H3 Falcon 4010 User manual

Falcon 4010
User Manual

1
Navigation
Device Allocation P.14
Firmware update P.24
GUI Introduction P.7
Hardware Installation (Quick Installation Guide)
LCD Operation P.34
License Key Activation P.31
Operation Safety P.45
PCIe Port Configuration P.16
Rest to Default P.44
Service Parts P.33
Specifications P.5
System modes P.6
Trouble Shooting P.46

2
Table of Contents
1. Introduction.................................................................................................................... 4
1.1 Key Features ..................................................................................................................... 4
2. Package Contents............................................................................................................ 4
3. Technical Specifications................................................................................................... 5
4. Requirements.................................................................................................................. 5
4.1 Host Server ....................................................................................................................... 5
4.2 Host OS/BIOs .................................................................................................................... 5
4.3 Web Browsers................................................................................................................... 6
5. Hardware Installation and Initial Settings........................................................................ 6
6. System Modes................................................................................................................. 6
6.1 Standard Mode................................................................................................................. 6
6.2 Advanced Mode................................................................................................................ 6
7. Graphical User Interface.................................................................................................. 7
7.1 Log-in................................................................................................................................ 7
7.2 Functions.......................................................................................................................... 7
7.2.1 Overview................................................................................................................................................ 8
7.2.2 Resource Management........................................................................................................................ 12
7.2.3 Port Configuration................................................................................................................................ 16
7.2.4 Monitor ................................................................................................................................................ 18
7.2.5 System Health ...................................................................................................................................... 20
7.2.6 Chassis.................................................................................................................................................. 23
7.2.7 Maintenance........................................................................................................................................ 24
7.2.8 Event Logs ............................................................................................................................................ 26
7.2.9 Setting .................................................................................................................................................. 27
8. Parts Replacement........................................................................................................... 33
8.1 Fans ................................................................................................................................ 33
8.2 Power Supply Units......................................................................................................... 33
9. LCD.................................................................................................................................. 34

3
9.1 Operation ....................................................................................................................... 34
9.2 Menus............................................................................................................................. 35
9.2.1 Main menu........................................................................................................................................... 35
9.2.2 Power control....................................................................................................................................... 36
9.2.3 Power reset.......................................................................................................................................... 36
9.2.4 System.................................................................................................................................................. 36
9.2.5 Slot ....................................................................................................................................................... 37
9.2.6 Devices................................................................................................................................................. 38
9.2.7 Hosts .................................................................................................................................................... 39
9.2.8 Health................................................................................................................................................... 40
9.2.9 Temperature......................................................................................................................................... 41
9.2.10 Network ............................................................................................................................................. 41
9.2.11 Password (Feature coming Soon) ................................................................................................... 43
9.2.12 Reset to default.................................................................................................................................. 44
10. Operation Safety............................................................................................................ 45
11. Trouble Shooting ........................................................................................................... 46
Symptoms or Errors.............................................................................................................. 46
Solution ................................................................................................................................ 46

4
1. Introduction
Falcon composable GPU solution consists of Falcon Composable Chassis and the management GUI. The system
is applicable to most multi-GPU applications and the software-defined approach greatly simplifies the device
managements.
1.1 Key Features
GPU composability
Device surprise-add and remove
GPU peer-to-peer
PCIe port configuration
Real-time GPU cluster topology
System performance monitoring
Role-based authentication and access control
2. Package Contents
Falcon 4010 GPU Expansion Chassis *1
Main chassis (1)
GPU drawer (2)
PSU (2)
Fan (6)
Power cord (PSU) *2
Mini-SAS HD external cable *8
Host bus adapter set *2
HBA (2)
Full-height bracket (2)
Half-height bracket (2)
PCIe power cable (8 to 8 pin) *8

