Item
OSS-AMPERE8 AI Training Platform
Rack Scale
[One Stop Systems]
- 10 PetaTOPS INT4*; 2.5 PetaFLOPs Tensor Float 32*, 156 TeraFLOPS FP64
Product
OSS-AMPERE8
Model
OSS-AMPERE8 AI Training Platform
Vendor
One Stop Systems
Category
Automation & Industrial
Product Group
One Stop Systems
Type
Rack Scale
Consulting
Requirements analysis and custom system design
Implementation
Integration guidance and lifecycle technical support
Verification
Testing against specifications and industry standards
Features
- 4U Chassis
- Dual Future Generation Intel® Xeon® Scalable processors
- Up to 6TB DDR4 LRDIMM System Memory
- Six 2.5" NVMe SSD drive bays
- Eight Ampere GPU SXM4 with 600GB/s NVLink 3.0
- Ten x16 PCIe 4.0 slots
- Four 2200W Titanium Power Supplies
- GPU Management and Monitoring pre-installed
- Software bundle pre-installed
Specifications
Details
Dimensions:
4U Rack Units 7" H x 17.2" W x 35.2" D (39.3" with rails)
CPUs:
Dual Future Generation Intel® Xeon® Scalable processors
Drive Bays:
Hot-
System BIOs:
BIOS Type: 128Mb SPI Flash EEPROM with AMI® BIOS
Cooling Fans:
8x 92mm + 4x 80mm heavy duty counter-
Power Supply:
4x 2200W Redundant Power Supplies with PMBus
Total Output Power: 1200W/
Dimension (W x H x L): 73.5 x 40 x 265 mm
Certification: UL/
Environment:
Operating Temperature: 10°C to 30°C (50°F to 86°F)
Non-
Operating Relative Humidity: 8% to 90% (non-
Non-
Compliance:
RoHS 6 of 6, WEEE
System Memory:
Memory Capacity
- 32x 288-pin DDR4 DIMM slots
Memory Type
- 3200 MHz ECC DDR4 RDIMM/LRDIMM
GPUs:
8x Tesla A100 SXM4
- 10 PetaTOPS INT4*; 2.5 PetaFLOPs Tensor Float 32*, 156 TeraFLOPS FP64
- 320GB HBM2 Memory
- 4.8 TB/s Total Aggregate GPU Bandwidth
Software Bundle:
Choice of Operating System
- CentOS
- RHEL
- SLES
Choice of Machine Learning Framework
- Caffe2
- Pytorch
- Mxnet
- Microsoft Cognitive Toolkit
- Tensorflow
- Theano
- MLPython
- ML Dependencies (400MB Python)
- cuDNN
- DIGITS
- Caffe on Spark
- CUDA & NVIDIA driver
- CUB (CUDA building blocks)
- NCCL
GPU Management from Bright Computing
- Health Management
- Workload Integration
Expansion Slots:
PCI-Express
- Ten PCIe 4.0 x16 LP slots
- Six PCIe 4.0 x4 U.3 2.5" drive bays
On-board devices:
Chipset: Dual Future Intel® Xeon® Scalable Processors
SATA: SATA3 (6Gbps) with RAID 0, 1, 5, 10
IPMI
- Support for Intelligent Platform Management Interface v.2.0
- IPMI 2.0 with virtual media over LAN and KVM-over-LAN support
Network Controllers
- Intel® X540 Dual Port 10GBase-T
- Virtual Machine Device Queues reduce I/O overhead
- Supports 10GBASE-T, 100
Graphics: ASPEED AST2400 BMC
Front Panel:
Buttons
- Power On/Off button
- System Reset button
LEDs
- Power LED
- Hard drive activity LED
- Two Network activity LEDs
- System Overheat LED / Fan fail LED /
- UID LED
Resources