This is how you measure the load on your processor and video card

Your cpu and gpu largely determine how your system performs. If these performances sometimes turn out to be below average, then you naturally want to find out which process is responsible and what the impact is. Such a check can take place in real time, but can also take place over a shorter or longer period.

You can actually do two levels of checking how hard your cpu and gpu have to endure. Per application, where you check how much a specific process, service or application loads your processors, and system-wide, where you monitor cpu or gpu usage more globally. Both methods are discussed in this article.

We also briefly discuss what you can do with the analysis results. After all, in addition to major interventions such as installing a new processor or overclocking a current processor, processor use can also be adjusted less radically.

At the application level

You can check the processor load of a specific process from the Windows Task Manager. Press Ctrl+Shift+Esc and open the tab Processes† Through the menu Image / Update Rate you can increase or decrease the scan frequency. At the top of the columns Processor and GPU you see the global, current use of cpu and gpu, but you also get this nicely split over the active processes. Click on a column title to sort it by that criterion.

For some processes, such as browsers, you will see an arrow, click on it to see the underlying processes. Unfortunately, with Google Chrome it is always the same name (Google Chrome) although the column PID †Process ID) indicates that there are different processes involved. To identify them, open the browser and choose More Tools / Task Manager†

From the tab Details You can find a lot more information in the Task Manager. Right click on a column title and choose Select columnsafter which you can designate items as Processor time† GPU engine† Dedicated GPU memory and Shared GPU memory (see the ‘Gpu components’ box). You can also select various items related to disk I/O and memory usage here.

GPU parts

A GPU engine is an independent GPU unit, where each unit can consist of multiple GPU cores. It is perfectly possible that such an engine can run in parallel with another engine and even share the same underlying cores (similar to hyperthreading with a CPU). For example, there are engines for data transfer, encryption, 3D rendering and so on.

The designation Dedicated GPU memory refers to memory that is exclusively reserved for the GPU, either VRAM (with a discrete graphics card), or DRAM, or reserved system memory. The latter occurs with integrated GPUs, although they can also use memory that they share with the CPU (the designation Shared GPU memory†

Windows won’t let the GPU claim more than half of the available physical DRAM.

Affinity and priority

When you’re on the tab Details right-clicking on a process, you can set both the affinity and the priority. With affinity you indicate on which processor cores the process threads should be executed (by default, Windows uses all cores). The priority refers to the CPU time that a process is allocated. A process is given priority by default Normal assigned, but there are five other levels available, from Low until real time†

It is also possible to start an application by default with a specific affinity and priority. Right click on the corresponding shortcut and choose Properties† On the tab ShortcutBee Goalthen enter the following:

cmd.exe /c run “” / /affinity ““

This could be, for example:

cmd.exe /c run “EventSentry” /high /affinity 3F ” “D:Program Files (x86)ESLeventsentry_gui.exe”

The value behind /affinity requires some explanation. Suppose there are sixteen logical processors and you only want to use the first six (0000000000111111), then you press this with hexadecimal value 3F off (a converter can be found via https://kwikr.nl/rthex†

System wide (short)

To monitor your processors system-wide, you can go back to the Task Manager, on the tab Performance† Select Processor for CPU usage. Through Image / Update Rate switch between a measured time of 30 seconds to 4 minutes.

For gpu use, select GPU : GPU 0 for your integrated gpu and GPU 1 or higher for external graphics cards. If you use CrossFire or SLI for parallel GPU processing, you will notice this from the indication (Link 0) behind GPU † By default, you will see the most used engines here, but you can also select others with the arrows on the left above the graphs.

If you want to monitor gpu usage for a little longer, consider it portable GPU-Z to use. Start the tool and select your graphics adapter at the bottom left. Open the tab Sensors: via the button Settings select the desired data as well as the scan frequency. With arrow buttons you indicate which values ​​you want to see.

Place a checkmark Log to file bottom left to save the data in a txt file. You can study this with Excel, but Generic Log Viewer works more clearly. Start the tool, click Open Fileindicate GPU-Z and refer to your log file. Indicate how many diagrams and columns you want, after which you indicate in a drop-down menu which information you want in each diagram.

System wide (long)

If you want to measure processor usage over a longer period of time, you can use Windows’ built-in Performance Monitor. Press Windows key+R and run perfmon.exe.

Open the section Data Collector Sets and right click on User defined† Select New / Data Collector Sets† Enter a name, choose Create manually (advanced) and click Next one†

You choose from three data types:

– Components of Windows or of applications that report events (Event Tracking Data†

– Registry changes (System configuration information†

– Performance measurements (performance counter†

We choose the latter type here. Press Next one and on To addafter which your items from GPU or Processor marks to those with Add >> to the right pane. Once you have added all the necessary items, press OK and determine the sampling interval. Finish with Next one (2x) and Complete†

open now Data Collector Sets / User Definedright click on your set and click Start (or Stop† Open the section Reports and double click on the report to see the cpu and gpu usage in a graph.

The performance meter measures CPU usage over a longer period of time

Power Schemes

Through so-called power schemes you can somewhat optimize the global CPU and GPU usage. Launch the Power Options module from the Windows start menu, choose Create a power plan and select a schedule.

Enter a name, set the desired options and confirm with To make† Click on your schedule Change the plan settings / Change advanced power settings† Go through the available options here, such as Processor power management and […] Graphics settings and set it as desired. Confirm with OK†

†