2 stacks in total and also 8 passive die stiffeners on each GPU. The GPU also makes use of 11 EMIB dies that are featured underneath the HBM2 and I/O link chips.

Here are the actual EU counts of Intel’s various MCM-based Xe HP GPUs along with estimated core counts and TFLOPs:

  • Intel Xe HP (12.5) 1-Tile GPU: 512 EU [Est: 4096 Cores, 12.2 TFLOPs assuming 1.5GHz, 150W]
  • Intel Xe HP (12.5) 2-Tile GPU: 1024 EUs [Est: 8192 Cores, 20.48 assuming 1.25 GHz, TFLOPs, 300W]
  • Intel Xe HP (12.5) 4-Tile GPU: 2048 EUs [Est: 16,384 Cores, 36 TFLOPs assuming 1.1 GHz, 400W/500W]

Intel Xe class GPUs would feature variable vector width as mentioned below:

  • SIMT (GPU Style)
  • SIMD (CPU Style)
  • SIMT + SIMD (Max Performance)


- Advertisement -Follow-Google-News


OnePlus 12R Anticipation: A Sneak Peek into the Upcoming Flagship Marvel

In the paced world of smartphones, OnePlus continues to...

U Mumba unveils PKL Season 10 Jersey with Vicky Kaushal

In a resplendent event on Wednesday, U Mumba brilliantly...

Nintendo Switch 2’s New Update On Reconstruction

Recently there have been rumours on the system revealed...

iQOO 12 Price For India Revealed: Seems to be under Rs 60,000

By releasing a picture of the phone's retail box,...

Don't miss

The Sleep Company [CPS] IN


Nivedita Bangari
Nivedita Bangari
I am a software engineer by profession and technology is my love, learning and playing with new technologies is my passion.

The Best Spotify Alternatives in 2023 that You Must Know

Best Spotify Alternatives in 2023: The well-known music streaming service Spotify has completely changed how we listen to and appreciate music. Spotify satisfies the...

HBO Max in India: Here’s how you can watch the service using VPN (30th November)

HBO Max in India might launch soon but still, we cannot deny that we want to enjoy our favourite HBO shows as soon as...

Top 10 IT Companies in World: Here are the leading IT companies in the world (30th November)

Top 10 IT company in world: Over the last two years, there has been an increase in IT expenditure, which has resulted in the...


Please enter your comment!
Please enter your name here

This site uses Akismet to reduce spam. Learn how your comment data is processed.