|
gptkbp:instance_of
|
gptkb:microprocessor
|
|
gptkbp:ai
|
gptkb:Tensor_Flow
gptkb:Caffe
gptkb:MXNet
gptkb:Py_Torch
|
|
gptkbp:airesearch_applications
|
gptkb:Yes
|
|
gptkbp:architecture
|
gptkb:NVIDIA_Pascal
|
|
gptkbp:cooling_system
|
air-cooled
|
|
gptkbp:data_usage
|
high
|
|
gptkbp:dimensions
|
3 U rack size
|
|
gptkbp:fan_engagement
|
N+1 redundancy
|
|
gptkbp:first_generation
|
gptkb:Yes
|
|
gptkbp:form_factor
|
rack-mounted
|
|
gptkbp:gpu
|
gptkb:NVIDIA_Tesla_P100
8
|
|
gptkbp:has_programs
|
gptkb:NVIDIA_GPU_Cloud
|
|
https://www.w3.org/2000/01/rdf-schema#label
|
NVIDIA DGX-1
|
|
gptkbp:is_a_framework_for
|
gptkb:Yes
|
|
gptkbp:is_scalable
|
gptkb:Yes
|
|
gptkbp:management
|
NVIDIA System Management Interface (nvidia-smi)
|
|
gptkbp:manufacturer
|
gptkb:NVIDIA
|
|
gptkbp:network
|
10 Gb E
|
|
gptkbp:number_of_cores
|
gptkb:Yes
|
|
gptkbp:operating_system
|
gptkb:Ubuntu_Linux
|
|
gptkbp:pciexpress_version
|
gptkb:4
|
|
gptkbp:performance
|
1.3 petaflops
|
|
gptkbp:power_consumption
|
3 k W
|
|
gptkbp:power_supply
|
dual redundant power supplies
|
|
gptkbp:price
|
approximately $149,000
|
|
gptkbp:ram
|
512 GB
|
|
gptkbp:release_date
|
gptkb:2017
|
|
gptkbp:storage
|
8 TB SSD
|
|
gptkbp:successor
|
gptkb:NVIDIA_DGX_A100
|
|
gptkbp:support_for_nvidiaclara
|
gptkb:Yes
|
|
gptkbp:support_for_nvidiadali
|
gptkb:Yes
|
|
gptkbp:support_for_nvidiadeep_stream
|
gptkb:Yes
|
|
gptkbp:support_for_nvidiajarvis
|
gptkb:Yes
|
|
gptkbp:support_for_nvidiamaxine
|
gptkb:Yes
|
|
gptkbp:support_for_nvidiamerlin
|
gptkb:Yes
|
|
gptkbp:support_for_nvidiarapids
|
gptkb:Yes
|
|
gptkbp:support_for_nvidiatensor_board
|
gptkb:Yes
|
|
gptkbp:support_for_nvlink
|
gptkb:Yes
|
|
gptkbp:support_for_tensor_rt
|
gptkb:Yes
|
|
gptkbp:target_market
|
data scientists
enterprise
research institutions
|
|
gptkbp:training_time_reduction
|
up to 10x
|
|
gptkbp:use_case
|
deep learning
|
|
gptkbp:virtualization_support
|
gptkb:Yes
|
|
gptkbp:weight
|
90 kg
|
|
gptkbp:bfsParent
|
gptkb:NVIDIA_P100
|
|
gptkbp:bfsLayer
|
5
|