Just a note that I'm sure you're already aware of, but the Windows auto-installing GPU driver in the background after a DDU is why the DDU instructions urge you to disconnect your internet connection prior to cleaning and rebooting (you need to have the driver installer pre-downloaded and waiting before you go into safe mode to run DDU). This way you can be 100% sure that the GPU driver installed is only the one you manually told it to install.
I had a similar issue a couple years ago. Windows update was trying to install an Nvidia driver that was different than the one GeForce Experience installed and they basically ended up fighting each other and making my PC a nightmare to use. I ended up disabling Windows update, disabling GE auto-update, and have been manually downloading/installing the latest drivers because these auto-update functions have lost my trust.
@@2435Bits True. but if you're using DDU you may as well do it there. For the record though, the first time you run DDU it asks you if you want to disable automatic driver installs, otherwise its in the settings.
Bulldozer was developed under the premiss that more compute stuff will be shifted over to the GPU. AMD bought ATI, so this seemed like a good idea. Unfortunately is wasn't. But looking at the situation today, AMD was not wrong, but a few years too early.
@@johnscaramis2515 no, Bulldozer was built under the premise that more cores would be used, which didn't happen when Bulldozer was new. The only saving grace of Bulldozer is, when 8 integer cores are used, then it's relatively fast. Otherwise it behaves like a weak 4 core CPU.
I appreciate the appreciation for DDU. I was one of the earliest contributors when Wagnard (the creator) first posted about the program on Guru3d sometime in 2013 (I think it was 2013?). The guy was hated on big time, and nobody liked what he was doing, and to see how far he managed to take it, and to know that I was supportive and very involved in the code early on, and to now see it appreciated for what it is and for Wagnard to have had that success... It's nice to see.
DDU is fine, but nothing at all beats a total wipe and provided you've setup your PC properly (set partitions if only 1hdd) all your games and stuff and still stored, you can fully reinstall windows 10/11 and be back gaming within 30mins.
DDU is useless, just un-install the drivers like normal. since DDU has come out i used it a total of 1 time, and thought it was dumb because the way the drivers are installed by manufacturers like Nvidia and AMD now, when you remove them it takes all the crap with them. back in 2013? yeah things would be left behind but not anymore, making DDU irrelevant.
Seems like a problem with the connection between the die and the substrate. This can happen, because the die and the substrate are made from different materials so they expand slightly differently under high temperatures and this causes tiny cracks over a long period of time. Buildzoid made a video about this called "Rambling about chip degradation"
The correct answer is the one of vital vram solder ball just touching the pad not melting when the card is pass in factory oven reflow. When new the pad not accumulate moisture from air. Sometimes baking it will vaporised the moisture from the pad.
Hey Jay, the 770 (TI) / 780 (TI) are very likely to have a bad microBGA solder connection from the die to the substrate of the chip (not the substrate to the board). You can circle the die with hotair at about 250-275°C for about 5 minutes and usually this fixes these cards perfectly. I did this to 2 or 3 of those cards and one was this specific one you have there iirc.
@@ArtisChronicles Cool. And that relates to the solder bump problem how? Oh right.. it doesn't. At all. Not to mention the 750 and 750Ti aren't even the same GPU generation as the rest of the higher-end 700 series cards (760/770/780/780Ti are Kepler, 750 and 750Ti are first gen Maxwell, so they're not even comparable)
It's a common issue on older nvidia's. That card is done for, you can't fix it without a new GPU. "Baking" it in an oven might temporarily revive it, but the issue is inside the BGA chip, and won't go away. Basically the connections between the silicon and it's carrier, which actually gets soldered onto the Board, crack due to thermal expansion. Louis rossman has videos which explains the issue better, and also why it isn't fixable without a new chip.
You're correct. I ran into the same issue with my Zotac 970, and the solution was simply that the GPU chip/die itself needed BGA re-solder or re-balling.
Ok this video confirms it, Jay is omniscient because: 1) This week I needed to learn how to flash a BIOS because I converted my FTW3 to a hybrid and needed to convert it back but the prepackaged tool to revert back didn't work. This video would've been a great intro into the risks etc. 2) Months earlier I was building my first custom loop and Jay did a Water-cooling for beginners video with Corsair. 3) Also a few months ago I needed to understand if an OC on the 11900K would see any improvement vs a 10900K and Jay did a video on that as well. Jay clearly is operating on a different plane of existence... Keep it up though! The info has been invaluable and taught me so much!
@@TLM-Nathan hey, so EVGA allows you to put the hybrid BIOS on an air card if you convert because the air version has 3x fans that cool different parts of the card vs the hybrid with only 2 fans and you still need to control the pump speed on the AIO (also different LEDs). In the end I was being stupidly optimistic about fitting it in my case without draining my existing loop and figured it wasn't worth the effort and wanted to revert back (I borrowed the conversion kit and I was doing it to see whether putting a waterblock on the card was worth it). The program on the forum to convert to stock wouldn't work so I had to source the original BIOS, remove the write protection which would allow it to flash the original version (the flashing tool would see it as me trying to flash a BIOS from a different type of card which can brick things hence it trying to stop you).
@Laydn. This time no, but in the end yes, albeit being pointless. In the end I went for the normal (400W) BIOS on the "main" switch and XOC BIOS on the second switch with a 450W power limit and +118% power draw. Turns out it was pointless because the card never draws more than +-380W thanks to EVGA using analog voltage controllers instead of digital so it's an Ultra Overclocking card with a binned chip hampered by poor voltage delivery. Kinda sucks but you live and learn.
Oven it!!! My 770 windforce did the same thing and I “reflowed” it and still works like a charm…. Unfortunately haven’t been able to update my card now so I’m still stuck with that 😅
Yeah for me that method worked for a while.. but i had to do it multiple times. It'd crash after 4 days then i'd reflow it back to life, repeating the cycle until i decided to flash which i did wrong i think and ended up bricking the damn thing. An r9 270x 4gb..
Pretty relaxing to see you diagnosing the GPU issue. I did recently lose my 760 to a GPU fan controller fault that made my graphics memory overheat and die. I was not as calm as you diagnosing that issue 😅
This is really interesting, because my partner has a 770 that was in his case and transported basically across the country for a long time with no support, so it literally BENT under it's own weight. It worked for me for a while, then started acting pretty similar to this one, and I'd almost bet it's got something similar going on. I never checked with CPU-Z, but I'm almost tempted to pull it out of mothballs and see next time I have to get into my case.
I had a 770 the driver started to crash the windows desktop on a very frequent basis. I blamed it on corrosion because of humidity, but who knows might have been the same problem as Jay
Jay should run Nvidia Modular diagnostic software - MODS, that should tell him which memory module is acting fucky, since this indeed is a memory issue.
This exact Zotac cooler can be put on GTX 9xx cards, probably GTX 10xx and newer too if some small changes are done to the heatsink.. A really good cooler for it's age btw.
I have a Zotac 1080 that was doing the same thing and I went through all of the same steps that you just did. What I ended up doing to fix it was reflow the BGA. So what that means is the chip is having trouble communicating through the solder joints with certain memory lanes. The easy way to fix it would be put in in a toaster oven to 250C
I love how you talk down the guy who send you the card in the beginning, yet your only approaches on how to fix the card are the most basic i can think of and they don’t even work. gj
I have one that's a few steps down, a 5870, but something happened when sending it to a friend, I need to try it on my new system and see if anything works on it.
@@MarkAntony01 you are right, it doesn't mean that I didn't try. it turns out that the address bar in explore has more features that connect it to explorer (or now Edge) so I think we really can't.. BUT I'm willing to share another "shortcut" that not all knows - if you are in a folder on cmd prompt (or power shell) and you want to get its folder on windows GUI, you can run the command "start ." (the dot is the folder in this case, you can set any folder that you want) and explore is pop-up in the folder you selected, it's kind of the opposite operation to what he showed.
