NVIDIA Groq 3 LPX
The NVIDIA Groq 3 LPX is an inference accelerator designed for NVIDIA Vera Rubin, optimized for low-latency, large-context agent systems and large-scale content generation. Each rack supports 256 LPX units, delivering 128GB of SRAM, 40PB/s of memory bandwidth, and 640TB/s of rack-level scaling bandwidth.
In Stock
Inventory:99itemCore Specifications
GPU Architecture
Vera Rubin
LPX Quantity
256 per rack
SRAM
128GB
Memory bandwidth
40 PB/s
Expand Bandwidth
640 TB/s
Usage
Low-latency inference acceleration
CompleteSpecifications
| GPU Architecture | Vera Rubin |
| LPX Quantity | 256 per rack |
| SRAM | 128GB |
| Memory Bandwidth | 40 PB/s |
| Expand Bandwidth | 640 TB/s |
| Usage | Low-latency inference acceleration |
| Shape | Rack-level |