5
3. Technical Specifications
Chassis
BMC/mCPU
Aspeed AST 2500
PCIe Switch
PEX 88096; PCIe 4.0
PCIe Slots*
8x PCIe4.0 x16 dual-width, FHFL 2x PCIe 4.0x16 low-profile
Host Interface
SFF-8644 connectors
Power
1600W; 220V AC; 80+ Platinum; hot swap
Fan
120x120x38mm; 6700 RPM; hot swap
Operating Temp.
0°C ~ 35°C (32°F ~ 95°F)
Dimension
5U; 219(H) x 435(W) x 450 (D) mm
Net Weight
34.5 kg
Mini-SAS HD external Cable
Connector
SFF-8644 to SFF-8644
PCIe
PCIe 4.0 x4 each
Length
2 meters
＊Each PCIe slot supports up to 300 watts (75W from slot + 225W from the 8pin PCIe power)
4. Requirements
4.1 Host Server
Minimum of one vacant PCIe x16 (PCIe 3.0 or higher) slot for HBA installation.
4.2 Host OS/BIOs
Standard Mode:
No limitations
Advanced Mode:
Ubuntu 16.04 LTS, 18.04 LTS, 20.04 LTS
Windows build 1903-20H2
Cent OS/ RHEL 7.3-8.0
Note:
Advanced mode is not limited to the OS listed above. The listed OS are recommended as they have
been tested to support PCIe device hot plug.

6
4.3 Web Browsers
Mozilla Firefox 3.5 (or higher)
Google Chrome ver. 12 (or higher)
5. Hardware Installation and Initial Settings
Please see Falcon 4010 Quick Installation Guide for system set up.
6. System Modes
There are two system modes for Falcon 4010. The Standard Mode is limited to single host connection and
does not support device dynamic allocation or host port bifurcation. The Advanced mode supports multiple
host connection and could allocated devices to hosts dynamically. You could activate the Advanced mode with
a premium license key.
Please contact [e-mail] for purchase.
6.1 Standard Mode
System monitor
Power control from GUI
Download system performance data from GUI
Firmware update
User management
Limited to single host (per GPU drawer)
*Does not support device dynamic allocation and host port bifurcation
6.2 Advanced Mode
System monitor
Power control from GUI
Download system performance data from GUI
Firmware update
User management
Multiple host
Device hot plug (if the OS supports this feature)
Port configuration
Device dynamic allocation
Mode switch

7
7. Graphical User Interface
7.1 Log-in
7.2 Functions
Every time you access GUI, you will be
asked to log in.
Please enter your username and
password.
The drop-down menu is at the top-left of
the page
Please find details of each function in the
relative section.

8
7.2.1 Overview
Resource List
The Resource List provides PCIe device usage and host port usage information.
Usage of specific device types (GPU, NVMe, and NIC) features can be accessed with premium license
activated.
“Used” indicates the number of devices that are currently assigned to hosts.
e.g. Used 8 of 10.
There are 10 devices installed in Falcon 4010, 8 of them are assigned to the host(s).
The overview page sorts out the basic
performance data of the Falcon 4010
system in charts and graphs.

9
GPU Utilization Rate(%)
In the GPU utilization chart, users can check the GPU utilization of a specific GPU in a specific period.
Y-axis represents the utilization rate and X-axis represents a specific GPU. The data is read from PCIe
devices directly, only the compatible devices with the out-band information will be shown here.
1. Graph title: GPU Utilization Rate(%)
2. Utilization rate: The average GPU utilization scaled from 0~100%
3. Bar graph: Utilization rate of a specific GPU displayed in bar graph
4. Device number: displayed as [Drawer#] : [Slot#]. E.g. 1:1 indicated GPU on drawer 1 slot 1.
5. Display period: The graph will display the utilization rate of the GPUs in the past hours.
(1, 12, 24, or 72 hours options available)
6. Download: Download the GPU utilization data (up to the past 72 hours)
PCIe Throughput(MB/s)
The PCIe Throughput graph shows the throughput of each device in MB/s. The data is read from PCIe
devices directly, only the compatible devices with the out-band information will be shown here.
1. Graph title: PCIe Throughput (MB/s)
2. Throughput rate: The numbers on throughput rate scale (MB/s) will increase as throughput

