Can This PC Be Saved?

fruitysalad on 04/04/2023 - 08:15
Last edited 04/04/2023 - 13:14

Hi Ozbargainers and PC experts,

I have a 4 year old PC and its GPU seems to be dying - crashes regularly, more often when playing games, but also in normal use. I'm wondering if it can be 'saved' with minimal spend by replacing just the GPU.

Here's the spec:
CPU: AMD Ryzen 5 3500X 6-Core Processor, 3593 Mhz, 6 Core(s), 6 Logical Processor(s)
Motherboard: BIOSTAR B45M2
RAM: 16 GB
GPU: AMD Radeon RX 5700 8GB
PSU: Allied 750W

I don't need to play games at their highest settings but I do need a reliable PC … can I save this PC? Or is it time to get a new one?

Thanks for your advice!

Computing Computer Desktop Computer

Comments

  • geekcohen on 04/04/2023 - 08:22
    +4

    Might be a motherboard issue. I had an issue with two PCs that was about 4 years old (same hardware for both as I built 4 of them for work) and they were freezing. I thought it was the GPU, turns out it was the motherboard.

    If it is the motherboard, best for a new build, which is what I ended up doing.

    • lew380 on 05/04/2023 - 09:19

      Recently thought my 3090 (from ebay) was dying. replaced the cheapo PSU which didn't fix it, then replaced the cheapo MBO which fixed it.

    • groook on 17/04/2023 - 03:02

      More than likely, the cheapo Biostar boards that come with the Techfast PC's are absolute junk

  • Crosscade on 04/04/2023 - 08:29
    +2

    I had a problematic GPU that I was able to that I was able to make limp along until a got a new computer by downclocking its core and memory clock speeds.

    Try underclocking the GPU using an overclocking utility to see if that makes the computer more stable.

    If you have any friends with desktop PC's you could always take your GPU out and try to test it on their computer. If you get crashes, highly likely that it is your GPU.

  • iDroid on 04/04/2023 - 08:46
    +13

    What makes you think it was the GPU?

    Can you elaborate on what "crashes regularly" entails?

    • mskeggs on 04/04/2023 - 08:54
      +3

      Agree. Start with proper trouble shooting. Much more likely to be software than hardware.

  • EightImmortals on 04/04/2023 - 08:57

    GPU, have you tried totally removing the GPU drivers/reinstalling?

  • djsweet on 04/04/2023 - 09:37
    +2

    I would try the following steps in approximately the order they are listed. Alternatively, if you are looking for a reason/excuse to buy a new computer, skip to step 6.

    1) Turn it off and on again
    2) Reinstall up-to-date GPU drivers
    3) Clean install Windows and up-to-date drivers
    4) Ensure all components and cables are properly seated/connected.
    5) Test GPU on another "known good" computer (if available) and see if the issues are replicated. If they are, borrow a GPU from a friend. Put it in your box and see if you issues are resolved.
    6) Its pretty old, time for a new machine.

    • FireRunner on 04/04/2023 - 10:30
      +10

      4 years isn’t that old. I expect a PC to last longer than that

      • djsweet on 04/04/2023 - 10:38

        Yeah I would too. But… from a tax perspective, you depreciate desktops over 4 years. So if you are using it "for work" then maybe worth an upgrade ;)

        • FireRunner on 04/04/2023 - 10:46

          Being an AM4 system you can get a Ryzen 5600 CPU upgrade plus new GPU for a great upgrade without needing to replace the whole system.
          Assuming the issue is the GPU, it could be the motherboard or PSU.

  • Mechz on 04/04/2023 - 09:39
    +5

    Could be the PSU, could be the RAM, lots of things to test and check before spending money.

    • MilkDrinker on 04/04/2023 - 10:32
      +2

      ^ This, I reckon its more likely the PSU slowly failing. A bit early for his GPU unless he was blasting it on bitcoin mining.

