How To Test AMD Radeon GPU RAM memory using Linux Scripts

แชร์
ฝัง
  • เผยแพร่เมื่อ 19 ก.ย. 2024
  • LER #239 How to use Linux scripts to test GPU RAM
    Wanna know? You just have to press PLAY.
    Software links in this video:
    Rufus (english)
    rufus.ie/en/
    Arch Linux
    archlinux.org/...
    memtest & dmgg scripts:
    drive.google.c...
    I work in collaboration with:
    The Electronics Channel (live streams with Carlos and Detlef)
    / @theelectronicschannel
    Gran Canaria Uncovered
    / @grancanariauncovered (with Detlef and Julie. Not electronics related)
    Det Builds Stuff - / @detbuildsstuff (Detlef)
    Retro Upgrade - / @retroupgrade (Carlos)
    For All Your PCB needs: free $5 discount coupon
    www.pcbway.com...
    Equipment used in my videos. These are affiliate links, you pay the normal price and I make a small commission.
    TEST METERS
    ANENG AN8009 MULITIMETER
    s.click.aliexp...
    amzn.to/4da4Q7h
    amzn.to/4fsjkkm
    amzn.to/3yuf03q
    amzn.to/4fwQOOD
    www.banggood.c...
    KAIWEETS HT118E MULTIMETER
    s.click.aliexp...
    amzn.to/3yC0ZRe
    amzn.to/3SHIEcy
    www.banggood.c...
    VC480C+ MILLIOHM METER
    s.click.aliexp...
    amzn.to/46Gg1Cm
    amzn.to/3YwUjyn
    MESR-100 ESR METER
    amzn.to/3yxQ0Zd
    amzn.to/3WSr73A
    s.click.aliexp...
    XC6013L CAPACITOR METER
    s.click.aliexp...
    amzn.to/4cf2lzm
    amzn.to/4dw9cp5
    TM-902C TEMPERATURE METER
    s.click.aliexp...
    amzn.to/4fAz44L
    amzn.to/4fBxD69
    LCR-T4 COMPONENT ANALYZER
    s.click.aliexp...
    www.banggood.c...
    amzn.to/3X4rOHv
    amzn.to/3Agcj6h
    FNB58 USB ANALYZER
    s.click.aliexp...
    amzn.to/3MgBFnn
    amzn.to/4dyLU1H
    www.banggood.c...
    PCI POST ANALYZER
    s.click.aliexp...
    amzn.to/4dd8ihw
    amzn.to/3LTpeO7
    TL460S PLUS PCI_E ANALYZER
    s.click.aliexp...
    amzn.to/3WEsdi6
    TOOLTOP ET120MC2 SCOPE
    s.click.aliexp...
    amzn.to/4dCaz5L
    amzn.to/4dxOYLt
    FNIRSI 1014D SCOPE
    s.click.aliexp...
    amzn.to/3AdZqK1
    www.banggood.c...
    NPS3010W 30V 10A PSU
    s.click.aliexp...
    amzn.to/4dzl2yM
    T12 STATION WITH M8 9501 HANDLE
    s.click.aliexp...
    amzn.to/46E73W6
    M8 9501 HANDLE
    s.click.aliexp...
    YIHUA 982 Soldering Station C210/C245
    s.click.aliexp...
    FNIRSI HS-02 PORTABLE SOLDERING C210/C245
    s.click.aliexp...
    amzn.to/3Yv7Ma5
    QUICK 861DW
    s.click.aliexp...
    amzn.to/3WDF4B5
    amzn.to/4ci7rdW
    PROS'KIT SS-331H
    s.click.aliexp...
    INFIRAY P2 PRO IR CAMERA + MACRO LENS
    s.click.aliexp...
    www.banggood.c...
    amzn.to/4coSRle
    amzn.to/4coSV4s
    TOOLTOP ET13S THERMAL CAMERA METER
    s.click.aliexp...
    ET13S MACRO LENS
    s.click.aliexp...
    OPTICAL MICROSCOPE
    s.click.aliexp...
    usa.banggood.c...
    amzn.to/4dzBA9G
    TOMLOV TM4K AF FLEX
    amzn.to/3LY0YKs
    tomlov.com/pro...
    TL866 II+ PROGRAMMER
    s.click.aliexp...
    amzn.to/3SGKL05
    CH341 PROGRAMMER
    s.click.aliexp...
    NC-559-ASM FLUX
    s.click.aliexp...
    DESOLDER BRAID (I use size 8045)
    s.click.aliexp...
    s.click.aliexp...
    HX-T100 SOLDER (0.6MM)
    s.click.aliexp...
    ESD-11 TWEEZERS
    s.click.aliexp...
    Learn Electronics Repair is now on Discord! Come and join the fun, it's free.
    / discord
    If you would like to support this channel
    You can send donations
    www.paypal.com...
    You can subscribe to Patreon
    / learnelectronicsrepair
    You can click Join to become a channel member
    Thank you
    Richard