I'm always fascinated when techtubers challenge themselves with problems not encountered on the daily and explains why they do the stuff they do. Jay really shows his wisdom in these videos.
I would have looked at the little capacitors on the back and front (near the slot) as sometimes they can get knocked off with carelessness or a tight case. Long shot but it would cause issues you are seeing if the memory/die is getting power just not enough power (I expect wattage to increase if driver initializes).
@Jay, why you are choosing to start Safe Mode with 2 restarts? In Win 10, start -- msconfig -- Click on Boot tab -- Check Safe boot and press OK -- restart pc into safe mode right away. When done in safe mode do the same but uncheck the Safe boot. Restart and voila Win 10 starts normally
Jay, there's a faster way to open command prompt at a folder location. When you said "you can click up here and just be like bup"... just type "cmd" there and hit enter. It will open command prompt at that folder for you.
1. You really don't need to keep the Shift Key pressed when doing the safe mode thing. Once the "Please Wait" screen shows up, it's done. 2. Windows always reinstalls the GPU drivers when initializing. To stop this, you have to kill the internet connection before booting up.
You can stop this behavior in Windows by clicking Start and typing into the "Find a Setting" search box and type "Device Installation Settings" and then select "NO"
Hi Jay, I used to rock two of these zotac 780ti in sli. One day I went to play a game and mine done exactly the same thing. I tested them individually and then pulled the defective one apart and popped it in the oven for a bit. When the time was up I then put some thermal paste on and popped it back in the system and watched it boot up as normal. I then opened the game I was trying to play and all was well. I done this about 6 years ago now. While watching your video I spoke to my mate who I gave them to when I got an upgrade and he still uses them now and they're still both going strong. I loved these cards back in the day so it was a cool little trip down memory lane. 😎
I love these troubleshooting videos, it’s fun and interesting watching your process on attempting to solve random computer problems, even if it doesn’t get fixed in the end
Now this is content that resonates with me! Even if I don’t have an older card like this. This is the shit I stay subscribed to Jay for. Love it! Can’t wait for the next one.
I few random extras to Jays short cuts, 1. If you navigate to a folder and click the file path at the top and type cmd and hit enter it will open cmd to that directory. 2. To open anything as Admin, find the application in start or the shortcut, hold ctrl and shift and then open the app and it will open as admin. 3. Another way to get into safe mode, open run, type msconfig and hit enter. click the Boot tab and select Safe boot from the lower options and click ok, now the system will always boot into safemode. you have to undo this by unticking safe boot in msconfig but if you are wanting to be in safemode to do a few things and required to restart a few times it can be useful.
Also, you can right click in the empty space in the folder while pressing CTRL and it will have an option to open a CMD/Powershell window in that location. That also works.
I always adore seeing videos like this, showing in-depth troubleshooting for components that people often think are broken but in reality, just have a bit of a software hiccup, though unfortunately for this card it's very much damaged. Excellent content as always! I'd certainly love to see more of this type of stuff from you!
@@ObscenePizza "you can also SHIFT+Right-click inside the directory and open from there" Shift+rightclick will not allow you to open the command-prompt but powershell or linuxsub-shell.... windows is just that strange.
I think you're right here. I remember the older 4870 X2 cards better since I had to revive them almost on the daily at work (RMA cases). A speck of dust between the fins and the card would overheat. A good blow out with a compressor took care of that 95% of the time :D
Of course it is a 5970, it's obvious by the fact that the DIEs are very near each other, the 6990 had the dies spread out on opposing sites with space in the middle, same with 7990. Aside I owned this card and know how it looks. Also there's a fucking sticker on the side and easy to check, it simply will say HD 5970 2 GB.
tried this on my 780Ti with device manager error code. after baking it twice and already prepared to sell for parts, the bios flash worked!! thanks jay :D
Since you showed a way to change the cmd path by opening a folder and copying its path into CMD: a shorter way is holding shift + pressing the right mouse button in that folder and then choosing option to open a cmd window in that location (again, requires holding shift while pressing the right mouse button). Just a handy shortcut, nothing more ;)
Hey Jay, friends exact same card had these same issues back at the beginning of the year. Went through these same steps based on yours and others videos on these sort of issues. Eventually went with the nuclear option, stripped it down to the board, baked it at 210°C for 8 minutes and it’s been running fine ever since. Used PCB standoffs in the corners to keep the board in free space and the die facing up. Seemed safer than using scrunched up balls of aluminium foil like Linus did.
I had a 780, same thing happened to it so I sold it to a guy for cheap. Then he told me he reflowed the solder on the gpu die and worked perfectly after that.
@@runescapefan0001, cheap is better than just e-waste. I had so many ups and down with it appearing to work but as soon as I installed the drivers it died. Once I checked the mount I figured it was likely a bad solder joint. Screws were not tight at all and some spots off the die looked like they had never seen paste. So it was worth a shot, and as Jay was trying to say, you miss 100% of the shots you don’t take 😂.
im thinking of doing this since my GPU presents this exact same problem, whenever i install any kind of driver it just dies, i tried flashing it and all but nothing, any advice?
Thankyou! I've been trying to fix a similar problem all weekend. Even though you didn't fix it was really cathartic to see the steps and thought processes you worked through.
That card was post AMD buyout, AMD bought ATi in iirc 2006, but didn't start putting their name on cards until the HD7000 series I think. Edit: you can hold shift while right clicking on an empty spot in a folder to get an option to open location in cmd (in older versions of Windows) or powershell (newer versions) as long as you don't need an admin prompt.
@@weavercs4014 yeah I figured as much, and I was actually gonna put that then decided not to :D Edit: I looked it up and it seems it's a HD5970, the HD6990 has the fan central and has more aggressive styling than the one in the video.
4:30 When you find and confirm that a video port is corrupt, never trow that issue out the window. Clean the port and try to make it work, if you can't make it work. Do a full clean on the whole GPU, you have a ultrasonic cleaner. Put the GPU in the ultrasonic cleaner for 1 hour, and make sure you see 0 dust or corrosion after. After that dry the GPU with a blower, and let it dry for 24 hours. Next day bake the GPU in the oven for 2 sessions of 10 minutes. Let the GPU fully cool off between the sessions. Assemble the GPU fully and test the broken port, still not working. If you have the skills de-solder the broken port. And try to do the Bios flash again. Only after that you can call the GPU dead JayzTwoCents. You gave up fast one this one.
@@vespa7961 heating up to around 250°C should melt most available solders, so broken solder points could come back together. Sort of a discount reflow.
If it's not memory it could be electromigration. Buildzoid explains it in a very good way. They gifted me a 780ti classified with the same problems, I cooked with the oven method and it's working since lol
@@DesiredHeadshot He do that and he'll get Louis Rossmann on his ass! Jokes aside Louis did a video on the in his opinion silly idea of reflowing solder in a oven. It was pretty scathing, though I have heard from several people for whom that worked at least temporarily. In my opinion it's a why not thing. If the alternative is tossing the card then why not try baking it in the oven? Now I've got a heat gun and some nozzles but that doesn't mean I know what TF to do with them. And it especially doesn't mean I'm going to lift every component, clean the connections and re-flow it. I would probably fuck something like that up something awful.
@@blahorgaslisk7763 Yeah its not a fix, but good luck replacing a giant BGA chip without decent equipment (often times buying a replacement chip will cost nearly as much as a working card). Heat gun should only be used when you have a diagnosed bad core tho.