10
increases.
3. Time: The X-axis display system times (per hour)
4. Devices: List all the devices installed. Every device has a unique color indicator.
5. Throughput curve: The curve of PCIe throughput of each device, distinguished by the color.
6. List down menu: Displays throughput of each device at a specific time point.
Move the mouse over the curve to see this menu.
7. Select Traffics: Select traffic types to display on the throughput graph.
There are three types: Ingress, Egress, and Sum.
8. Display period: The graph will display the PCIe throughput rates in the past hours.
(1, 12, 24, 72 hours options available)
9. Download: Download the PCIe throughput data (up to past 72 hours)
PCIe Link Health
The PCIe Link Health chart shows the link health condition of every PCIe port in use.
1. Chart title: PCIe Link Health
2. Health indication: Green indicates healthy (Bad TLP=0, Bad DLP=0)
Red indicates errors existed (Bad TLP>0, BAD DLP>0)
Gray indicates no link
3. PCIe port number: PCIe ports are listed in order.
4. Port type: indicates that whether the port is a device or host port
5. Device name: The name of the devices installed on the specific PCIe ports.

11
Thermal(°C)
System Profile
The Thermal chart displays the average temperature of
each component (in °C) in the Falcon 4010 chassis is
displayed.
Green Good thermal performance
Amber Moderate thermal performance
Red Overheat. Please check out the system.
* Falcon 4010 will shut down automatically when the
system detects any device temperature >85°C for over
10 seconds.
The System Profile chart displays basic system
information of the chassis being operated.
Model: chassis model name
Serial number: the serial number of the chassis
Mac address: mac address of the chassis
Firmware: current BMC firmware version
System up time: time since the system is powered-on
Last Login: The last user that logged in
Online user: The number of users currently on

12
7.2.2 Resource Management
The topology view shows the graph of hosts, devices, and PCIe switch. The list view lists all the devices and
hosts in a table.
Under advanced mode, users can provision or re-provisioned the PCIe devices to connected hosts using
topology view.
Topology view
1. Display mode: PCIe resources can be displayed in either the topology mode or the list
mode.
2. Allocate: This button is used when allocating resource to the hosts. See
Device Allocation section for details.
3. Drawer 1 PCIe ports: PCIe ports of drawer 1 are in green background.
4. Drawer 2 PCIe ports: PCIe ports of drawer 2 are in blue background.
5. Legends: Help users to clarify the components in the topology mode
6. Refresh: Click to refresh the topology display
7. System mode: Display the current system mode of the drawers.
8. Port label aid: Click the icon, the chassis diagram with port labels will pop-up for aid.

13
Port Information:
1. Link icon: This icon indicates that the device is assigned to a host already.
2. Color tag: Each host has a colored frame. This color tag indicates that the device is assigned to
the host with the same color.
E.g.
The device 1:1 is assigned to the host 1:H1 (color = red)
3. Port & Device: Port number and the device name is displayed in the white box.
List view
*Each roll contains the information of a PCIe slot in use, the PCIe slots that are empty will not be listed.
1. Slot#: This column shows all the PCIe slots with device installed.
2. Assigned host: This column shows the hosts that the devices are currently assigned to.
3. Device name: This column shows the device name
4. Type: This column shows the device type. (GPU, NVMe SSD, or NIC)
5. UUID/Serial number: This column shows the UUID and serial number of the devices
6. Temperature: This column shows the temperature of the devices
7. Link capability: This column shows the link capability of the devices
[Device link capability] / [Slot link capability].

14
8. Device details: Click on the drop-down arrow to see detailed information of the selected
device.
9. Port label aid: Click the icon, the chassis diagram with port labels will pop-up for aid.
*The data is read from PCIe devices directly, only available if the device provides the out-band information.
Device Allocation
This feature is only enabled in Advanced mode.
Go to Resource Management page
(Use Topology mode)
1. Select the target host
2. Check the box beside the vacant device
3. Click “Allocate” to assign the device to the
host
If multiple PCIe devices should be provisioned to one connected host, users can also select multiple
devices at one time then allocate to one connected host.
The confirmation message will pop-up to ask users for confirmation.
Click “Yes” to confirm. Click “OK” to finish the provisioning processes.
After you have assigned the device to a host, the link icon and the color tag should appear.
e.g.

