Diagnose Your PC For Faulty Hardware

Picture this—you’re playing a game and about to pull off a headshot when all of a sudden—BAM!—it crashes to the desktop without warning. Fine, maybe it’s sloppy code on the part of the developer, but then it happens again, and in a different game this time. Uh oh, something’s wrong.

You decide it’s a sign that you should be finishing that PowerPoint presentation you’ve been working on instead of playing games, so you ignore the issue. But as you load up the program, your PC restarts itself. What is going on? Do you have a bad stick of RAM? Possibly, or maybe your graphics card is dying. All you know is that something is ailing your PC, you just don’t know what.

Well, we are here to help. Just as a doctor can run an X-ray to see a broken bone or an MRI to determine ligament damage, there are ways of determining if a component is toast. Here’s how.

Symptoms: Random restarts; crashing to the desktop; blue screen of death errors; system won’t boot; corrupted files; full amount of RAM is not being recognized.

Diagnostic Testing: Microsoft includes a free memory testing tool in Windows, which saves you the hassle of burning a third-party utility to a bootable CD or USB drive. In Windows 7, click Start and type mdsched.exe, and in Windows 10 type Windows Memory Diagnostic in the search bar. Windows 7 gives you the option of running it now or the next time you restart your PC; choose the latter for best results. Depending on your PC and amount of RAM, this should only take a few minutes. Your system will reboot when it finishes. To view the results, load up the Event Viewer (Start > Run > eventvwr.exe), expand the Windows Logs folder, right-click System and select Find, and type MemoryDiagnostics-Results. If the report indicates any errors, rerun the test with just one stick of RAM installed, if that’s an option. This way you can isolate exactly which memory module is bad.

Symptoms: Makes a grinding or clicking noise; blue screen of death errors; error messages when moving files; scrambled file names; long wait when opening a folder or file; random crashes.

Diagnostic Testing: If you hear grinding or clicking noises coming from your hard drive, further diagnosis is not needed—backup your data immediately and replace the drive, it’s on its last legs. Otherwise, a little bit of testing can let you know if you have a problem. Every hard drive maker has their own utility, and that is where you should start. Here are some links:

Western Digital: Data Lifeguard Diagnostics for WindowsSeagate: SeaToolsHGST: Windows Drive Fitness Test (WinDFT)Samsung: HUTIL (or SHDIAG and SUTIL for older drives)Toshiba: Storage Diagnostic Tool

In Windows, there is also a built-in tool called Chkdsk that scans disks for corrupt files and bad sectors, the latter of which can be software based or represent actual physical damage. The easiest way to run it is to right-click on your drive, select Properties, click on the Tools tab, and press the Check button. Another way is to open up a command prompt and type chkdsk x: /f, where “X” is the letter of the drive you want to check. Adding the “/f” command instructs the utility to try and repair any errors it finds.

Symptoms: Slower than usual file transfers; blue screen of death errors; error messages when moving files; scrambled file names; long wait when opening a folder or file; random crashes.

Diagnostic Testing: Unlike mechanical hard drives, there are no moving parts on a solid state drive. That means you will never hear a grinding or clicking noise from your SSD. That said, some of the same testing methods apply. There are way too many SSD manufacturers to list them all out, but check your SSD maker’s website for a diagnostic utility. Intel, for example, offers a Solid State Drive Toolbox that can report the health of your SSD and even estimate the remaining drive life. We also recommending running CrystalDiskMark, a free benchmarking program that also examines the drive health of your SSD.

Symptoms: Crashing out of games; artifacting in games or on the desktop; black screens; slower than expected framerates.

Diagnostic Testing: A bad graphics card or one that is overheating will usually make itself known rather quickly, typically by presenting screen abnormalities. You might see flashing colors, textures could start stretching in extremely weird ways, or the screen may even go blank at times. The best way to determine if your graphics card is defective (either the GPU or graphics card memory) or overheating is by running a stress test or a graphically demanding benchmark, such as 3DMark. The free version (Basic Edition) will suffice. Run the Time Spy test, which is a DirectX 12 benchmark, and any others that are appropriate for your system (as outlined by each test’s description). You can also stress test your card with FurMark, though don’t run it for more than a few minutes. FurMark is designed to overheat your GPU, and heat is not good for components. In short spurts, however, FurMark is a useful tool for exposing a bad graphic chip or memory.

Symptoms: Computer won’t boot to Windows; PC powers on and turns off almost immediately; Windows freezes; blue screen of death errors.

Diagnostic Testing: It’s pretty rare that a CPU is actually defective, but it does happen. In most cases, a bad CPU will prevent your system from booting. That makes troubleshooting a little tricky, because there are a million and one things that can stop your PC dead in its tracks. If you can’t get your system to post, carefully reinstall the CPU, checking for bent pins in the process. This will also give you an opportunity to reseat the heatsink, which can be another culprit.

If you’re able to boot into Windows but still suspect your processor is faulty, you have two options. One is to swap out the processor for a known good one and see if the symptoms disappear, but you might not have a spare chip laying around. The other way is to stress test. A popular program for that is Prime95. It was primarily created to find new Mersenne prime numbers, but it is widely used to stress test CPUs, mostly to see if an overclock is stable. Run either the Small FFTs or In-place large FFTs test. If you’re not overclocking and it crashes, you might have a bad CPU. You can also run OCCT. It was designed to test for overclocks, but like Prime95, it can help identify potentially faulty silicon.

Symptoms: System won’t boot; random crashes and reboots; frequent freezing/lockups; makes a grinding noise (PSU fan).

Diagnostic Testing: A bad power supply can cause all sorts of problem, everything from instability to actually frying your components. We have seen PSUs go up in smoke (literally), and it’s not a pretty sight.

If you happen to have a spare PSU, swapping it out is one way to determine if you have a defective unit. It’s also possible that your components are trying to draw more wattage than your PSU is capable of providing. To get an idea of what size PSU you should be running, use an online PSU calculator. There are several out there, many of which are powered by OuterVision.

Assuming your PSU is beefy enough for your setup, you can purchase a power supply tester to see if there’s a problem with the unit itself. The cheapest options cost less than $10 and have a series of diagnostic LEDs, while $15 will get you a fancier version with a built-in LCD.

Symptoms: System won’t boot; peripherals are not recognized; blue screen of death errors; instability.

Diagnostic Testing: Since the motherboard is the command center for your entire PC, practically every symptom imaginable can potentially point to a bad board, though the most common symptom is a failure to boot. Keep things simple when troubleshooting. Start with a visual inspection. Look for obvious signs of trouble, such as cracks, scratches, and leaking or bulging capacitors.

If everything checks out, make sure your display cable is securely fastened to both your monitor and PC’s output. You can also try connecting your display cable to your motherboard’s video output instead of your graphics card. Sometimes a BIOS can be configured to only output through a PC’s onboard graphics. Once you’re receiving a video signal, you can go into the BIOS and change that setting so that you can use your graphics card instead.

Barring a video issue, the only other thing you can do at this point is rule out your other hardware by checking them with another system, if you have one handy. For example, if your graphics card works in a spare PC but not the one you’re troubleshooting, then you can assume the GPU is not the issue. The same goes for your RAM, CPU, and power supply. It’s a pain in the backside, but less of a headache than RMA’ing components at random.