@@blahorgaslisk7763 Why that? I don't think anyone who knows anything about soldering believes that you can "reflow" it with baking. That's simple physics, 'cause most alloys melt around 180 - 215 °C. I think the misconception is based on that, that people think if they heat up the oven to 210 °C, they think the solder/the GPU heats up equally. Maybe because they also "think" to smell solder scents when opening the oven, despite being just (another bunch of toxic) fumes. But: Baking a GPU is causing the PCB to flex and press this way (maybe) broken solder joints back together. That's why you often need multiple "baking" attempts - and why this is usually just a temporary fix, that doesn't replace a real repair. And sometimes, if thinking of some cards, maybe baking GPUs at a high temperature is just burning and blowing away the dirt - what, in my experience, is usually the cause of 90 % of all errors. I did "repair" so much electronics in all my life, from dishwashers to cars to PCs, by simply cleaning (all) components.😂
This knowledge is definitely something I'll keep in my troubleshooting toolbox. One thing I would like to know is if there's any benefit to upgrading gpu bioses to newer versions for an already working card.
I was thinking the exact same thing.... Got rx580 that's working but have had quite a few crashes just from normal use - gaming, might try to update the bios on that. Just don't want to brick it because of the prices for the newer stuff..... That's the reason why I'm still with rx580 f scalpers!!!!!!!
Man, that Jaytober bump is excellently done. Great sound fidelity on that - great spooky jingle. Actually, shout out to the whole production of this thing. You can tell you guys have been doing it for a long time now. Cheers :)
4:30 When you find and confirm that a video port is corrupt, never trow that issue out the window. Try to clean and fix it, i had in the past that a dirty port would cause no image on the monitor.
It's easier to get to safemode by going into msconfig, then boot, and then select boot in safemode. Wish the F5 or F8 or whatever it was, was still a thing in Windows 10 myself tho!
It is still a thing if you do the following: Command prompt (Administartor) bcdedit /set bootmenupolicy legacy. You get the F8 back this way. bcdedit /set bootmenupolicy standard. To revert back to the default (original) behavior. Be mindful of the spaces: one after "bcdedit" , one after "set" and one after "bootmenupolicy". If your version of windows is upgraded to the next one, say from 20H2 to 21H1, it will revert back to the standard mode. Same thing by doing a repair/re-install.
@@markstephens8632 If it is an Asus motherboard, you may have to use F11 instead of F8, and only after a restart, not from a cold boot, unless Fastboot is disabled in the Bios. I have both an Asus and Gigabyte motherboards. Gigabyte follows the established F8 convention, but not Asus.
@@garyr7027 it’s funny you mention that, because my 700 series Titan had a bad cap and was doing some weird stuff including artifacting But it’s usually memory in my experience
had similar issues before thought was memory and every other issue. Was just GPU sag related .. try pushing it a bit up with your hand and see if anything happens with the display. Gives indication anyway, then work it out from there
11:30 you can also just cd or even skip the tedious process of launching command line shell separately at all: open elevated shell from current path (explorer) by accessing file menu > open powershell (as admin) - ALT+F,S,A (on English win10 at least).
Great stuff. Troubleshooting and beyond repair are always interesting. Sometimes you just can't repair it (at least for a reasonable cost) and only way forward is replacing. The path to that is worth learning. Thanks!
Brian: "hey do you have that motherboard I was asking about?" Jay: "yeah totally... Aren't you missing something? ....What have you got for me? Brian: "um... I have.. um... 780 ti mildly broken, an old unsupported potentially unfunctioning ATI card... Three used condoms, oh and a stapler!" Jay: "Sold!"
The DP Port doesn't have to be broken just because you don't get an issue in there on first boot, I had a similiar issue with the GTX 1660, it first didnt work but after installing the correct drivers it did. So yeah it might still work
Hey Jay, I think you should try removing the cooler and taking a heat gun to the dye and memory. I've had good luck with that in past years on macbooks, other gpus, and they all work reliablly to this day.
its faulty vram, if you go to device manage properties, there’s error code 43, which 90% of the time its vram fault, do a mats scan it’ll reveal which channel ram faulty, can be replaced with heat gun soldering
GPU Die solder bump issues (Geforce 8400-8600 and the shitty Radeon HD6xxx come to mind). the heatgun treatment just has the potential to "accidentally" fix the actual issue when you try to pointlessly reflow the VRAM cause it's literally right next to the GPU Die, so it's gonna get hot as well...
The displayport that is broken hints at this card being connected and disconnected a lot. HDMI and other ports have low voltage on them, and that voltage will short out ports and blow things on the board. This happens on my CCTV systems I sell also.
Had same problem on an iMAC, Placed the card for 6 minutes in the oven at 230Celsius and then let it cool down together with the oven, and it's still running 2 years later.
Jay....for your info, some of these old ATI-cards are specific of which port they boot to screen from. I had an HD7990 which ONLY showed boot image from 1 specific DVI-port on the card, never from the HDMI or Displayports. So that may not be a faulty port, but hardware based. If the display port works when you extend the screen to it then it is OK!
"This is an old ATi graphic card from 2010 prior from AMD buying them" [Cue facepalm scene] AMD bought ATi in 2006 but the branding wasn't changed to AMD until 2010
My guess is an IC chip fried. The card probably had a loop leak and got fried. Its also possible there was water damage the owner wasn't aware of after changing some parts out.
780 ti had unlocked bios , why didn't you try to undervolt/underclock the memory and give it a go "bios editing" , that might have worked . pls make another video
Well if jay is willing to trade stuff for stuff, what will you take for my gtx 970? It has a problem I’ve seen nobody know or see before, it flickers the screen off and in whenever it’s usage is low
I have seen this a few times actually. Forgive me for suggesting a few things, in case you already tried “everything” but maybe one of these will help: Power management settings might be too aggressive - check Bios settings and windows to make sure you aren’t running some BS energy star power profile and disable all of the other power saving stuff. You can add them back in one by one to see which one might be the problem, if you care. I had a card that flaked out on lower power profiles AND had a weird case where a desktop had laptop style power management options in windows that it shouldn’t have had. Try setting even 1% higher voltage in afterburner. Re-seat your card Swap monitor and cable to test. I had a DVI cable act like this where the monitor was going to sleep randomly when the computer was not. Again, I am not suggesting that you didn’t try this stuff but I figured it couldn’t hurt to suggest a few things.
@@arnox4554 ya, but I have a heat station and I can remove at will and replace, I was just wondering if he had the required tools is all..don't get me wrong Jay is my fav!
11:53 the command prompt trick, there is a easiest way to do it if you don't need admin cmd. Just type "cmd" and enter on the path box (the same one that you copied the path), it will open the cmd in that folder
I have owned zero, what is this "lights" you speak of? But in all seriousness, I eschew LEDs because it's pointless. Not like I can avoid getting them on some hardware now.
The displayport issue could be a monitor setting. On my LG monitor, I have to toggle between DP 1.2 and 1.4 depending on what the thing I'm connecting supports. This card being old I'm sure is 1.2 only, so maybe that's that issue (I'm only 6 minutes into this video so if you figure this out later, my bad lol)
I'm not familiar with every monitor from all companies out there, And have not heard about this type of setting? But from a technical point of view, All versions of Display Port are backwards compatible plug and play normally... It's sounds like an odd implementation imo, Can't think of a reason anyone would do that? What model do you have out of curiosity?
@@danimayb It's an lg 34950f. Not very old honestly, just went back and checked and you can in fact choose 1.4, 1.2, 1.1. Never had to pick 1.1, but I have had to cycle between 1.2 and 1.4 depending on if my laptop or desktop was connected.
@@stephenbrothers9185 Very Nice monitor! I see, How strange? Still doesn't make sense as I rap my pea brain around why would such a thing be implemented unless someone can enlighten me lol But there you go, There must be a thought behind it somewhere.. Who knows if other monitors do this? Seems an unnecessary extra step when the port technology itself allows simple backwards play.
You could always give this thing a bath in the ultrasonic cleaner, and see if anything happens with it! Then if nothing else works, try re-flowing the solder!