15
Release Device from host
This feature is only enabled in Advanced mode.
Go to Resource Management page
(Use Topology mode)
1. click the link icon next to the target
device
You can only deallocate one device at a
time with this method.
The confirmation message will pop-up to ask users for confirmation.
Click “Yes” to confirm. Click “OK” to finish the provisioning processes.
After you have assigned the device to a host, the link icon and the color tag should disappear.
The check box should appear.
e.g.

16
7.2.3 Port Configuration
Falcon 4010 provides user defined PCIe port configurations. All PCIe ports are default to 16 lanes (PCIe 4.0).
The lanes can be configured into 2x8 lanes or 4x 4lanes depending on the custom requirements.
1. Undo and Apply: Undo or Apply configuration settings. See Configure Ports section for
details.
2. Drawer 1 PCIe Ports: PCIe ports of drawer 1 are in green background.
3. Drawer 2 PCIe Ports: PCIe ports of drawer 2 are in blue background.
4. Legends: Help users to clarify the components in the topology mode
5. System mode: Display the current system mode of the drawers.
6. Port label aid: Click the icon, the chassis diagram with port labels will pop-up for aid.
Configure Ports
This feature is only enabled in Advanced mode.
Go to Port Configuration page.
There are 4 configurable ports:
1:H1, 2:H1, 1:H2, and 2:H2
Note
1:H1 and 2:H1 are fixed host ports, users can only control the bifurcation setting.
1:H2 and 2:H2 can be configured into device or host port, bifurcation setting is only available for host
mode.
(continue next page)

17
1. Click the drop-down icon of the PCIe port to be configured and select the desired configuration.
2. Click “Apply” to apply the configuration, click “Undo” to discard the configuration
The confirmation message will pop-up to ask users for confirmation.
Click “Yes” to confirm. Click “OK” to finish the configuration processes.
After you finished the configuration, your new configuration will be displayed, and the text should turn
Black.
e.g.
Please reboot Falcon 4010 (or the drawer) for the new configuration to take effect
Red text indicates that
the configuration is not
applied yet.

18
7.2.4 Monitor
In the Monitor page, users can see the real-time traffic, link speed, and the error count of each PCIe port.
1. Sub-menu: Select the monitor information you would like to see.
2. Drawer 1 PCIe ports: PCIe ports of drawer 1 are in green background.
3. Drawer 2 PCIe ports: PCIe ports of drawer 2 are in blue background.
4. Legends: Help users to clarify the components in the topology mode
5. System mode: Display the current system mode of the drawers.
6. Port label aid: Click the icon, the chassis diagram with port labels will pop-up for aid.
Traffic
When select Traffic, the traffic information will show up on the right side of every white box (port)
1. Egress Traffic: PCIe switch to device traffics
2. Ingress Traffic: Device to PCIe switch traffics

19
Link Speed
When select Link Speed, the link speed information will show up on the right side of every white box
(port).
Display format
[PCIe generation] x[PCIe lanes].
e.g.
Nvidia A100 PCIe is a PCIe Gen4 x16lane device, under normal condition, the link speed should
be displayed as Gen4 x16
1. Current link speed: The current link speed of the device installed on the PCIe port.
2. Maximum link speed: The maximum link speed of the PCIe port
Note:
Max link speed should always be Gen4 x16, the current link speed is depending on the device
installed.
Error
When select Error, the PCIe error count will show up on the right side of every white box (port).
Display format
[Bad DLLP] – [Bad TLP] – [Port RX Error] – [Recovery Diag. Error]
e.g.
0-0-0-2 indicates that there are two Recovery Diag. Error count.
1. Error counts: displays the number of each error type.
2. Clear errors: Click the icon to reset all error counts. (back to 0-0-0-0)