Figure 5. PCI Express Cable (2 Meter Long) .......................................................................7
List of Tables
Table 1. System Configuration .........................................................................................3
Table 2. SKUs for Rack Mount Kit and x8 Host Interface Card.............................................3
Table 3. Environmental Specifications and Conditions.........................................................8
Table 4. Languages Supported......................................................................................... 9
January 14, 2008 | SP-03718-001_v01 iv
January 14, 2008 | SP-03718-001_v01 1
The NVIDIA® Tesla™ D870 Deskside Computing System features two Tesla
graphics processing units (GPUs) and can be used either alongside a workstation or
alternatively, two deskside systems can be rack-mounted for four Tesla GPUs in a
3U configuration (it is possible to rack-mount just one deskside unit).
The Tesla D870 connects to a host workstation via a low-power PCI Express ×16
or ×8 adapter card.
Key Specifications
NVIDIA Tesla D870
Overview
Computing Processors
Two Tesla C870 GPUs
¾ 128 thread processors per GPU, each with a single-precision floating point
3 GB of total memory (1.5 GB dedicated memory per GPU)
Mechanical Overview
Physical Dimensions
¾ Tower desktop: 9.54 inches high × 5.94 inches wide × 20.6 inches deep
¾ Rack Mount: 5.25 inches (3U) high × 8.5 inches wide × 18 inches deep
Host Adapter Card
¾ PCI Express ×16 or ×8: small form factor, passive heat sink (peak power
dissipation 10 W)
External Connectors
¾ 2 PCI Express cable connectors: 2 meter length
Operating Environment
Quiet operation: 40 dB, suitable for office environment
Power: 520 W (maximum), 100-240 VAC Autosensing
Temperature: 5 degrees Celsius to 35 degrees Celsius
Relative Humidity: 10 % to 80 % non-condensing
System Architecture
pply
The Tesla D870 GPU computing system is based on the 8-Series GPU from
NVIDIA. It can be connected to a single host system via a PCI Express connection
as shown in Figure 1.
CPU Host
Workstation
CPU
FSB
NVIDIA Tesla D870 GPU Deskside System Specification
Tesla D870 GPU
Deskside System
Power
Su
Core
Logic
Tesla
GPU
Tesla
GPU
PCI Express
Cables
PCI-E Gen1 x16
NVIDIA PCI-E
Switch
Figure 1. Tesla D870 GPU Computing System Architecture
January 14, 2008 | SP-03718-001_v01 2
Configuration
The Tesla D870 has one system configuration as listed in Table 1. Geographical
SKUs have localized power cords but the system configuration is the same.
The Tesla D870 can be used as a deskside unit or alternatively, two deskside systems
can be rack-mounted for four Tesla GPUs in a 3U configuration. The rack-mounted
deskside conforms to the EIA 310E specification for 19” 4-post racks with 900 mm
to 1000 mm depth.
Note: It is also possible to rack-mount a single deskside in a 3U configuration.
Mechanical
Specification
The chassis dimensions are:
Tower desktop: 9.54 inches high × 5.94 inches wide × 20.6 inches deep
Rack Mount: 5.25 inches (3U) high × 8.5 inches wide × 18 inches deep
NVIDIA Tesla D870 GPU Deskside System Specification
Figure 2. Tesla D870 Deskside
Figure 3. Tesla D870 Rack-Mounted
January 14, 2008 | SP-03718-001_v01 5
Host Adapter Card
The NVIDIA PCI Express interface card (Figure 4) comes with a standard (long)
bracket as well as a shorter low profile bracket that conforms to the PCI Express
low profile form factor. The specifications for the interface card are:
Form factor: 6.6 inches × 4.376 inches
Power draw: 10 W (maximum)
Cooling: Passive heat sink (no fan)
Bus: PCI Express ×16
An optional PCI Express ×8 host interface card can be purchased separately.