I have this exact same problem with GTX 1060M on hp omen laptop, when I disable nvidia GTX 1060 from device manager, it works fine but as soon as GPU is enabled, laptop freezes and continuously goes blank and flash. The laptop is just 3 years old and the issue started when I was playing “Horizon Zero Dawn” and suddenly it crashed while I was on chrome browser for some time. I tried DDU, complete reinstall win 10, rollback bios firmware, completely disassembled laptop and cleaned CPU GPU thermal paste but no luck, One thing I noticed is when you try to switch tabs GPU usage is instantly spiked to 100% but if you try to watch some 4k/8k hdr 60fps videos on TH-cam it works fine until you don’t switch tabs or hover mouse around, there GPU and clocks everything works as expected
@@danishprince2760 yeah I don't watch Twitter but I tell you what though I always watch every video he posts here on TH-cam and I thought more than 3-5k viewers would be enough I mean hell the wan show I feel like sometimes is about the same way there's not a lot of live streamers or I should say live viewers that watch it and hell there's a lot of times where they upload it at a time where I'm at work so I can't watch it I just watch it afterwards is all. That's bogus I really like watching them because it's like reminiscing with your friends and I feel like there was a lot of good moments on the show
I was having the same problem with an old sapphire hd7850, it seemed to work fine until i tried to install drivers then the system reset and went into a boot loop. I went into bios and noticed horizontal red dotted lines and figured the card is fried. It was a brilliant card that lived a long happy life, it almost got 20 years of constant use. Now i know why sapphire cards are so sought after.
Jay I'm glad you did this video. I had a card that was the same thing. Turns out that it was my PSU that caused it. The card was a Red Dragon Vega 56. Sold it broken with the GPU shortage at MSRP.
i had the same issues with my rx480 and I've tried to flash the bios like that but it didn't work 😢 when i check in device manager it detected Microsoft basic display adapter error code 31
Hey Jason I ran into this EXACT issue with my Zotac 970, and the solution was simply that the GPU chip itself needed BGA re-soldering AKA reballing. look it up.♥ Please thumbs up so he can see it.
I am so happy you did this video. I had the same problem I refreshed the rom and I tried to reload the driver.. and this video just told me that my old Nvidia HTC 770 afterburner is just dead do to memory corruption.. so THANK YOU
I had that problem with a r9 270, it would work perfectly in desktop as long as you didn't install a driver, once you did it would get artifact all over and crashed if you put some load. It was some thermal paste that got on top of the caps and resistors next to the GPU I cleaned it and it worked just like new. Btw the thermal paste was artic mx2 which is supposedly non conductive...
17:50 you dont actually have to have pressed down the key...you can just hold shift plus left click and as soon as you see the restart icon you are safe to let go of the pressed key....i have been using this a lot so im 1000% surtain it will work
Very surprised you 1) did not disconnect internet 2) did not install the driver while you were in safe mode. Unless I'm mistaken these are basics of driver installs especially in sensitive repair cases like this...
I have a card that works in BIOS OK, and even boots to a system OK with basic Microsoft GPU driver. But as soon as you install its driver - you get a Black Screen when the system boots to desktop. Nothing to do with games. Wonder what that's all about.
Want to see a part 2 with a teardown and examination of the card. This problem is too interesting to not get a definitive answer imo.
Get Louis Rossmann involved!
Does he try to sell me a 2k monitor that's known to be shit to fix it from Samsung because it was free????????
I cant remember the channel but there is another TH-camr who tries to do board diagnostics with GPU's. BPS has sent him cards before.
@@Zetharion1 Are you thinking of Buildzoid/Actually Hardcore Overclocking?
Don't worry there won't be a part 2 🤔😃.
Just a note that I'm sure you're already aware of, but the Windows auto-installing GPU driver in the background after a DDU is why the DDU instructions urge you to disconnect your internet connection prior to cleaning and rebooting (you need to have the driver installer pre-downloaded and waiting before you go into safe mode to run DDU). This way you can be 100% sure that the GPU driver installed is only the one you manually told it to install.
I had a similar issue a couple years ago. Windows update was trying to install an Nvidia driver that was different than the one GeForce Experience installed and they basically ended up fighting each other and making my PC a nightmare to use. I ended up disabling Windows update, disabling GE auto-update, and have been manually downloading/installing the latest drivers because these auto-update functions have lost my trust.
Good point!
DDU Gives you the option to Disable automatic driver installs so disconnecting your internet is not required.
You can also disable windows update grabbing drivers under settings (somewhere...)
@@2435Bits True. but if you're using DDU you may as well do it there. For the record though, the first time you run DDU it asks you if you want to disable automatic driver installs, otherwise its in the settings.
AMD bought ATI in 2006, Bulldozer didn't exist til 2011. That looks to be a 5970 which is one of the best cards from its era.
Bulldozer was developed under the premiss that more compute stuff will be shifted over to the GPU. AMD bought ATI, so this seemed like a good idea.
Unfortunately is wasn't. But looking at the situation today, AMD was not wrong, but a few years too early.
it look very simular to my cousin's 5870, so i think you are right
@@johnscaramis2515 no, Bulldozer was built under the premise that more cores would be used, which didn't happen when Bulldozer was new. The only saving grace of Bulldozer is, when 8 integer cores are used, then it's relatively fast. Otherwise it behaves like a weak 4 core CPU.
This is HD 5970 Dual GPU , I still use one of this in my PC xd
I remeber i wanted to buy this card but couldn't afford it
I appreciate the appreciation for DDU. I was one of the earliest contributors when Wagnard (the creator) first posted about the program on Guru3d sometime in 2013 (I think it was 2013?). The guy was hated on big time, and nobody liked what he was doing, and to see how far he managed to take it, and to know that I was supportive and very involved in the code early on, and to now see it appreciated for what it is and for Wagnard to have had that success... It's nice to see.
DDU is fine, but nothing at all beats a total wipe and provided you've setup your PC properly (set partitions if only 1hdd) all your games and stuff and still stored, you can fully reinstall windows 10/11 and be back gaming within 30mins.
DDU is useless, just un-install the drivers like normal. since DDU has come out i used it a total of 1 time, and thought it was dumb because the way the drivers are installed by manufacturers like Nvidia and AMD now, when you remove them it takes all the crap with them. back in 2013? yeah things would be left behind but not anymore, making DDU irrelevant.
@@thelasthallow imagine being this miserable sounding lmao
@@thelasthallow ddu used to remove all drivers by mistake some time, and no, a fresh windows install is quicker.
@@Ionstorm explain how to move all my files to a separate partition
Seems like a problem with the connection between the die and the substrate. This can happen, because the die and the substrate are made from different materials so they expand slightly differently under high temperatures and this causes tiny cracks over a long period of time.
Buildzoid made a video about this called "Rambling about chip degradation"
The correct answer is the one of vital vram solder ball just touching the pad not melting when the card is pass in factory oven reflow. When new the pad not accumulate moisture from air. Sometimes baking it will vaporised the moisture from the pad.
Hey Jay, the 770 (TI) / 780 (TI) are very likely to have a bad microBGA solder connection from the die to the substrate of the chip (not the substrate to the board).
You can circle the die with hotair at about 250-275°C for about 5 minutes and usually this fixes these cards perfectly. I did this to 2 or 3 of those cards and one was this specific one you have there iirc.
So it's the same old Geforce 8400-8600 series shit all over again with the bump issues. Glad I skipped the 700 series back in the day
@@ArtisChronicles 750ti is the perfect one among the 700 series low power under heat and decent enough to play esport game
@@ArtisChronicles Cool. And that relates to the solder bump problem how? Oh right.. it doesn't. At all. Not to mention the 750 and 750Ti aren't even the same GPU generation as the rest of the higher-end 700 series cards (760/770/780/780Ti are Kepler, 750 and 750Ti are first gen Maxwell, so they're not even comparable)
It's a common issue on older nvidia's.
That card is done for, you can't fix it without a new GPU.
"Baking" it in an oven might temporarily revive it, but the issue is inside the BGA chip, and won't go away.