  • Loot N Plunder on 04/04/2023 - 09:52
    +4

    Rip out the gpu and try the onboard graphics to eliminate it is the gpu causing the problem

    https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&c…

    1 x HDMI Connector ,resolution up to 4096 x
    1 x VGA Port ,resolution up to 1920 x 1200 @60Hz

    • FireRunner on 04/04/2023 - 10:31
      +5

      The 3500X doesn’t have integrated graphics

    • Sinnerator on 05/04/2023 - 07:57
      +1

      But GPU may be causing the PSU to stress, which is causing the problem.

  • pharkurnell on 04/04/2023 - 09:55
    -1

    Does the mobo have onboard graphics?

    Remove card and use onboard graphics to see what happens?

    • mrkorrupt on 04/04/2023 - 10:31
      +1

      No onboard graphics as the CPU doesnt have it

  • PensionerXXL on 04/04/2023 - 10:00
    +3

    have you checked the temps? sometimes temps cause crashing, and high temps may be from a lot of dust build up in the pc/gpu heatsink
    so a good clean may drop temps & prevent crashes

    other steps may be to do maintenance on the gpu, replace thermal pads and apply new paste
    repaste the cpu

    the clock speeds at 3600mhz seem a little low, that kinda indicates thermal throttle (cpu being too hot) my old 3700x would boost to 4.2ghz with the crappy stock cooler on it

  • misfitt1911 on 04/04/2023 - 10:35
    +1

    I had similar issues when my SSD was failing.

    • FireRunner on 04/04/2023 - 10:53
      +1

      This one would be easy to check, use CrystalDiskInfo to check drive's health

      • jv on 04/04/2023 - 11:11
        +1

        use CrystalDiskInfo to check drive's health

        The Shizuku Edition?

  • jv on 04/04/2023 - 11:09

    I'm wondering if it can be 'saved' with minimal spend by replacing just the GPU.

    yes

    Thanks for your advice!

    you're welcome…

  • FoxJump on 04/04/2023 - 12:31
    +1

    Have you cleaned the fans?

  • Boioioioi on 04/04/2023 - 12:31
    +1

    I had a similar issue and it turned out to be the PSU. Couldn't provide power required for the GPU under high load

  • niknah on 04/04/2023 - 12:49
    +3

    Here's how to find which part is faulty…

    Memory: Press start, type in "MdSched.exe"

    Fans: Run whatever program is crashing your computer and check the temperature, some times fans get stuck when they're old. https://openhardwaremonitor.org/

    Dust: Get a dust blower, put on a facemask + glasses. Open it up outside and blow the stuff out, cockroaches also like to make their homes in a warm places too.

    Disk: Right click on the "my computer" > each of your drives and go to properties > tools, check the disk. Somewhere in the properties you can see how much the disk has been used and the longevity.

    CPU / GPU: You can get GPU test programs like furmark. There're similar programs for the CPU.

  • fruitysalad on 04/04/2023 - 13:00

    Wow thank you for all the great expert advice! Seems like there's a lot more troubleshooting that I should be doing before looking to replace - will definitely do those this weekend.

    More detail on the 'regularly crashing' and why I suspect the GPU:

    • When PC is turned back on for first time on a day, within an hour of normal use the sound would suddenly distort (as if something else in the system is struggling). If left like this, system would usually freeze, the GPU driver disables, and have to be restarted.
    • Strangely enough, if I catch this and restarts right away (without waiting for the freeze), the PC would run fine until it's turned off again.
    • It gets to the point that the first thing I do when turning on the computer is play random Youtube/music, wait for the sound distortion, and then restart the computer.
    • Playing some games (recently Gloomhaven on PC) would sometime freeze, the GPU driver disables, and has to be re-enabled and restarted.
    • PC also had a bit of a history - have been sent back to the builder twice - one time no issue was found, the second there was a PSU problem and was replaced.
    • I did the fresh GPU driver reinstall a few times, doesn't do a lot, will try that again and the other troubleshooting steps.
    • scrimshaw on 04/04/2023 - 13:05

      You haven't mentioned what the power supply was. What is the make / model of the PSU?

      • fruitysalad on 04/04/2023 - 13:14

        Ah right - it's Allied 750W.

        The old PSU was AYWUN 700W and apparently it was not the right PSU / cable - at that time there was actually a burnt cable as a result.

