Parsec

Uncategorized

Cloud Gaming With The Amazon G3 Instances Featuring Tesla M60 GPUs — Now Supported In Parsec And…

http://www.nvidia.com/object/tesla-m60.html

Amazon announced new GPU machines. Immediately, we wanted to add support for these in the Parsec Cloud Gaming Marketplace and via a Community AMI. The instructions for building the machines on your own are below.

More Details About The AWS G3 Instances

These cloud GPU machines are impressive and relatively inexpensive versus comparable machines offered through Azure. The Amazon G3 instances feature the Nvidia Maxwell Architecture (Tesla M60). This is a significant upgrade for Amazon over their Kepler-based GPU instances (G2s) that were released in 2013. With our Cloud Gaming Marketplace, you can build a gaming PC with the click of a button, but we know a lot of people want to build and manage their own hardware in the cloud, so we also added support for the G3 with a new Community AMI.

For cloud gaming, all you really need is the g3.4xlarge. It has 16 vCPUs using custom Intel Xeon E5–2686v4 (Broadwell) processors — I think you’ll be good there ;). As with other AWS instances, these use SSDs and take advantage of the infrastructure of AWS. They’re currently available in Ohio, Northern Virginia, California, Ireland, and California. Prices vary by region, but they’re roughly $1.14 per hour + bandwidth and hard drive costs.

https://www.microway.com/knowledge-center-articles/in-depth-comparison-of-nvidia-tesla-maxwell-gpu-accelerators/

https://aws.amazon.com/blogs/aws/new-next-generation-gpu-powered-ec2-instances-g3/

Launch Instructions For AWS G3.4xlarge Instances

  1. Before you begin, make sure the Parsec Client is installed on your computer.

  2. From the AWS EC2 Console, select Instances then Launch Instance. Make sure you are in the AWS region closest to you.

  3. When prompted to select an AMI, choose Community AMIs, then search for parsec-g3-ws2016–10.

  4. Choose GPU instances->g3.4xlarge as the instance type. This may require requesting a limit increase on this type of instance. It usually takes about 1 day for AWS to increase the limit to 1.

  5. On the Configure Instance Details page, scroll down to the Advanced Details section. Paste your configuration settings from the add computer flow and click the link at the bottom of the first step to reveal the configuration file.

  6. On the Add Storage page, specify the size of the root drive (C:). The default is 30 GB, which is just enough for Windows.

  7. When configuring your Security Group, make sure ports 8000–8040 for UDP are open.

  8. Just before launch, make sure to proceed without a key pair when prompted so the server automatically logs in.

  9. The server usually takes about 4–6 minutes to appear on your play list. During initial usage, disk access can be slow, probably due to some copy-on-write logic with AWS block storage — it should speed up over time. On our managed instances, we eliminate this problem.

  10. Please remember the g2.2xlarge instances are very expensive on an hourly basis. You should shut down your AWS instances when you’re done. Save your drive when spinning it down. You will have to pay for this storage. Also, a Parsec user created a Terraform project in Github to make this process easier. Again, we take care of this with our managed instances if you want an easier setup.

Borderless Windowed Mode

If you install games, when you launch the game, we recommend going into the game’s video settings and changing the default to Borderless or Fullscreen Windowed Mode. It will have much better performance.

Backup — Use VNC

The Parsec AMI comes pre-installed with VNC for troubleshooting purposes. VNC runs with elevated privileges and is able to function in certain situations where Parsec cannot. VNC uses port 5900, and has a default password of 4ubg9sde. Make sure to only allow connections to port 5900 from your IP, and change the default password immediately on login — please do these two things. It’s a major security risk if you don’t.

Power your remote workplace. Try Parsec for Teams