ความคิดเห็น • 84

  • @nicolasnicolas3889
    @nicolasnicolas3889 2 ปีที่แล้ว +7

    Just found your channel, amazing, I have some electrical experience, kinda basic but thorough and still have my tools and test gear, I'm gonna learn me some of this, need to find a few really exiting skips, thanks buddy, this is pure gold! ;D

  • @Chr0meTiGeR
    @Chr0meTiGeR ปีที่แล้ว +2

    This is great to know. I'm trying to repair old GPUs myself, just as a hobby, and I never knew tests like that existed. Time to scavenge through my (well-filled) GPU scrap metal crate again and maybe save a few more :D

  • @dejnio580
    @dejnio580 2 ปีที่แล้ว +2

    It is working on my HD5830 Sapphire Xtreme, big thanks Richard!

  • @andymark3664
    @andymark3664 11 หลายเดือนก่อน +5

    🆘 Help!
    09:27 At that point I fail.
    cd /run/archiso/bootmnt doesn't work, has something changed in Arch Linux?
    Says: no such file or directory.
    How do I get to the directory where dmgg.py is? It's like in the movie, on USB stick.
    I've already tried commands like: media, volume, but also with the message: no such file or directory.

    • @andymark3664
      @andymark3664 11 หลายเดือนก่อน

      Found the error. The script works.

    • @MedBuzzi
      @MedBuzzi 11 หลายเดือนก่อน +1

      What was the error i have the same one

    • @KyleSand
      @KyleSand 4 หลายเดือนก่อน

      @@andymark3664 what was the error? I'm trying to figure this one out too

  • @northwestrepair
    @northwestrepair 10 หลายเดือนก่อน +7

    these scripts are garbage. use tserver for amd or mods mats for nvidia
    thats it

  • @rafak9177
    @rafak9177 2 ปีที่แล้ว +4

    I'm a complete beginner when it comes to electronics repairs and really enjoy your videos, a real fountain of knowledge :). To the point though...
    Some time ago I've seen a video with an interesting method of identifying faulty memory chips on a graphics card (cannot find it anymore! :( ). The idea was to use a multimeter with a resistor attached to it (don't rember how, you'll figure it out with your knowledge) and probing the chips with such "device". It was causing artifacts on otherwise correct sections of the screen from the working chips, and also increasing artefacting/blinking on the faulty ones. Maybe you can make use of this method?

    • @renter84
      @renter84 ปีที่แล้ว

      This only works if you have vertical lines with artifacts. Just search "AMD Memory Testing Guide" in google. You basically short to ground a memory chip with a resistor (about 300-500 ohm - see for yourself which value works better). This will provoke vertical lines on top of bad chip vertical lines.

  • @anthonydenn4345
    @anthonydenn4345 2 ปีที่แล้ว +2

    Thanks for sharing more interesting card repair info. That software is great. I'll look forward to see how useful it is. I'm guessing you will be implementing it into future card repair video's.

  • @GregM
    @GregM 2 ปีที่แล้ว +1

    One point regarding using the mount command. As you are booting off a USB flash drive that device would have to be already mounted to read the various Linux files and scripts. The mount command would however be needed if say you were using a Linux machine and had these script files on an external USB flash drive. That would be when the mount command would allow accessing the USB flash drive.

    • @LearnElectronicsRepair
      @LearnElectronicsRepair  2 ปีที่แล้ว

      Thanks Greg. As mentioned I am no expert in Linux but thanks to your excellent explanation of the mount command I have learned something here 🙂

  • @uzairfarooq273
    @uzairfarooq273 ปีที่แล้ว +3

    I cannot go to directory cd /run/archiso/bootmnt. It says no such file or directory please help

    • @mberring1
      @mberring1 ปีที่แล้ว +3

      same problem

    • @uzairfarooq273
      @uzairfarooq273 ปีที่แล้ว +5

      @@mberring1 i fixed it by using older versions of arch linux.

    • @mberring1
      @mberring1 ปีที่แล้ว +1

      @@uzairfarooq273 thank you bro!

