How to configure ProteinLynx Global Server 3.0.3 to use a NVIDIA Quadro GPU - WKB52140

OBJECTIVE or GOAL

Enable ProteinLynx Global Server (PLGS) 3.0.3 to use compute unified device architecture (CUDA) enabled NVIDIA Quadro series graphics processor unit (GPU) to accelerate MSe, HD-MSe and SONAR data processing.

ENVIRONMENT

• PLGS3.0.3
• Windows 7
• Windows 10
• Lenovo P720 based PLGS processing PC
• Lenovo P720 based NLD Stage 2 LM processing PC (for Progenesis QI for Proteomics)
• Lenovo P520 based NLD Stage 1 LM processing PC (for Progenesis QI for Proteomics)

PROCEDURE

1. In Windows search type Device Manager
2. Open the Device Manager and in the list of device types select Display Adapters
3. Check the installed GPUs and make a note of the installed GPUs
   1. In a Lenovo P720 based PLGS processing PC or P520 based NLD Stage 1 Large Mol PC one of the GPUs should be either a NVIDIA Quadro P5000 or a NVIDIA Quadro RTX 5000. 
4. Close PLGS3.0.3
5. Open C:\PLGS3.0.3\lib\apex3d\
6. Check whether there are any exisiting Apex3D64_params.txt files in that folder 
   1. If there are no existing command files
      1. Open a new text file. 

      2. If you found a Quadro P5000 at step 3, type -preferredGPU P5000

      3. If you found a Quadro RTX 5000 at step 3, type -preferredGPU RTX 5000

      4. Save the command file as C:\PLGS3.0.3\lib\apex3d\Apex3d64_params.txt

NOTE The above command have to be types exactly as shown above

2. If there is an existing Apex3D64_params.txt,
   1. Open that file in a text editor such as notepad.
   2. Delete any existing '-preferredGPU ' commands
   3. Beneath the any other command type the relevant command for your GPU
   4. Save the command file

7. Restart PLGS3.0.3

ADDITIONAL INFORMATION

PLGS was originally developed to only use NVIDIA Tesla series GPUs for MSe processing. NVIDIA are increasingly selling the Tesla series cards as a 'server only' product. Server GPUs can't be used in workstation PCs as they tend to overheat due to the lack of a dedicated on board GPU cooling fan. However most NVIDIA GPUs are now CUDA enabled. As a result PLGS can in theory be configured to use other NVIDIA GPUs, as long as they have enough on board dedicated GPU RAM (12GB recommended). But you have to specifically tell PLGS that it's allowed to use a non-Tesla GPU using a command file as described in this article.

The command described in this article is specific to PLGS3.0.3. It does not work with PLGS 3.0.2 or earlier