NVIDIA Tesla D870 GPU Deskside System Specification
Figure 4. PCI Express x16 Host Adapter Card
January 14, 2008 | SP-03718-001_v01 6
PCI Express Cable
The Tesla D870 includes two PCI Express cables. These cables are 2 meters in
length and connect the Tesla D870 to the host system using the host adapter card.
Figure 5. PCI Express Cable (2 Meter Long)
NVIDIA Tesla D870 GPU Deskside System Specification
January 14, 2008 | SP-03718-001_v01 7
January 14, 2008 | SP-03718-001_v01 8
Environmental Specifications
Summary
The Tesla D870 system has passed thermal and shock vibration qualification testing
and is recommended to operate under the conditions listed in Table 3.
Table 3. Environmental Specifications and Conditions
Environmental
Specifications
Specifications Conditions
Operating temperature 5 °C to 35 °C
Operating humidity 10 to 80 % RH, non-condensing
Operating Altitude 5000 feet mean sea level (MSL)
Operating Shock Half sine 40g, 2ms duration, 3 axis
Operating Vibration Sinusoidal 0.25g, 10 to 500 Hz, 3 axis. Random 1.0
Acoustics 40 dBa at 1 meter in front of system
Non-operating temperature 0 °C to 65 °C
Grms, 10 to 500 Hz
January 14, 2008 | SP-03718-001_v01 9
Languages
Table 4. Languages Supported
Support
Information
English (US) X X
English (UK) X
Arabic X
Chinese, Simplified X
Chinese, Traditional X
Danish X
Dutch X
Finnish X
French X
French (Canada) X
German X
Italian X
Japanese X
Korean X
Norwegian X
Portuguese (Brazil) X
Russian X
Spanish X
Spanish (Latin America) X
Swedish X
Thai X
Windows
XP
Linux
Note: NVIDIA’s CUDA™ software is only supported in English (U.S.)
Certificates and Agencies
Certifications
There are no certifications planned at this time.
Agencies
Australia Communication Authorities (C-Tick)
Bureau of Standards, Metrology, and Inspection (BSMI)
Conformité Européenne (CE)
Federal Communications Commission (FCC)
Interference-Causing Equipment Standard (ICES)
Ministry of Information and Communication (MIC)
NVIDIA Tesla D870 GPU Deskside System Specification
Underwriters Laboratories (UL)
Voluntary Control Council for Interference (VCCI)
January 14, 2008 | SP-03718-001_v01 10
Notice
ALL NVIDIA DESIGN SPECIFICATIONS, REFERENCE BOARDS, FILES, DRAWINGS, DIAGNOSTICS, LISTS, AND
OTHER DOCUMENTS (TOGETHER AND SEPARATELY, “MATERIALS”) ARE BEING PROVIDED “AS IS.” NVIDIA
MAKES NO WARRANTIES, EXPRESSED, IMPLIED, STATUTORY, OR OTHERWISE WITH RESPECT TO THE
MATERIALS, AND EXPRESSLY DISCLAIMS ALL IMPLIED WARRANTIES OF NONINFRINGEMENT,
MERCHANTABILITY, AND FITNESS FOR A PARTICULAR PURPOSE.
Information furnished is believed to be accurate and reliable. However, NVIDIA Corporation assumes no
responsibility for the consequences of use of such informa ti on or f or any infringement of patents or other
rights of third parties that may result from its use. No license is granted by implication or otherwise under
any patent or patent rights of NVIDIA Corporation. Specifications mentioned in this publication are subject to
change without notice. This public ation supersedes and replaces all information previously supplied. NVIDIA
Corporation products are not authorized for use as critical components in life support devices or systems
without express written approval of NVIDIA Corp oration.
Trademarks
NVIDIA, the NVIDIA logo, CUDA and Teslaare trademarks or registered trademarks of NVIDIA Corporation in
the United States and other countries. Other company and product names may be trademarks of the
respective companies with which they are associated .