AIME T400 Manuale utente
User Guide
AIME T400 - WORKSTATION
02/12
AIME T400 Workstation
Artifi ial Intelligen e
Ma hines
Contents
1. Devi e Des ription & S ope of Delivery 03
2. Important Safety Instru tions 04
3. Te hni al Spe ifi ation 05
4. Components & Cooling System 06
5. Conne tivity 07
6. Cleaning & Care 08
7. Liquid Cooling 09
8. AIME ML Container Manager 10
9. Complian e & Servi es 11
10. Warranty 12
Copyright © 2020 AIME UG (haftungsbes hränkt) All rights reserved. AIME is a trademark of
AIME U (limited liability). Other brands may be trademarks of their respective owners.
This product was individually made to customer specifications. The ocially stated prices are
subject to change without notice. If you have any questions about the price, please contact
your AIME customer manager.
AIME is not responsible for incorrect images or printing errors.
03/12
AIME T400 Workstation
Artifi ial Intelligen e
Ma hines
1. Devi e Des ription & S ope of Delivery
Devi e Des ription
The AIME T400 is a workstation that is designed for
deep learning development and has been optimized
for maximum performance with regard to low
volume development and the lowest possible energy
consumption without compromising on durability and
reliability.
All installed components were selected for their
energy eciency, durability, compatibility and high
performance. They are perfectly balanced with each
other so that there are no performance bottlenecks.
The elobareted cooling system of the AIME T400
covers the CPU and all PUs with a stream of liquid
cooling. High thermal conduction blocks and the far
higher cooling capabilities of liquid compared to air
keeps the operation temperature of all components
well below 60°C.
The secured low operation temperature reduces
stressful overheating for PU and CPU silicons and
allows that all components keep operating at their
highest performance levels even under full load in
24/7 scenarios.
The large radiators in the front and on top of the
system are cooled by high durability noise reduced
fans with a guaranteed operation of more than 10
years MTBF. This setup keeps the system cooler,
more performant, durable and far less noisy than an
collection of many small fans for each component.
The AIME T400 enables you to train deep learning
models at up to four times the speed of a single PU,
without throttling the computing power due to heat.
This means AI supercomputing performance with
over 400 trillion tensor FLOPS, which is equivalent to
the performance of hundreds of conventional
servers.
AIME workstations are delivered with a pre-installed
Linux operating system and are preconfigured with
the latest drivers and frameworks such as
Tensorflow, Keras, PyTorch and Mxnet.
The pre-installed AIME ML Container Manager oers
a comprehensive software stack, with which
developers can comfortably set up AI projects and
navigate between frameworks and projects. The
libraries and PU drivers required for the respective
deep learning framework are loaded in
preconfigured Docker containers and can be
configured and started with simple commands. The
most common frameworks - i.a. TensorFlow, Keras,
PyTorch and Mxnet - are pre-installed and ready to
use.
The AIME ML Container Manager makes life easier
for developers, so they don't have to worry about
problems installing the framework versions.
AIME customers log in immediately after unpacking
and start development right away with their
preferred deep learning framework.
S ope of Delivery
1x AIME Workstation T400
1x power cable for power supply
(IEC connector on protective contact)
1x WLAN antenna
1x quick start guide with individual access data
1x technical specifications
04/12
AIME T400 Workstation
Artifi ial Intelligen e
Ma hines
2. Important Safety Instru tions
CAUTION - RISK OF ELECTRIC SHOCK!
WATCH OUT! When using the workstation, the basi
safety rules must always be followed to avoid
injuries, overvoltage, fire and ele tri sho k:
1. Follow the manufacturer's instructions and safety
instructions during installation. Failure to do so can
damage the device and cause serious injury or death.
2. The device operateson high voltage. Never open
the device and do not try to repair it. Above all, never
open the power supply - not even when it is switched
o - there is a risk to life and can lead to injuries,
overvoltage, fire and electric shock!
3. This product is only intended for indoor use.
4. Do not use the device near water or at high
temperatures or in environments with high humidity.
5. Do not use the device during thunderstorms. In the
event of a thunderstorm, pull the power plug from
the power supply to protect it from damage caused
by lightning and overvoltage and to protect people.
6. Do not place the device near heat sources such as
fan heaters, heaters, stoves or other devices that
generate heat.
7. Do not insert any objects into the open areas of the
ventilation or fan grilles of the device.
