GPUDatacenterReleased —

Intel Data Center GPU Max 1350

96 GB HBM2e · 14,336 cores

A 450 W OAM module announced in November 2022 as the mid-tier Ponte Vecchio part, with 112 Xe-cores and 96 GB of HBM2e for HPC and AI. Intel removed it from the line-up before it shipped, so the Max 1100 and Max 1550 are the only members of the family that reached customers.

General

Manufacturer: Intel
Model: Data Center GPU Max 1350
Segment: Datacenter
GPU chip: Ponte Vecchio
Architecture: Xe-HPC
Release date: —
Price: Search on Amazon

Memory

Memory size: 96 GB
Memory type: HBM2e
Bus width: —
Bandwidth: —
Memory clock: —

Clocks & silicon

GPU clock: —
Boost clock: —
Shader units: 14,336 ALUs (112 Xe-cores)
Tensor / AI cores: 896
RT cores: 112
Transistors: —
Technology (nm): 7 nm
Intel 7 + TSMC N5/N7 (multi-tile, Foveros/EMIB)
TDP / TBP: 450 W

Interface & APIs

Bus interface: OAM
SLI / multi-GPU: —
DirectX: N/A (compute-only)
OpenGL: N/A (compute-only)
Vulkan: N/A (compute-only)
Other APIs: oneAPI Level Zero, SYCL, OpenCL, OpenMP

AI support

Precisions: FP8FP16BF16INT8INT4TF32FP64
AI accelerator: 896 XMX Engines
Peak AI TOPS: —
Notes: Xe-HPC XMX engines cover TF32, FP16, BF16 and INT8; FP8 is not supported on this generation. FP64 runs on the vector engines at full rate, which is why the Max family targets HPC as well as AI.

All Intel GPUs →

Datacenter gpus from Intel announced around the same time as the Intel Data Center GPU Max 1350. Add any of them to your compare list to see the specs side by side.

gpu