Basically the connections between the silicon and it's carrier, which actually gets soldered onto the Board, crack due to thermal expansion.
Louis rossman has videos which explains the issue better, and also why it isn't fixable without a new chip.
You're correct. I ran into the same issue with my Zotac 970, and the solution was simply that the GPU chip/die itself needed BGA re-solder or re-balling.
Ok this video confirms it, Jay is omniscient because:
1) This week I needed to learn how to flash a BIOS because I converted my FTW3 to a hybrid and needed to convert it back but the prepackaged tool to revert back didn't work. This video would've been a great intro into the risks etc.
2) Months earlier I was building my first custom loop and Jay did a Water-cooling for beginners video with Corsair.
3) Also a few months ago I needed to understand if an OC on the 11900K would see any improvement vs a 10900K and Jay did a video on that as well.
Jay clearly is operating on a different plane of existence...
Keep it up though! The info has been invaluable and taught me so much!
Why did you have to mod / change the bios just to put a watercooler on the gpu?
Either that or he's checking your search history for content ideas.
@@TLM-Nathan my guess is so that it could control the pump more precisely than it would a fan
@@TLM-Nathan hey, so EVGA allows you to put the hybrid BIOS on an air card if you convert because the air version has 3x fans that cool different parts of the card vs the hybrid with only 2 fans and you still need to control the pump speed on the AIO (also different LEDs). In the end I was being stupidly optimistic about fitting it in my case without draining my existing loop and figured it wasn't worth the effort and wanted to revert back (I borrowed the conversion kit and I was doing it to see whether putting a waterblock on the card was worth it). The program on the forum to convert to stock wouldn't work so I had to source the original BIOS, remove the write protection which would allow it to flash the original version (the flashing tool would see it as me trying to flash a BIOS from a different type of card which can brick things hence it trying to stop you).
@Laydn. This time no, but in the end yes, albeit being pointless. In the end I went for the normal (400W) BIOS on the "main" switch and XOC BIOS on the second switch with a 450W power limit and +118% power draw. Turns out it was pointless because the card never draws more than +-380W thanks to EVGA using analog voltage controllers instead of digital so it's an Ultra Overclocking card with a binned chip hampered by poor voltage delivery. Kinda sucks but you live and learn.
Oven it!!! My 770 windforce did the same thing and I “reflowed” it and still works like a charm…. Unfortunately haven’t been able to update my card now so I’m still stuck with that 😅
Did the same thing with my 1080, although it's still a bit janky so there's probably a solder that need proper fixing
tätä kommenttia tulin etsimään ;D
Yeah for me that method worked for a while.. but i had to do it multiple times. It'd crash after 4 days then i'd reflow it back to life, repeating the cycle until i decided to flash which i did wrong i think and ended up bricking the damn thing. An r9 270x 4gb..
What ever you can fix by cooking it, you can fix better by knowing what you're doing.
@@Dayto_G in your place i dont cook anything in this oven right now.... The solders make toxic fumes with heat. Much solders has Pb (Plumbum)
Pretty relaxing to see you diagnosing the GPU issue. I did recently lose my 760 to a GPU fan controller fault that made my graphics memory overheat and die. I was not as calm as you diagnosing that issue 😅
I have the same issues that is shown in that video with my gtx 760 and i dunno what to do
old monitors use to show those kind of red lines when ground was missing in their data cable...i repaired so many of them...
This is really interesting, because my partner has a 770 that was in his case and transported basically across the country for a long time with no support, so it literally BENT under it's own weight. It worked for me for a while, then started acting pretty similar to this one, and I'd almost bet it's got something similar going on. I never checked with CPU-Z, but I'm almost tempted to pull it out of mothballs and see next time I have to get into my case.
I had a 770 the driver started to crash the windows desktop on a very frequent basis. I blamed it on corrosion because of humidity, but who knows might have been the same problem as Jay
@ainsley evergarden before you run around looking for cpu-z the utility is gpu-z ☺️
that mean you maybe have some welding point broken, cooking the card sometime work
@@AustinHeard55 To add, if you think gpu-z has been closed, check your running applications. It runs in the background and you can open 100 of them.
Maybe some solder balls on the memory modules shook loose.
"this one of those demon cards where youre just like 🤷♂️" well it is 31 days of Jaytober
True
Well bake it
Jay should run Nvidia Modular diagnostic software - MODS, that should tell him which memory module is acting fucky, since this indeed is a memory issue.
I loved when GPUs looked like this! I really want UV reactive shrouds and fans to make a come back
Imagine, a double 3090 for cooling good luck
MS Windows Update (regular ~ Auto pilot ~ routine) is the main cause
I think you'll have to print your own because I don't think that's ever coming back, it's gone forever just like transparent casings.
This exact Zotac cooler can be put on GTX 9xx cards, probably GTX 10xx and newer too if some small changes are done to the heatsink.. A really good cooler for it's age btw.
@@Kattmat literally the cards I had
I have a Zotac 1080 that was doing the same thing and I went through all of the same steps that you just did. What I ended up doing to fix it was reflow the BGA. So what that means is the chip is having trouble communicating through the solder joints with certain memory lanes. The easy way to fix it would be put in in a toaster oven to 250C
I love how you talk down the guy who send you the card in the beginning, yet your only approaches on how to fix the card are the most basic i can think of and they don’t even work. gj
Jay, I still have that ATI card in a box
I really wanted him to play around with that ATI 5970, brings back some memories.
Omg
I have one that's a few steps down, a 5870, but something happened when sending it to a friend, I need to try it on my new system and see if anything works on it.
@@kevinsteinman8967 As an (unfortunately) life long Nvidia customer, AMD GPU designations are a mystery to me, what is that card equivalent to?
@@voltaicfire1825 probably a GTX 295
about the "shortcut" to the path in the command line, you can just write "cmd" in the folder address and get the prompt form where.. no copy+paste :)
But not with admin rights.
@@MarkAntony01 you are right, it doesn't mean that I didn't try.
it turns out that the address bar in explore has more features that connect it to explorer (or now Edge) so I think we really can't..
BUT I'm willing to share another "shortcut" that not all knows - if you are in a folder on cmd prompt (or power shell) and you want to get its folder on windows GUI, you can run the command "start ." (the dot is the folder in this case, you can set any folder that you want) and explore is pop-up in the folder you selected, it's kind of the opposite operation to what he showed.
Or hold shift and right click in the folder and you'll get an additional menu option to open a command prompt at that location.
I'm always fascinated when techtubers challenge themselves with problems not encountered on the daily and explains why they do the stuff they do. Jay really shows his wisdom in these videos.
I would have looked at the little capacitors on the back and front (near the slot) as sometimes they can get knocked off with carelessness or a tight case. Long shot but it would cause issues you are seeing if the memory/die is getting power just not enough power (I expect wattage to increase if driver initializes).
Yeah I have vmemory and gpu driver boot fails, and when using my cpu onboard graphics it doesn't happen. sux
@Jay, why you are choosing to start Safe Mode with 2 restarts? In Win 10, start -- msconfig -- Click on Boot tab -- Check Safe boot and press OK -- restart pc into safe mode right away. When done in safe mode do the same but uncheck the Safe boot. Restart and voila Win 10 starts normally
Jay, there's a faster way to open command prompt at a folder location. When you said "you can click up here and just be like bup"... just type "cmd" there and hit enter. It will open command prompt at that folder for you.
That's sick! Nice tip.
1. You really don't need to keep the Shift Key pressed when doing the safe mode thing. Once the "Please Wait" screen shows up, it's done.
2. Windows always reinstalls the GPU drivers when initializing. To stop this, you have to kill the internet connection before booting up.
You can stop this behavior in Windows by clicking Start and typing into the "Find a Setting" search box and type "Device Installation Settings" and then select "NO"
@@whpony96 Windows still tries to enforce display driver install even with this setting off unfortunately
Or you can disable windows from updating the driver in DDU.