8. The device needs adequate ventilation. Make sure
there is a distance of at least 10 cm on both sides,
front and back. Do not place any objects on the top,
the fans must remain free.
9. Do not modify any cables or connectors included
with this device.
10. Only use power cords that are compatible with
the device's power supply model. When using other
cables, the contacts may not be compatible, which
can seriously damage your device.
11. Modifications inside the device are only to be
carried out by a specialist. Before starting such work,
switch o the device and pull out the main power
plug. ESD protective measures must be observed
during all work.
12. Failure to observe the manufacturer's
instructions or one of these safety instructions, as
well as opening the device or removing, severing or
damaging one of the guarantee seals leads to the
immediate loss of all warranty and guarantee claims.
13. This device has been developed in such a way
that all specified EN classifications are only met if it
is used within a completely closed metal housing.
14. Make sure that your data is backed up regularly
to prevent data loss due to faulty data carriers,
memory modules and other hardware and software
errors or malware. In the event of a defect, the data
can then quickly be restored.
*: Notes on ESD prote tion
Electronic components are very sensitive to electrostatic
discharge. Therefore, never touch chips and contacts on circuit
boards directly, but only on the outer edges of the circuit board
without touching contacts or components. Ensure electrical
grounding on the housing or on metallic objects, e.g. radiators.
05/12
AIME T400 Workstation
Artifi ial Intelligen e
Ma hines
3. Te hni al Spe ifi ation
PERFORMANCE
CPU ( onfigurable)
Threadripper 1920X (12 cores, 4 Hz)
Threadripper 1950X (16 cores, 4 Hz)
Threadripper 2920X (12 cores, 4.3 Hz)
Threadripper 2950X (16 cores, 4.3 Hz)
Threadripper 2970X (24 cores, 4.2 Hz)
Threadripper 2990X (32 cores, 4.2 Hz)
RAM
64 or 128 B, ECC optional
GPU
2 to 4 NVIDIA RTX 2080 TI
Cooling
CPU and PU liquid cooled
Low-noise radiator fan > 100000h MTBF
Main Memory
up to 2 X 2TB NVMe SSD, configurable:
QLC: 1500 MB/s read, 1000 MB/s write
TLC: 3500 MB/s read, 1750 MB/s, write
MLC: 3500 MB/s read, 2700 MB/s write
Chipset
AMD X399
Power Supply
1600 or 1200 watts of power (depending on the
number of PUs installed),
80 PLUS Titanium or Platinum certified (96% Eciency):
92% at 20% utilization
94% at 20% utilization
90% at 80% utilization
Noise Level
Idle < 30dBA, Full Load < 40dBA
DESIGN
Type
Mid Tower Workstation
Dimensions (WxHxD)
233 x 465 x 538 mm
Weight
approx. 23 kg
SOFTWARE
Preinstalled
Linux OS (Ubuntu)
AIME ML Container Manager, Docker,
AI-Frameworks (TensorFlow, Keras, PyTorch, Mxnet)
CONNECTIVITY
Front
1 x USB Type-C™ port with USB 3.1 en 1
2 x USB 3.1 en 1 ports
Ba k
PU outputs depending on configuration
(HDMI 2.0b, USB Type-C)
1 x USB Type-C™ USB 3.1 en 2
1 x USB 3.1 en 2 Type-A
8 x USB 3.1 en 1
1 x PS/2 keyboard/mouse port
1 x RJ-45 port
2 x MMCX antenna connectors (2T2R)
1 x optical S/PDIF Out
5 x audio jacks (Center/Subwoofer Speaker Out, Rear
Speaker Out, Line In, Line Out, Mic In)
Network
2 x 1000 MBit LAN
Intel® Wi-Fi 802.11, 2.4/5 Hz Dual-Band
BLUETOOTH 4.2, 4.1, BLE, 4.0, 3.0, 2.1+EDR
Ele tri ity
1x 230V IEC connector
ENVIRONMENTAL CONDITIONS
Ambient Temperature
In operation: 9°C - 37°C
Out of operation: 1°C - 60°C (liquid cooling)
Ambient Humidity
rel. Air humidity without condensation
In operation: 20 - 80%
Out of operation: 20 - 85%
06/12
AIME T400 Workstation
Artifi ial Intelligen e
Ma hines
4. Components & Cooling System
Cooling blo ks with
fans (blowing out)
Fan
(su king in)
Flux Capa itor
(surge tank)
Pump
(liquid ooling)
