Product overview
Purpose-built accelerator for enterprise AI workloads
Designed for dense inference, model adaptation, and private AI infrastructure where predictable availability, localized supply, and software compatibility are critical.
Architecture
GPGPU/XCORE
Memory & Performance
High-bandwidth memory profile for AI inference and training.
VRAM capacity64
Memory typeHBM2e
Memory bandwidth440
Interconnect typeMetaXLink
Interconnect speed384
Architecture
Compute architecture and software execution model.
ArchitectureGPGPU/XCORE
Compute units-
Power & Thermal
Data center integration requirements.
Thermal design power350
CoolingПассивное
Form factorPCIe
Pixel Rate-
Texture Rate-
Benchmarks
Peak theoretical compute for common AI precisions.
FP64
-
FP32
30
FP16
240
TF32
120
BF16 Tensor
240
FP8 Tensor
-
INT8 Tensor
480
Compatibility
Interfaces, frameworks, and deployment environment.
PCIe interfacePCIe 4.0x16
Video encoding-
Video decoding-
Physical Dimensions
Card dimensions for server platforms.
Slots2
Length- mm
Height- mm
Width- mm
Pricing
On request
Volume pricing available for cluster deployments and pilot batches.
Documentation