GPU Details Report

In GPU enabled virtual environments, if users to the VMs/desktops complain of slowness while accessing graphic applications, administrators should be instantly able to figure out the slowness and detect the root cause of such slowness. Administrators should be instantly able to identify which type of GPU is allocated to the users - whether Shared GPU or Pass- through GPU? Also, administrators should be able to identify which type of graphics card is utilized and how well the GPU resources are utilized? The GPU Details report helps administrators in this regard. By generating this report, administrators would be able to detect the server that is utilizing the maximum GPU resources and needs additional resources to function in an error-free manner.

To generate this report, do the following:

  1. Select the GPU Details option under the Virtualization Reports node of the REPORTS BY FUNCTION tree. Figure 1 then appears.

    Figure 1 : Specifying the criteria for generating the GPU Details report

  2. In Figure 1, select a criterion to build filter condition, so that you can selectively view the GPU Details report. The first step towards building these filter conditions is selecting a basis for the filter. This can be achieved by picking an option from the Analysis By list. The options provided by the Analysis By list box are discussed hereunder:

    • Component: Select this option to choose the component(s) from across all the managed components in the environment. For instance, for a report on details of the GPU cards used by all the managed VMWare vSphere VDI hosts in the environment, select Component from the Analysis By list, select VMWare vSphere VDI from the Component Type list, and then select all the virtual hosts listed in the Components list. If the Components list consists of too many components, then viewing all the components and selecting the ones you need for report generation could require endless scrolling. To avoid this, you can click the button next to the Components list. The pop up window will then appear using which you can view almost all the components in a single interface and Select the ones for which the report is to be generated. You can narrow your search further by using the Search text box. Specify the whole/part of the component name to search for in this text box, and click the icon next to it.
    • Segment: Choose this option if the virtual hosts to be evaluated are part of a segment. Then, pick a Segment for analysis.
    • Service: Select this option if the components for which a report is to be generated are involved in the delivery of a business service. Then, select a Service.
    • Zone: Pick this option for a report on the performance of virtual components that are included in a zone. Then, choose a Zone. A Zone drop-down list will then appear, from which you would have to select the zone to which the components of interest to you belong. A Sub Zone flag also appears. Indicate whether the components present within the sub-zones of the chosen zone are also to be to be considered for report generation, by setting the Sub Zone flag to Yes.
  3. Provide a Timeline for the report. You can either provide a fixed time line such as 1 hour, 2 days, etc., or select the Any option from the list to provide a From and To date/time for report generation.


    For every user registered with the eG Enterprise system, administrator can indicate the maximum timeline for which that user can generate a report. Once the maximum timeline is set for a user, then, whenever that user logs into eG Reporter and attempts to generate a report, the Timeline list box in the report page will display options according to the maximum timeline setting of that user. For instance, if a user can generate a report for a maximum period of 3 days only, then 3 days will be the highest option displayed in the Timeline list - i.e., 3 days will be the last option in the fixed Timeline list. Similarly, if the user chooses the Any option from the Timeline list and proceeds to provide a start date and end date for report generation using the From and To specifications, eG Enterprise will first check if the user's Timeline specification conforms to his/her maximum timeline setting. If not, report generation will fail. For instance, for a user who is allowed to generate reports spanning over a maximum period of 3 days only, the difference between the From and To dates should never be over 3 days. If it is, then, upon clicking the Run Report button a message box will appear, prompting the user to change the From and To specification.

  4. In addition to the settings discussed above, this report comes with a set of default specifications. These settings are hidden by default. If you do not want to disturb these default settings, then you can proceed to generate the report by clicking the Run Report button soon after you pick the criteria for generating the report. However, if you want to view and then alter these settings (if required), click on the icon. The default settings will then appear in the MORE OPTIONS drop down window (see Figure 2). The steps below discuss each of these settings and how they can be customized.

    Figure 2 : The default settings for generating the GPU Overview report

  5. If the timeline specified for the report needs to exclude the data collected during the Weekends, then set Exclude weekends to Yes. If not, select No.


    You can configure the days of the week that need to be considered as a ‘weekend’ using the Days parameter in the [exclude_weekend] section in the eg_report.ini file (in the <EG_INSTALL_DIR>\manager\config directory).  The Days parameter is set to Saturday,Sunday by default. To change this weekend specification, enter two other days of the week against the Days parameter.

  6. Next, specify the start time and end time for report generation against the Time period field in Figure 2.
  7. In large environments, reports generated using months of data can take a long time to complete. Administrators now have the option of generating reports on-line or in the background. When a report is scheduled for background generation, administrators can proceed with their other monitoring, diagnosis, and reporting tasks, while the eG manager is processing the report. This saves the administrator valuable time. To schedule background processing of a report, you can select the Background Save - PDF option from the Report Generation list. In this case, a Report Name text box will appear, where you would have to provide the name with which the report is to be saved in the background. To process reports in the foreground, select the Foreground Generation - HTML option from this list.


    • The Report Generation list will appear only if the EnableBackgroundReport flag in the [BACKGROUND_PROCESS] section of the eg_report.ini file (in the {EG_INSTALL_DIR}\manager\config directory) is set to Yes.
    • The default selection in the Report Generation list will change according to the Timeline specified for the report. If the Timeline set is greater than or equal to the number of days specified against the MinDurationForReport parameter in the [BACKGROUND_PROCESS] section of the eg_report.ini file, then the default selection in the Report Generation list will be Background Save - PDF. On the other hand, if the Timeline set for the report is lesser than the value of the MinDurationForReport parameter, then the default selection in the Report Generation list will be Foreground. This is because, the MinDurationForReport setting governs when reports are to be processed in the background. By default, this parameter is set to 2 weeks - this indicates that by default, reports with a timeline of 2 weeks and above will be processed in the background.
  8. Finally, click the Run Report button to generate the report. Figure 3 then appears.

    Figure 3 : A GPU Details report

  9. From the report that appears in Figure 3, administrators can figure out the Hypervisor Shared GPU Status section. By merely viewing this report, administrators can figure out the following:

    • The type of the GPU cards used by each hypervisor;
    • The name of the GPU cards;
    • The number of VMs that are currently available on the hypervisor;
    • The average and maximum utilization percentage of each GPU card in terms of frame buffers, BAR1 memory, GPU memory etc.

      Analyzing this report, administrators would be able to figure out the servers that are utilizing the maximum GPU resources and provide additional GPU resources to the needy servers.