This CloudFormation stack is based on Amazon Linux 2 along with NICE-DCV software, enabling a full interactive Linux desktop over HTML5.
- Dynamic resolution based on browser size
- Remote Audio
- Remote Text Copy/Paste
- EBS multi-volume stripe
- Sudo for custom software installation.
- Full Matlab suite (if selected)
- Support for up to 488GIB of memory, 96 vCPUs and 25 Gigabit network.
- Nvidia Driver and CUDA 9.2 for GPU based compute.
Click the following links to launch a streaming Desktop instance in your own AWS account via console and select Next to begin configuring Parameters for your desktop. Launching the Matlab stack will first require IMSS support to allow your AWS account access to the custom machine image. Please email a request, include your AWS account number to email@example.com. The base stack may be launched without any prior support request and may be launched in any availability zone. Matlab machine image must be launched in US-WEST-2 availability zone.
Click Next to confirm the self populated template link.
- Enter a unique stack name for your desktop.
- Enter a secure Login and Password.
- Select the instance type (default is M5.2Xlarge with 8 vCPUs and 32GB of memory).
- Select the SSH key name that is already uploaded or created in AWS. See here for more info https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/ec2-key-pairs.html
- For Matlab OS Volume is 32GB. Recommended to leave this setting unless there is a requirement for larger OS volume. (Data Volume is much faster and should be used for all data Input/Output).
- Larger OS Volume sizes will require longer EBS volume initialization times for optimal performance.
- ConnectLocation. Default is to allow Caltech wired, wireless and Caltech VPN only (using Tunnel-All Setting). THIS IS THE RECOMMENDED SETTINGS FOR SECURE ACCESS.
- DataVolume Size. A 6 volume EBS stripe is automatically created which each volume defaulting to 10GB for a 60GB striped volume.
- You will usually not need to set these.
- Verify the parameters and select Create to initiate the stack.
- Matlab stack will take approximately 10 minutes to configure itself and initialize the OS Volume. Base stack will take around 6 minutes. Once complete a access URL is listed under the Outputs tab. It is recommended to right click the URL and add a browser bookmark for ease of access.
- Once the URL is selected a self-signed Certificate warning will show up. Accept the certificate and continue to the site. This warning will only show up once on new stack launches.
- Login with the login username/password entered on the parameters page.
- Click within the display window and hit enter. Login with the login username/password entered on the parameters page.
- You should now be presented with your linux Desktop. Disconnect and reconnect to the running session by either closing the browser tab or by clicking the ip-xxx-xxx-xxx-xxx in the upper right-hand corner. Software processes will continue uninterrupted while session is disconnected. Copying and Pasting of text back and forth can be accomplished using the copy/paste link in the upper left-hand corner. The default account is granted sudo access which is useful for installing additional packages.
- You will need to click on the top bar of the MathWorks login menu before entering your MathWorks username and password. Failing to do so will prevent text from being entered into the field. The MathWorks username/password is not the one entered to login to the desktop itself, rather the one used for Math Works accounts online.
- Resizing the display (i.e. Browser) during an active Matlab session will cause errors as Matlab’s interface does not handle resolution changes well in a virtual environment. It is advisable to hardset a specific resolution before running Matlab by clicking the interface preferences icon in the upper-left > Display Resolution > choose a static Resolution. If however, you have already started Matlab and resizing caused errors during a critical process, right-click the Matlab toolbar, select Maximize followed again by UnMaximize. Running the optional Matlab command ‘clc’ after doing the previous procedure will clear any remaining errors on the display.