    • @platinumsteel6458
      @platinumsteel6458 ปีที่แล้ว +4

      2020 version works fine in case anyone is wondering and have this same issue. Thanks to @Uzai Farooq for pointing out this.

  • @HajeejiAsjedh
    @HajeejiAsjedh 10 วันที่ผ่านมา

    When I am watching this, I had a good sleep 😅😅

  • @silmarlis9449
    @silmarlis9449 2 ปีที่แล้ว

    Thank you for your useful instructional videos

  • @seeithappen1
    @seeithappen1 2 ปีที่แล้ว

    This is really indeep going diagnosing, thanks for showing us.

  • @imousschannel
    @imousschannel 2 ปีที่แล้ว

    Thanks for sharing , i will try it on my RX570 🙏

  • @nunoduarte419
    @nunoduarte419 ปีที่แล้ว +1

    In my case I tested with 10 Mb and the test was ok...then I put 50 Mb and give an error in chip 7/8.

  • @fromhell1
    @fromhell1 2 ปีที่แล้ว +1

    when I run "python3 dmgg.py e0000000 10 16" (my r9 290 have 16 chips) i got a error "xz decompression failed, data probably corrupt" OSerror errno 5

  • @jasmall5503
    @jasmall5503 2 ปีที่แล้ว

    It’s amazing… God bless you 🙏

  • @SkippyTheCanadianKangaroo
    @SkippyTheCanadianKangaroo ปีที่แล้ว

    Thank you sir for this video, i have tried many others methods to detect on which VRAM module is broken on my 5700xt. it has told me that slot 7 and/or 8 is broken. but it only says 1 is faulty. hopefully this works!

  • @NewbFixer
    @NewbFixer 2 ปีที่แล้ว

    Outstanding Sir.

  • @vasiliynet3425
    @vasiliynet3425 ปีที่แล้ว

    Awesome! Thank you!

  • @milanterzic3980
    @milanterzic3980 2 ปีที่แล้ว +1

    Hello Sir, what do you think, is it better to use this program for test amd cards or "the lord of gpu" program that you show us earlier... Wish you the best.

    • @bjones1794
      @bjones1794 ปีที่แล้ว +1

      Lord of the GPU won't work with Navi. Just tried it myself and the Linux system can't recognize the card and won't post.
      Even if using onboard graphics to post though, kings_overkill replied to comments in his own video that Lord of the GPU does not work for Navi.
      Appears this is the only option at the moment

  • @fernandofgg
    @fernandofgg ปีที่แล้ว +1

    Dont work, error: wrong fs type, bad option, bad superblock on /dev/sdb, missing codepage or helper program, or other error. Dmesg1 may have morr information after failed mount system call. 😢

    • @ch.wey.4406
      @ch.wey.4406 10 หลายเดือนก่อน +1

      I have the same bullshit. Everything tried! 15 USB sticks 390 different ISO softwares CSM to UEFI in the bios .... Damn bullshit!!!

    • @max0r4axor
      @max0r4axor 6 หลายเดือนก่อน

      Sdb is the device you need to mount the partition it should be something like /dev/sdb1

  • @fernandofgg
    @fernandofgg ปีที่แล้ว

    I will test it now!

  • @jeong1201
    @jeong1201 2 ปีที่แล้ว +1

    Thanks for the guide video. I tried it with 6800xt. 10mb works fine with no errors. From 13mb onwards, there is noise on the screen and an error appears. Is there any problem?

    • @rafaelmateus9158
      @rafaelmateus9158 ปีที่แล้ว

      The same here with rx 470 that works fine. I belive this is a bug, so test all gpus with only 10mb.

  • @nicolasnicolas3889
    @nicolasnicolas3889 2 ปีที่แล้ว

    I have an old radeon 7080 or something from the early noughties, maybe I'll dig it out....

  • @jasonbevan6275
    @jasonbevan6275 2 ปีที่แล้ว

    I wait with enthusiasm for each Learn Electronics Repair videos, and this was a good one but can you tell me why do you test on 10MB of the 4GB of RAM?

    • @SaschaElble
      @SaschaElble 2 ปีที่แล้ว +1

      Typically the whole chip goes bad, testing 10mb of each chip is usually enough to catch it. (It likely also spreads the 10mb per chip, across the whole chip.)
      This is how the nvidia version does it.

    • @LearnElectronicsRepair
      @LearnElectronicsRepair  2 ปีที่แล้ว +4

      Yeah, what Sascha Elble says. Any RAM fault sufficiently severe to stop the GPU displaying an image, artifact or crash when driver loads, would show up very quickly once you start the mem test. Having said that you are free to rest as much of the RAM as you like - but it will take longer the more you test.

