System Configuration

  • Home
  • System Configuration
HPC System — Xeon Gold 6544Y · NVIDIA L40S

System Overview

Enterprise-grade HPC node featuring Intel® Xeon® Gold 6544Y, NVIDIA L40S Ada Lovelace accelerator, and 256 GB high-speed RAM — engineered for AI/ML training, inference, scientific simulation, and ray-tracing workloads.

Peak AI Performance
1466 TOPS
INT8 Inference · Tensor Cores
System Memory
256 GB
High-Speed RAM
Total Storage
~18 TB
High-Capacity Data Storage
Total Endpoints
111
Machines + 1 HPC Node

CPU — Intel® Xeon® Gold 6544Y

Compute Node
x86_64 · 64-bit Enterprise Architecture · ASUS Server Platform
Processor
Intel® Xeon® Gold 6544Y
Architecture
x86_64 (64-bit Enterprise)
Total CPU Cores
64 High-Performance Cores
Physical Address Space
46-bit Physical
Virtual Address Space
57-bit Virtual
System Memory
256 GB High-Speed RAM

GPU — NVIDIA L40S

Ada Lovelace
Data Center Grade Accelerator · CUDA 13.0 · 4th Generation Tensor Cores
GPU Model
NVIDIA L40S
GPU Architecture
Ada Lovelace
VRAM
46 GB GDDR6 ECC
Usable VRAM
46068 MiB
Memory Bandwidth
~864 GB/s
Max Power
350 Watts TDP
CUDA Version
13.0
Tensor Core Generation
4th Generation
Total Storage
~18 TB

Compute Performance

Benchmarks
Metric Value Relative Scale
FP32 Compute (CUDA Cores)
General-purpose floating point · simulation · physics
~91.6 TFLOPS
FP16 Tensor (4th Gen Tensor Cores)
Mixed-precision training · deep learning models
~733 TFLOPS
INT8 AI Inference
Quantized inference · production serving · high throughput
~1466 TOPS
RT Core (Ray Tracing)
Real-time ray tracing · scientific visualization · rendering
~212 TFLOPS

System Capabilities

Summary
Intel® Xeon® Gold 6544Y
NVIDIA L40S Ada Lovelace
64 CPU Cores
256 GB RAM
46 GB GDDR6 ECC VRAM
CUDA 13.0
4th Gen Tensor Cores
~18 TB Storage
864 GB/s Bandwidth
Data Center Grade
AI/ML Optimized
Scientific HPC

Machine Configuration

111 Units Total
Lab endpoints available for compute, development, and AI workloads — including 1 dedicated HPC node in the Windows cluster
30
Mac Mini Units
80
Windows Units
1
HPC Node
111
Total Endpoints
Apple Mac Mini
×30
macOS · Apple Silicon
ProcessorApple M2
Memory16 GB Unified
ArchitectureARM64
Units30 Machines
Lenovo — Windows
×50
Windows · Intel 12th Gen
ProcessorIntel Core i5 (12th Gen)
Memory8 GB RAM
OSWindows
Units50 Machines
Dell — Windows
×30
Windows · Intel 12th Gen
ProcessorIntel Core i5 (12th Gen)
Memory8 GB RAM
OSWindows
Units30 Machines
HPC Node — Windows
×1
Windows Host · Data Center Grade · NVIDIA L40S
CPUXeon® Gold 6544Y
Cores64 Cores
RAM256 GB
GPUNVIDIA L40S
VRAM46 GB GDDR6 ECC
CUDA13.0
Storage~18 TB
Bandwidth~864 GB/s
FP32~91.6 TFLOPS
FP16~733 TFLOPS
INT8~1466 TOPS
RT Cores~212 TFLOPS
Tensor Cores4th Generation

Network Switching Infrastructure

Switching Spec
Managed switch fabric connecting all lab endpoints and the HPC node
Managed Network Switch
24 PORT · ×7 UNITS · 1 GBPS
Switch Units
7 Switches
Ports per Switch
24 Ports
Total Ports
168 Ports
Port Speed
1 Gbps
Total Aggregate Bandwidth
168 Gbps
Endpoints Supported
111 Machines + HPC