@@Mildly_Amused Disabling from DDU still doesn't stop Windows from doing it, belive me, I have it from experience 🤣
@@viduani Something is very wrong then. It works perfectly on all my systems.
When the Nvidia driver loads it then becomes possible for Windows to address all the memory, but since the memory is bad it can't display an image.
+1 for heating up/reflowing the die. I've resurrected 3 out of 3 gpus with similar issues that way.
Same here I resurrected a few with same problem heating / reflow the die using a heat gun for 6 min.
Hi Jay, I used to rock two of these zotac 780ti in sli. One day I went to play a game and mine done exactly the same thing. I tested them individually and then pulled the defective one apart and popped it in the oven for a bit. When the time was up I then put some thermal paste on and popped it back in the system and watched it boot up as normal.
I then opened the game I was trying to play and all was well. I done this about 6 years ago now. While watching your video I spoke to my mate who I gave them to when I got an upgrade and he still uses them now and they're still both going strong. I loved these cards back in the day so it was a cool little trip down memory lane. 😎
Really liked the whole video Jay and honestly, would be happy to see more GPU repair stuff, preferrably for NVIDIA (since I use their GPU)
I love these troubleshooting videos, it’s fun and interesting watching your process on attempting to solve random computer problems, even if it doesn’t get fixed in the end
Now this is content that resonates with me! Even if I don’t have an older card like this. This is the shit I stay subscribed to Jay for. Love it! Can’t wait for the next one.
YT showing off they can even get old cards that we can't get..how nice
I few random extras to Jays short cuts,
1. If you navigate to a folder and click the file path at the top and type cmd and hit enter it will open cmd to that directory.
2. To open anything as Admin, find the application in start or the shortcut, hold ctrl and shift and then open the app and it will open as admin.
3. Another way to get into safe mode, open run, type msconfig and hit enter. click the Boot tab and select Safe boot from the lower options and click ok, now the system will always boot into safemode. you have to undo this by unticking safe boot in msconfig but if you are wanting to be in safemode to do a few things and required to restart a few times it can be useful.
Also, you can right click in the empty space in the folder while pressing CTRL and it will have an option to open a CMD/Powershell window in that location. That also works.
Knew someone would have beaten me to it 👍
I always adore seeing videos like this, showing in-depth troubleshooting for components that people often think are broken but in reality, just have a bit of a software hiccup, though unfortunately for this card it's very much damaged. Excellent content as always! I'd certainly love to see more of this type of stuff from you!
For that cmd command you can just type cmd in the path of the file explorer, it will open de command line in that folder.
Ah, was looking for this comment, don't have to write one then \o/
True, you can also SHIFT+Right-click inside the directory and open from there, but in both cases it won't be elevated :(
@@ObscenePizza "you can also SHIFT+Right-click inside the directory and open from there"
Shift+rightclick will not allow you to open the command-prompt but powershell or linuxsub-shell.... windows is just that strange.
Looks like a 5970 possibly, I had two in Quad Crossfire, it was insane when it worked and a massive pain when it didn’t
im kinda inclined to believe that, it looks right. same shroud, fan, etc
I think you're right here. I remember the older 4870 X2 cards better since I had to revive them almost on the daily at work (RMA cases). A speck of dust between the fins and the card would overheat. A good blow out with a compressor took care of that 95% of the time :D
it is.
Of course it is a 5970, it's obvious by the fact that the DIEs are very near each other, the 6990 had the dies spread out on opposing sites with space in the middle, same with 7990. Aside I owned this card and know how it looks.
Also there's a fucking sticker on the side and easy to check, it simply will say HD 5970 2 GB.
tried this on my 780Ti with device manager error code. after baking it twice and already prepared to sell for parts, the bios flash worked!! thanks jay :D
Since you showed a way to change the cmd path by opening a folder and copying its path into CMD: a shorter way is holding shift + pressing the right mouse button in that folder and then choosing option to open a cmd window in that location (again, requires holding shift while pressing the right mouse button). Just a handy shortcut, nothing more ;)
It’s actually really interesting seeing you test old hardware like this!
Hey Jay, friends exact same card had these same issues back at the beginning of the year. Went through these same steps based on yours and others videos on these sort of issues. Eventually went with the nuclear option, stripped it down to the board, baked it at 210°C for 8 minutes and it’s been running fine ever since. Used PCB standoffs in the corners to keep the board in free space and the die facing up. Seemed safer than using scrunched up balls of aluminium foil like Linus did.
I had a 780, same thing happened to it so I sold it to a guy for cheap. Then he told me he reflowed the solder on the gpu die and worked perfectly after that.
@@runescapefan0001, cheap is better than just e-waste. I had so many ups and down with it appearing to work but as soon as I installed the drivers it died. Once I checked the mount I figured it was likely a bad solder joint. Screws were not tight at all and some spots off the die looked like they had never seen paste. So it was worth a shot, and as Jay was trying to say, you miss 100% of the shots you don’t take 😂.
im thinking of doing this since my GPU presents this exact same problem, whenever i install any kind of driver it just dies, i tried flashing it and all but nothing, any advice?
No useless lighting on the card seems like a good thing to me...
Based
Regarding the cmd trick at 11:45 - you can also just select the editor adress bar and enter cmd in the folder that you want cmd to be run in
Thankyou! I've been trying to fix a similar problem all weekend. Even though you didn't fix it was really cathartic to see the steps and thought processes you worked through.
That card was post AMD buyout, AMD bought ATi in iirc 2006, but didn't start putting their name on cards until the HD7000 series I think.
Edit: you can hold shift while right clicking on an empty spot in a folder to get an option to open location in cmd (in older versions of Windows) or powershell (newer versions) as long as you don't need an admin prompt.
It's a 6990 from what I can tell
@@weavercs4014 yeah I figured as much, and I was actually gonna put that then decided not to :D
Edit: I looked it up and it seems it's a HD5970, the HD6990 has the fan central and has more aggressive styling than the one in the video.
@@PaulTheFox1988 oh yeah :D
4:30 When you find and confirm that a video port is corrupt, never trow that issue out the window.
Clean the port and try to make it work, if you can't make it work.
Do a full clean on the whole GPU, you have a ultrasonic cleaner.
Put the GPU in the ultrasonic cleaner for 1 hour, and make sure you see 0 dust or corrosion after.
After that dry the GPU with a blower, and let it dry for 24 hours.
Next day bake the GPU in the oven for 2 sessions of 10 minutes.
Let the GPU fully cool off between the sessions.
Assemble the GPU fully and test the broken port, still not working.
If you have the skills de-solder the broken port.
And try to do the Bios flash again.
Only after that you can call the GPU dead JayzTwoCents.
You gave up fast one this one.
What is this obsession with using the oven? Don't bake your graphics card unless you're clueless and desperate.
@@vespa7961 heating up to around 250°C should melt most available solders, so broken solder points could come back together. Sort of a discount reflow.
@@HappyBeezerStudios 250 is way too high
If it's not memory it could be electromigration. Buildzoid explains it in a very good way. They gifted me a 780ti classified with the same problems, I cooked with the oven method and it's working since lol
I was hoping Jay would do something along those lines and reflow the board then again he might not want to do it himself
MS Windows Update (regular ~ Auto pilot ~ routine) is the main cause
@@DesiredHeadshot He do that and he'll get Louis Rossmann on his ass!
Jokes aside Louis did a video on the in his opinion silly idea of reflowing solder in a oven. It was pretty scathing, though I have heard from several people for whom that worked at least temporarily.
In my opinion it's a why not thing. If the alternative is tossing the card then why not try baking it in the oven? Now I've got a heat gun and some nozzles but that doesn't mean I know what TF to do with them. And it especially doesn't mean I'm going to lift every component, clean the connections and re-flow it. I would probably fuck something like that up something awful.