  • @mconfig2630
    @mconfig2630 2 ปีที่แล้ว

    hello, I have an Asus dual rx580 8gb, Windows recognizes it, but when installing the drivers it gives error 43, in gpuz the clock is at 0, I did the test with this script and it gives me an error in three channels, that is 3 and/or 4, 5 and/or 6, 7 and/or 8, the question is, is it possible many memories to fail?

  • @dmalka81
    @dmalka81 2 ปีที่แล้ว

    Great video. Have you seen examples of vmemory not being detected at all? Will that tool help in this case too?

    • @LearnElectronicsRepair
      @LearnElectronicsRepair  2 ปีที่แล้ว

      I can't say I have seen that. It would probably mean all the RAM was short, The RAM or memory controller had no supply voltage or the GPU memory controller was completely trashed. All of these would be easy enough to spot with a multimeter so you would never get to the point of trying to test the video ram.

    • @dmalka81
      @dmalka81 2 ปีที่แล้ว

      @@LearnElectronicsRepair i noticed that for Polaris cards when one VRAM is burnt or removed memory is not detected at all in testing software or Windows. So it was a bit painful sometimes to identify the one that failed

  • @EnergySeeker
    @EnergySeeker 2 ปีที่แล้ว

    i have a card which is in my pool cleaner and it stopped working can i send it to you to repair please ? and maybe make a video on youtube for it ?

  • @pieteboe
    @pieteboe 2 ปีที่แล้ว

    give me a few, one day i'll be just like you ... ty!

  • @inwerp
    @inwerp 2 ปีที่แล้ว

    Interrsting. Do you use tserver?

  • @Maxximilian
    @Maxximilian 2 ปีที่แล้ว

    I have seen this dmgg in Russian forum but the content was visible only if registered ... struggling through google translate and the non-English registering process I almost got a headache and let it go, have you also tested the card with tserver on that 70gig image ? unfortunately that image for amd wont work on my pc (4th line) and after some loaded lines it would stop reaching to root directory line although the first 1-3 lines containing mats work like a charm..

    • @Maxximilian
      @Maxximilian 2 ปีที่แล้ว

      Tested the dmgg code on RX470 and I think its not suitable for Polaris ones or there is something wrong with the code, the card is healthy and passes the furmark and tserver easily but with 8 chips argument mentioned the dmgg announces there are problems with all 8 chips, if I confine the test with 1-7 chips it would pass.

  • @mberring1
    @mberring1 ปีที่แล้ว

    on my archiso usb there is no bootmnt, there is only airootfs, copytoram, and cowspace . Any idea what I need to do?

    • @usamawaqas5596
      @usamawaqas5596 ปีที่แล้ว +1

      use exact version of arclinux mentioned in the video. it will solve your issue.

  • @dannymx69
    @dannymx69 6 หลายเดือนก่อน

    Does anyone have memtune files for RX 6800 XT, RX 6900 XT?

  • @nunoduarte419
    @nunoduarte419 ปีที่แล้ว

    Hi, when I try to use this linux test on my EVGA GTX 980 Ti Classified Gives me a black screen and then reboot's the computer, this means that is something wrong with my Graphic's card or is just because it is an NVIDIA? Thanks and congrats for all the videos that you make.

    • @JeffMacPherson-w8r
      @JeffMacPherson-w8r 9 หลายเดือนก่อน

      bit late here but for nvidia cards you want to use mats/mods not this software.

    • @nunoduarte419
      @nunoduarte419 9 หลายเดือนก่อน

      @@JeffMacPherson-w8r Hi. Thanks a lot for the info 🙏

  • @tarkaktas9003
    @tarkaktas9003 2 ปีที่แล้ว

    thanks

  • @bienhelado204
    @bienhelado204 ปีที่แล้ว

    How do you have integrated graphics on a mobo new enough to support UEFI? What am I not understanding?

    • @Ray-uc8ij
      @Ray-uc8ij 9 หลายเดือนก่อน

      It has in IGPU in the CPU.

  • @thunderstorm6314
    @thunderstorm6314 2 ปีที่แล้ว

    hi, does this also work for an r9 390 8gb?