        • Boioioioi on 04/04/2023 - 13:19
          +4

          That is a shitty generic power supply, my money would be on that dying. Wouldn't hurt to replace it with something reliable even if it isn't broken, because it probably will soon.

          Something from the likes of EVGA, Corsair etc.

        • scrimshaw on 04/04/2023 - 13:23
          +6

          The Allied power supply is a low-cost budget part, it's a re-labeled TechFast or AlliedGaming product that is manufactured by some other low-tier manufacturer.

          There's a medium chance that it could be a power supply issue given that it's not high quality.

          What you should do is test the parts that are easy to test first — the graphics card should be swapped out (check FB or eBay and see if you can grab a cheap graphics card for less than $30 or so). Example. See if you can replicate the crashing after swapping the GPU.

          RAM issues are not very common, but it also costs nothing for you to run the system with just 1 RAM stick. If you want to, remove one of the sticks and try them in different slots.

          If RAM / GPU not at fault then you'll need to start suspecting either the power supply or the motherboard.

    • deme on 04/04/2023 - 13:36

      Have you reinstalled Windows?

    • aileron on 04/04/2023 - 16:04

      Razer software was the reason the last time I saw this happen. I fixed it by reinstalling razer.
      Razer or not, I do think it's a software/driver issue for you - hardware problems either work or don't work, no wishy washy in-between

      • FireRunner on 04/04/2023 - 16:37
        +1

        I had a PSU that would intermittently crash the PC. Can be still be a hardware issue

  • deme on 04/04/2023 - 13:35
    +1

    Run something like https://www.ocbase.com/

  • garetz on 04/04/2023 - 13:56
    +3

    The first thing you do on a pc of that age is replace the cpu thermal paste, after 4 years you can almost guarantee that it has dried out.

    • Boioioioi on 04/04/2023 - 17:04
      +1

      You could use software to check whether or not it is overheating

    • AndyC1 on 04/04/2023 - 17:41

      I have not seen thermal paste supplied or used on most stock CPU's for a long long time.

  • AndyC1 on 04/04/2023 - 17:40

    Ask your friends who know about computers if they have a GPU that will fit in the PC and try it to see if it is the GPU.

  • shmoney on 04/04/2023 - 20:14
    +1

    sounds like a power supply issue.

  • CascadeHush on 05/04/2023 - 07:42

    Last time I had this issue it turned out to be a loose Sata cable and the fact that I had virtual PC installed.

    PC troubleshooting is bananas.

    • bamzero on 05/04/2023 - 10:21
      +1

      Sometimes just reseating everything is all it takes.

  • Sinnerator on 05/04/2023 - 08:01

    https://www.youtube.com/watch?v=IPsi6DGkrIA
    This is for troubleshooting a laptop, but if you follow the steps it will fix your problem also.

  • OBEY YOUR MASTERS on 05/04/2023 - 08:42

    Replace the thermal paste.

  • skillet on 05/04/2023 - 09:36
    +2

    Start off by doing some free $0 test.

    Do a test on your RAM, let it run for 24 hours. Problematic RAM will usually show up within the first hour.
    Memtest86
    "But I've never seen any broken RAM before".
    Then what is Memtest for?

    Do a smart info on your storage
    Crystaldiskinfo

    Do a stress test on your GPU
    Furmark
    This also some how tests the PSU because your RX5700 can draw up to 180watts. You can run CPU benchmark at the same time to max out power draw. This will not only push the PSU and also the power management components especially the VRM on your mobo. However this will not narrow down the problem but only confirm it's a hardware issue.

    I do not recommend Furmark or any GPU stress test for long hours on a laptop. Your GPU does not constantly max out during a game play. Unless your two inch thick laptop most has rubbish cooling and localised heat will cause thermal stress on the surrounding components especially the MLCC caps. Once they crack you get a short then say goodbye to your laptop.

    Do a test on the OS.
    LinuxMint
    Download and use it for a couple of days.
    If it doesn't crash it could be a problem with your Windows install.

    Have you overclocked and overvolt before? Borrow a spare GPU and PSU will help narrow down the problem.

Login or Join to leave a comment
Login Join