@@blahorgaslisk7763 Yeah its not a fix, but good luck replacing a giant BGA chip without decent equipment (often times buying a replacement chip will cost nearly as much as a working card). Heat gun should only be used when you have a diagnosed bad core tho.
@@blahorgaslisk7763 Why that? I don't think anyone who knows anything about soldering believes that you can "reflow" it with baking. That's simple physics, 'cause most alloys melt around 180 - 215 °C. I think the misconception is based on that, that people think if they heat up the oven to 210 °C, they think the solder/the GPU heats up equally. Maybe because they also "think" to smell solder scents when opening the oven, despite being just (another bunch of toxic) fumes.
But: Baking a GPU is causing the PCB to flex and press this way (maybe) broken solder joints back together. That's why you often need multiple "baking" attempts - and why this is usually just a temporary fix, that doesn't replace a real repair. And sometimes, if thinking of some cards, maybe baking GPUs at a high temperature is just burning and blowing away the dirt - what, in my experience, is usually the cause of 90 % of all errors.
I did "repair" so much electronics in all my life, from dishwashers to cars to PCs, by simply cleaning (all) components.😂
so... in this scenario which pc part is broken?
Probably the wrong bios after all, with different memory timing messing up the visuals., use the bios on the sticker.
jay you should match the memory types when searching for the bios
samsung, hynx, etc have different bios for the same card (basically memory timings)
This knowledge is definitely something I'll keep in my troubleshooting toolbox. One thing I would like to know is if there's any benefit to upgrading gpu bioses to newer versions for an already working card.
I was thinking the exact same thing.... Got rx580 that's working but have had quite a few crashes just from normal use - gaming, might try to update the bios on that. Just don't want to brick it because of the prices for the newer stuff..... That's the reason why I'm still with rx580 f scalpers!!!!!!!
@@nicklity so have u tried it ?
Man, that Jaytober bump is excellently done. Great sound fidelity on that - great spooky jingle. Actually, shout out to the whole production of this thing. You can tell you guys have been doing it for a long time now. Cheers :)
4:30 When you find and confirm that a video port is corrupt, never trow that issue out the window.
Try to clean and fix it, i had in the past that a dirty port would cause no image on the monitor.
A 780ti.. wow I thought this video would be years older than it is.
Love these tech support videos because you're a trustworthy source that goes over so many steps without sounding condescending
It's easier to get to safemode by going into msconfig, then boot, and then select boot in safemode. Wish the F5 or F8 or whatever it was, was still a thing in Windows 10 myself tho!
It is still a thing if you do the following:
Command prompt (Administartor)
bcdedit /set bootmenupolicy legacy. You get the F8 back this way.
bcdedit /set bootmenupolicy standard. To revert back to the default (original) behavior.
Be mindful of the spaces: one after "bcdedit" , one after "set" and one after "bootmenupolicy".
If your version of windows is upgraded to the next one, say from 20H2 to 21H1, it will revert back to the standard mode. Same thing by doing a repair/re-install.
@@claudest-gelais8456 nice, i will have to try it out!
@@markstephens8632 If it is an Asus motherboard, you may have to use F11 instead of F8, and only after a restart, not from a cold boot, unless Fastboot is disabled in the Bios. I have both an Asus and Gigabyte motherboards. Gigabyte follows the established F8 convention, but not Asus.
@@claudest-gelais8456 Thanks works Great
@@Trevorodunne Glad I could help!!! :-).
The first boot when you see the artifacting, instantly, memory issue
Bad capacitors will do the same thing.
@@garyr7027 it’s funny you mention that, because my 700 series Titan had a bad cap and was doing some weird stuff including artifacting
But it’s usually memory in my experience
@@chincemagnet memory was my first thought, of course a simple GPU memtest could have saved this guy a lot of wasted time.
had similar issues before thought was memory and every other issue. Was just GPU sag related .. try pushing it a bit up with your hand and see if anything happens with the display. Gives indication anyway, then work it out from there
11:30 you can also just cd or even skip the tedious process of launching command line shell separately at all: open elevated shell from current path (explorer) by accessing file menu > open powershell (as admin) - ALT+F,S,A (on English win10 at least).
Or directly type powershell or cmd in explorer's address bar to launch non-admin
Great stuff. Troubleshooting and beyond repair are always interesting. Sometimes you just can't repair it (at least for a reasonable cost) and only way forward is replacing. The path to that is worth learning. Thanks!
Brian: "hey do you have that motherboard I was asking about?"
Jay: "yeah totally... Aren't you missing something? ....What have you got for me?
Brian: "um... I have.. um... 780 ti mildly broken, an old unsupported potentially unfunctioning ATI card... Three used condoms, oh and a stapler!"
Jay: "Sold!"
Lol Jay got scammed
@@DeCo85 I mean, he's getting content out of it which is probably more valuable than that motherboard was.
@@KevinIrish Yes, but it's funnier to think that Jay just got scammed :)
@@DeCo85 i dunno, staplers are pretty handy
Was it the 3 used condoms?
Pro tip, you can just type “cmd” in the address bar on explorer, and a cmd will open in that location (although it’s not on admin privileges)
oh... learning something new everyday!
You know, i probably should've read a bit further down the comments before i typed mine...
Nice trick, easier than shift-right click on the folder especially if you get a shortcut to powershell instead of command...
I don't understand why they seem to hate power users who need an elevated cmd
The DP Port doesn't have to be broken just because you don't get an issue in there on first boot, I had a similiar issue with the GTX 1660, it first didnt work but after installing the correct drivers it did. So yeah it might still work
Quicker tip for opening CMD with the desired directory. Go to the desired folder in explorer. Highlight the address bar, type CMD, hit enter. Done.
Hey Jay, I think you should try removing the cooler and taking a heat gun to the dye and memory. I've had good luck with that in past years on macbooks, other gpus, and they all work reliablly to this day.
Thought it was the 3070 noctua for a second
Same
its faulty vram, if you go to device manage properties, there’s error code 43, which 90% of the time its vram fault, do a mats scan it’ll reveal which channel ram faulty, can be replaced with heat gun soldering
GPU Die solder bump issues (Geforce 8400-8600 and the shitty Radeon HD6xxx come to mind). the heatgun treatment just has the potential to "accidentally" fix the actual issue when you try to pointlessly reflow the VRAM cause it's literally right next to the GPU Die, so it's gonna get hot as well...
The displayport that is broken hints at this card being connected and disconnected a lot. HDMI and other ports have low voltage on them, and that voltage will short out ports and blow things on the board. This happens on my CCTV systems I sell also.
Had same problem on an iMAC, Placed the card for 6 minutes in the oven at 230Celsius and then let it cool down together with the oven, and it's still running 2 years later.
Appreciate the uploads Jay!
"We'll hang this one on the wall with the others" - Jay's basically saying to all his HW: "Behave or you'll be hung next" 😂
my Biggest gripe even in RAM "Hynix". iI cannot stress enough, a painful memory to deal with. Like Jay said its like negoiating w the devil.
“RIP in peace
ATF fluid
New NT Technology “
All identically ironic statements
Don't forget the classic LCD Display
Jay....for your info, some of these old ATI-cards are specific of which port they boot to screen from. I had an HD7990 which ONLY showed boot image from 1 specific DVI-port on the card, never from the HDMI or Displayports. So that may not be a faulty port, but hardware based. If the display port works when you extend the screen to it then it is OK!
People like you and linus tech tips I just never wanna see your channels end
"This is an old ATi graphic card from 2010 prior from AMD buying them"
[Cue facepalm scene]
AMD bought ATi in 2006 but the branding wasn't changed to AMD until 2010
My guy
"There's no lighting on this card, that's so weird!"
Me with a PNY 1060 that has no lights at all: lmao ye
First, read about cool projects such as Flirt Invest and then tell tales about growth
My guess is an IC chip fried. The card probably had a loop leak and got fried. Its also possible there was water damage the owner wasn't aware of after changing some parts out.