Power Adapter SSD’s
RAM
CPU
GPU’s
Liquid Cooling
(System)
Air Cooling
(System)
Air Cooling
(Power Adapter)
07/12
AIME T400 Workstation
Artifi ial Intelligen e
Ma hines
5. Conne tivity
Ba k
1x PS/2 keyboard/mouse port
2x USB 3.1 en 1 port mit DAC--UP
1x USB 3.1 en 1 port (white) with Q-Flash Plus for BIOS Upgrade
5x USB 3.1 en 1 ports (Standard)
1x USB Type-C™ port, with USB 3.1 en 2 Support
1x USB 3.1 en 2 Type-A port (red)
2x RJ-45 ports (Netzwerk, 1000 MBit LAN)
2x MMCX antenna connectors (2T2R)
1x optical S/PDIF Out connector
5x audio jacks (Rear, Center/Subwoofer Speaker Out,
Mic In, Speaker/Line Out, Line In)
PU outputs, depending on the individual configuration
(Display Port, HDMI, USB Type-C)
Network
2x 1000 MBit LAN
Intel® Wi-Fi 802.11, 2.4/5 Hz Dual-Band
BLUETOOTH 4.2, 4.1, BLE, 4.0, 3.0, 2.1+EDR
Front
1x HD-Audio Out
1x HD-Audio In
1x USB Type-C™ port with USB 3.1 en 1
On / o switch
Reset switch
2x USB 3.1 en 1 ports
1
1
2
2
3
3
4
4
5
a
b
c
d
e
f
5
6
6
7
7
8
9
8
10
10
9Speed LED
Activity LED
a b c d e f
08/12
AIME T400 Workstation
Artifi ial Intelligen e
Ma hines
6. Cleaning & Care
Filter Cleaning
The filters on the front and top of the AIME T400 Workstation are removable and
can be cleaned with a vacuum cleaner.
09/12
AIME T400 Workstation
Artifi ial Intelligen e
Ma hines
7. Liquid Cooling
Che k Liquid Level
The liquid cooling of the AIME T400 is designed for problem-free operation of
7-10 years. If the PU temperature rises to such an extent that individual PUs
cause self-throttling or if a gurgling noise can be heard continuously, please
contact AIME customer support.
If you have an AIME T400 as "White lass" edition, you can visually check the liquid
level yourself. If the fill level is below 50%, the tank should be filled with coolant. To
do this, please contact AIME customer support for recommended and compatible
coolant and instructions for filling the expansion tank.
Cloudiness of the coolant over time is normal and does not impair its function.
After transporting the AIME T400 Workstation, there may initially be some noise
("gurgling") when the device is put into operation - if the device was on its side
during transport or "upside down". If these noises persist for a long time, it can be
assumed that there is too little liquid in the cooling circuit. Then please contact
AIME customer support.
AIME Customer-Support: [email protected]
10/12
AIME T400 Workstation
Artifi ial Intelligen e
Ma hines
8. AIME ML Container Manager
AIME ML Container Manager
AIME has developed a container management
system that helps to easily install, run, execute and
manage Docker containers for the most common
deep learning frameworks.
Features
• Setup and run a specific version of Tensorflow,
Pytorch or Mxnet with one simple command
• Run dierent versions of machine learning
frameworks and required libraries in parallel
• Manage required libraries (CUDA, CUDNN, CUBLAS,
etc.) in containers, without compromising the host
installation
• Clear separation of user code and framework
installation, test your code with a dierent
framework version in minutes
• Multi Session: Open and run many shell session on
a single container simultaneously
• Multi User: Separate container space for each user
• Multi PU: Allocate PUs per user, container or
session
• Runs with the same performance as a bare metal
installation
• Repository of all major deep learning framework
versions as containers
Available Frameworks
• Tensorflow
• Mxnet
• Pytorch
Available Versions
Tensorflow: 2.0.0, 1.15.0, 1.15, 1.14.0, 1.13.2
Mxnet: 1.5.0, 1.4.1, 1.3.1
Pytorch: 1.3.0, 1.2.0, 1.1.0
For more information, visit the AIME itHub website:
https://github. om/aime-team/aime-ml- ontainers
Indice
Altri manuali AIME Scrivania