  • @psioniclt
    @psioniclt ปีที่แล้ว

    Only useful if have second integrated graphics. I have rx6900xt and good graphics output, but as soon trying to install drivers... only system restore brings me back to bootable windows

    • @Ray-uc8ij
      @Ray-uc8ij 9 หลายเดือนก่อน

      Hi, I also have a 6900xt Asus Tuf with an EKWB water block on it. I bought it used and it was working fine on the system of the seller, but as soon as I put it in my system and installed fresh Radeon drivers, the driver would crash. I would then get a no signal showing up on my monitor. The card works fine with generic microsoft driver, but not the radeon driver.

    • @psioniclt
      @psioniclt 9 หลายเดือนก่อน

      Sorted, just washed with alcohol and WD contact cleaner, then letter to dry for 3 days

    • @Ray-uc8ij
      @Ray-uc8ij 9 หลายเดือนก่อน +1

      @psioniclt interesting. Well I just sent my card out to a board repair place, so hopefully they can figure out what's wrong with it.

  • @tomaszkorytkowski1399
    @tomaszkorytkowski1399 2 ปีที่แล้ว

    Will it work on Nvidia?

  • @Dutch_off_grid_homesteading
    @Dutch_off_grid_homesteading ปีที่แล้ว

    Heya, I have heart off linux buut never worked with it, it looks a bit like ms-dos/machine laugauge as I like to coll it lol

  • @aziztcf
    @aziztcf 2 ปีที่แล้ว

    Should've made everyone jump through the arch install hoops first!

  • @marcellipovsky8222
    @marcellipovsky8222 2 ปีที่แล้ว +2

    Hello Richard,
    just an idea here - Is it possible to tell the script to test 1,2,3,4,5,6,7 chips even when there are 8? If so, in theory you could find the faulty one by running the script 8 times and increasing the number of chips parameter by one on each iteration. The only thing then would be to find the chip nr. 1. (as shown in the graphic)
    For other people here - A previously released guide:
    HOW TO DIAGNOSE FAULTY VIDEO RAM CHIP ON ALMOST ANY NVIDEA / AMD GPU RTX, GTX, VEGA, R9, R7 & More
    th-cam.com/video/a4QUkoBvxuY/w-d-xo.html
    And there is one more, just can't find the link now

    • @LearnElectronicsRepair
      @LearnElectronicsRepair  2 ปีที่แล้ว +4

      Marcel Lipovsky
      The script file is open source and written in python, so anyone with some programming knowledge is free to modify or improve it. I'm not an expert but I think the chips 1-8 are all at the same address but the databus is the width of all the chips together. So for example if each chip is 32 bits then 8 chips would give a 256 bit wide word.
      Meaning at address f0000000 you would write and read 8 x 32bits simultaneously (or sequentially?), and could not address and test each chip individually.
      Having said that, maybe if you found errors at some address then you could start writing data at that address such that 7x 32 bits had for example 0x00 0x00 0x00 0x00 and the 8th one had some other test pattern sequence, and then you *might be* able to determine which chip was likely to be faulty by the bit positions where you are/are not seeing errors?
      Maybe someone with an understanding of GPU memory architecture and experience in python (or any other language you could program in) wants to reverse engineer the source code and try such a technique?

    • @marcellipovsky8222
      @marcellipovsky8222 2 ปีที่แล้ว +1

      @@LearnElectronicsRepair I am no python expert, but have a bit of programming experience in other stuff. I should be able to modify the script/add a new one, but the GPU memory architecture and addressing is out of my scope of current knowledge.

  • @Kengtrading-j9o
    @Kengtrading-j9o 2 ปีที่แล้ว

    Good

  • @nice6144
    @nice6144 2 ปีที่แล้ว

    6900XT - КАК ПРОВЕРИТЬ

  • @lukegrossotron8015
    @lukegrossotron8015 ปีที่แล้ว

    How do I disable a memory module? rx5700xt

  • @06Angry
    @06Angry 2 ปีที่แล้ว

    your right is not paid

  • @ch.wey.4406
    @ch.wey.4406 10 หลายเดือนก่อน +1

    Ich krieg langsam die Krätze mit dieser Scheisse ! Bad Superblock oder readonly bla... man kann nicht auf die Ordner zugreifen. Was eine ROTZE !!

  • @danielsatko-
    @danielsatko- 4 หลายเดือนก่อน

    not work useless ap

  • @chiptuning.rostov5608
    @chiptuning.rostov5608 2 ปีที่แล้ว +1

    a very bad test, also used it several times, in fact writes all sorts of nonsense, there is no trust in him, disappointed.

    • @ramim.5856
      @ramim.5856 5 หลายเดือนก่อน

      same result on 5 different graphics cards! 😁😁