780 ti had unlocked bios , why didn't you try to undervolt/underclock the memory and give it a go "bios editing" , that might have worked . pls make another video
Well if jay is willing to trade stuff for stuff, what will you take for my gtx 970? It has a problem I’ve seen nobody know or see before, it flickers the screen off and in whenever it’s usage is low
I have seen this a few times actually.
Forgive me for suggesting a few things, in case you already tried “everything” but maybe one of these will help:
Power management settings might be too aggressive - check Bios settings and windows to make sure you aren’t running some BS energy star power profile and disable all of the other power saving stuff. You can add them back in one by one to see which one might be the problem, if you care. I had a card that flaked out on lower power profiles AND had a weird case where a desktop had laptop style power management options in windows that it shouldn’t have had.
Try setting even 1% higher voltage in afterburner.
Re-seat your card
Swap monitor and cable to test. I had a DVI cable act like this where the monitor was going to sleep randomly when the computer was not.
Again, I am not suggesting that you didn’t try this stuff but I figured it couldn’t hurt to suggest a few things.
I feel that when you get to the point of visual artifacts, your memorry is likely done for.
All he has to do is take it apart and swap the memory if he thinks that's the problem....buildzoid would 🤣
@@deminybs It wouldn't surprise me if it's a bad solder joint that's decided enough is enough. Just reflowing the memory might be enough.
@@blahorgaslisk7763 true... but that's the boring way 😂
Do you have the ability to swap memory modules? that way these demonstrations can have a happy ending haha
Memory is soldered onto the board.
@@arnox4554 and? Doesn’t mean it can’t be replaced. Would it be worth it? Probably not on a card this old
@@arnox4554 ya, but I have a heat station and I can remove at will and replace, I was just wondering if he had the required tools is all..don't get me wrong Jay is my fav!
It can be done but I'm not sure Jay has the skills and I don't think he has the equipment
@@ffwast he def has the ability, but it can be tedious as I habe found out repairing a few 1060 and 70s
These videos trying to diagnose weird shit that's gone wrong with a driver are so fascinating! Thanks for the video!
11:53 the command prompt trick, there is a easiest way to do it if you don't need admin cmd. Just type "cmd" and enter on the path box (the same one that you copied the path), it will open the cmd in that folder
Jay said: "So strange no lights on the card." Me: I've only owned one video card with lights.
I have owned zero, what is this "lights" you speak of?
But in all seriousness, I eschew LEDs because it's pointless. Not like I can avoid getting them on some hardware now.
you could've tried to reflow the memory modules using a heat gun, I've had great luck with that.
The displayport issue could be a monitor setting. On my LG monitor, I have to toggle between DP 1.2 and 1.4 depending on what the thing I'm connecting supports. This card being old I'm sure is 1.2 only, so maybe that's that issue (I'm only 6 minutes into this video so if you figure this out later, my bad lol)
I'm not familiar with every monitor from all companies out there, And have not heard about this type of setting? But from a technical point of view, All versions of Display Port are backwards compatible plug and play normally... It's sounds like an odd implementation imo, Can't think of a reason anyone would do that? What model do you have out of curiosity?
@@danimayb It's an lg 34950f. Not very old honestly, just went back and checked and you can in fact choose 1.4, 1.2, 1.1. Never had to pick 1.1, but I have had to cycle between 1.2 and 1.4 depending on if my laptop or desktop was connected.
@@stephenbrothers9185 Very Nice monitor! I see, How strange? Still doesn't make sense as I rap my pea brain around why would such a thing be implemented unless someone can enlighten me lol But there you go, There must be a thought behind it somewhere.. Who knows if other monitors do this? Seems an unnecessary extra step when the port technology itself allows simple backwards play.
You could always give this thing a bath in the ultrasonic cleaner, and see if anything happens with it! Then if nothing else works, try re-flowing the solder!
I have this exact same problem with GTX 1060M on hp omen laptop, when I disable nvidia GTX 1060 from device manager, it works fine but as soon as GPU is enabled, laptop freezes and continuously goes blank and flash. The laptop is just 3 years old and the issue started when I was playing “Horizon Zero Dawn” and suddenly it crashed while I was on chrome browser for some time. I tried DDU, complete reinstall win 10, rollback bios firmware, completely disassembled laptop and cleaned CPU GPU thermal paste but no luck, One thing I noticed is when you try to switch tabs GPU usage is instantly spiked to 100% but if you try to watch some 4k/8k hdr 60fps videos on TH-cam it works fine until you don’t switch tabs or hover mouse around, there GPU and clocks everything works as expected
Brian has been getting smacked around by the algorithm lately I think, everyone go give him a view! He does some cool builds.
Who is brian
Who is brian
God I miss the RTFM show.
.... they just started that what do you mean?
@@cleysoptique8749 It's cancelled :)
@@danishprince2760 any insight on this did he talk about it? Do you know why?
@@cleysoptique8749 He's talked about it on the Twitter. Not enough viewers/listeners to make up for the amount of work and it it took
@@danishprince2760 yeah I don't watch Twitter but I tell you what though I always watch every video he posts here on TH-cam and I thought more than 3-5k viewers would be enough I mean hell the wan show I feel like sometimes is about the same way there's not a lot of live streamers or I should say live viewers that watch it and hell there's a lot of times where they upload it at a time where I'm at work so I can't watch it I just watch it afterwards is all.
That's bogus I really like watching them because it's like reminiscing with your friends and I feel like there was a lot of good moments on the show
I was having the same problem with an old sapphire hd7850, it seemed to work fine until i tried to install drivers then the system reset and went into a boot loop. I went into bios and noticed horizontal red dotted lines and figured the card is fried. It was a brilliant card that lived a long happy life, it almost got 20 years of constant use. Now i know why sapphire cards are so sought after.
Jay I'm glad you did this video. I had a card that was the same thing. Turns out that it was my PSU that caused it.
The card was a Red Dragon Vega 56. Sold it broken with the GPU shortage at MSRP.
i had the same issues with my rx480 and I've tried to flash the bios like that but it didn't work 😢 when i check in device manager it detected Microsoft basic display adapter error code 31
Hey Jason
I ran into this EXACT issue with my Zotac 970, and the solution was simply that the GPU chip itself needed BGA re-soldering AKA reballing. look it up.♥
Please thumbs up so he can see it.
I am so happy you did this video. I had the same problem I refreshed the rom and I tried to reload the driver.. and this video just told me that my old Nvidia HTC 770 afterburner is just dead do to memory corruption.. so THANK YOU
I had that problem with a r9 270, it would work perfectly in desktop as long as you didn't install a driver, once you did it would get artifact all over and crashed if you put some load. It was some thermal paste that got on top of the caps and resistors next to the GPU I cleaned it and it worked just like new.
Btw the thermal paste was artic mx2 which is supposedly non conductive...
17:50 you dont actually have to have pressed down the key...you can just hold shift plus left click and as soon as you see the restart icon you are safe to let go of the pressed key....i have been using this a lot so im 1000% surtain it will work
Wish I had seen your earlier video, ended up blowing my 3090 and currenting waiting on EVGA RMA. Thanks for the research and recommendations!
Same issue with a GTX780 FTW i still have. Was messing with it when my GTX1080 was giving me issues. Thx for the vid.
Very surprised you
1) did not disconnect internet
2) did not install the driver while you were in safe mode.
Unless I'm mistaken these are basics of driver installs especially in sensitive repair cases like this...
Hey Jay, just run msconfig and un tick the box saying normal boot, then just restart until you reverse this, it will always go into safe mode then.
I’ve got an old ATI 6970 in my old old rig, rocks a Phenom X2, beast.
I have a card that works in BIOS OK, and even boots to a system OK with basic Microsoft GPU driver. But as soon as you install its driver - you get a Black Screen when the system boots to desktop. Nothing to do with games. Wonder what that's all about.