Data Center GPU Max Subsystem

Key Metrics

Save this GPU, then open another GPU page to compare them side by side.

Identity
Entry Type	Data Center GPU
Official Name	Intel Data Center GPU Max Subsystem
Architecture	Generation 12.5
Launch Date	2023-01-10
Core Specs
GPU Die	Ponte Vecchio
Process Node	Intel 10 nm
Transistor Count	100 billion
Die Size	1280 mm²
Execution Units	16384
TMUs	1024
RT Units / RT Cores	128
Base Clock	900 MHz
Boost Clock	1,600 MHz
Texture Fillrate	1638.4 GT/s
FP32 Throughput	52.43 TFLOPS
FP16 Throughput	52.43 TFLOPS
FP64 Throughput	52.43 TFLOPS
Memory
Memory Size	128 GB
Memory Type	HBM2e
Memory Bus Width	8192-bit
Memory Clock	1,565 MHz
Memory Bandwidth	3210 GB/s
L2 Cache	408 MB
L1 Cache	64 KB
Board and Display
TDP / TBP	2400 W
Recommended PSU	2800 W
Interface	PCIe 5.0 x16
Slot Width	Dual-slot
Card Length	267 mm
Software
Shader Model	6.6
DirectX Support	DirectX 12.1
OpenGL Support	OpenGL 4.6
OpenCL Support	OpenCL 3.0

Sources