This remind me of the time the same thing happened to my grandpa. He died, and came back to life using the defibrillator method . He worked fine for a little bit, but died again. Unfotunately I tried the oven method afterward but it didn't seem to work too well
@@faazdoesstuff Being “older than dirt” myself, I approve of the oven method. I’m going to put it in my will that when I die, my children should try to resuscitate me by stuffing me in my oven. The result may make me more kind and tender.
couldn't just be capacitors that are giving up? they would be reasonably easy to identify and fix, if i remember right LTT did a video with a gpu with broken caps and it had problem like no texture, crash and flickering
In this case the video card heats up and "fixes itself". Dying caps could also be affected by heat. I think it's an electrical continuity issue somewhere on the card. A thorough check for any obvious and less obvious cracks in soldering joints usually does the trick in my experience. As long as the main chip doesn't have any issues, it's pretty doable to get this card fixed.
@@RandomGaminginHD Maybe get a hold of Buildzoid at Actually Hardcore Overclocking, and see if he can test it and fix it for you. He's also in the UK, so I'm sure hooking up should be easy.
Seems very similar to what my 780Ti did. I always thought the vRAM could be faulty, since all artifacts on screen are similar to a failed vRAM overclock and the vRAM is clocked very high for the time on GK110 cards. AMD used a 512 bit bus in order to achieve similar bandwidth on the 290 and 290X in order to keep the clock speeds reasonable.
From my own experience, it's 90% the memory. You could try heating the vram chips with a heatgun, or if don't have that then an oven might be worth a try just for fun. Might give it an extra hour or month.
When using an oven you'll want to heat up the oven from stone cold with the GPU in it from the very start, otherwise you may heat-shock various components and the board won't heat up uniformly and that can make things worse and cause warping + solder joints to lossen even more, or worse, crack!
@@longnamedude3947 Yup, reflowing the solder, but slowly so that stuff doesn't expand or shrink rapidly. (Which makes me guess that lots of mining cards are actually fine. They run under permanent load and go through less heating and cooling cycles and are often run below their advertised clock for better efficiency)
@@HappyBeezerStudios Just reball the chips, it is not that hard ;) Also the "mining cards" usually fail due to a dead VRM where switches degrade with time, and a subsequent short will, unless fused, just cause a burn-through so it isn't much to be done.
I had an Strix GTX 780 that had same similar issues, but since I was upgrading to a new card anyways, I tried a fix I read in an forum. Which is to reflow the card in an oven and it actually worked. I still have it running in an emulation machine I use till to this day.
The 780 partially supports DX12 feature level 11.0 so it will run some DX12 games but not all. Older cards like the GTX 480 also partially support DX12
@@RandomGaminginHD GTX 400-500-600-700-800 were all DX12 Feature Level 11_0, but that doesn't tell the full story. There are varying bits of support for hardware features. You'd still need to test on a game by game basis.
Some games require the driver support. A good example are Fermi GPUs in Red Dead Redemption II. Despite having the same feature level as Kepler, even the GTX 580 runs RDR2 about the same as a theoretically much weaker GTX 650.
This also shows the story of an older mobility radeon HD3650 laptop I have. Died, got cleaned at some point, starts out with artifacts, FIXES ITSELF, and dies when I went to make a proper windows install on the thing.
I got a Gigabyte GTX 780 OC all the way back in 2013 or so and passed it away to a relative somewhere in 2016 or 2017. They still use it to run most of their PC activity, including gaming and it seems to still hold up pretty well, despite the lack of driver support. Really good card, especially for the time.
I had the Gigabyte too until it died in mid 2020 :'( At first I was dissapointed but finally it was the right time to upgrade, prices were still normal back then
Used properly it should last 15 years or more. But many do not pay attention at temps and just use it to full potential all the time. For example there is no need to have 250 FPS in some games. Cap the FPS to something nice like 60 or 85 or 144 and the GPU will not work 100% all the time. Use fan curve in MSI Afterburner and keep the card under 65C.
My 760 did exactly the same thing! Artifacts, buggy, dead, not dead, dead again. To it's credit though, it lasted me 8 years, and right up until about 2 years ago it could run pretty much anything I threw at it with some minor adjustments. The 700 series really was very solid.
My 660 Ti is still working fine. Less memory bandwith (192 bit bus instead of 256) and more shaders (1344 instead of 1152) but otherwise pretty similar. Kepler was a pretty fine architecture.
5:05 by default, overwatch 2 uses dynamic res scaling with FSR 1.0 so that might be the reason. also it looks like you might benefit from turning down the in-game mouse sensitivity
My son had a GTX 970 with similar graphical glitches, I stripped it down, cleaned the cooler and fans and put some new thermal paste on and it worked a treat 👍🏻
I had a 780 fail somewhat similarly, where it would crash on occasion with artifacting. Reducing the memory frequency helped for a bit, but after a while it wouldn't work at all whenever the drivers were installed. It's unfortunate, because it failed earlier this year when GPU prices were quite high. On another note, it seems that the R9 200 series have aged a fair bit better than the NVIDIA 700 series. Perhaps the extra 1 GB VRAM on the R9 290 and slightly longer driver support have helped it.
That is also why the HD 7950/7970 did a bit better over time, 3 GB does better nowadays than 2 GB. Makes me wonder how well the 4 GB GTX 680/770 would do.
Cyberpunk is DX12 ONLY. It uses an additional dll file to run on Windows 7 as Win7 has limited DX12 support (it's on a per game basis), but now Win7 support has been dropped for Cyberpunk, even though that does not mean it will refuse to run on that OS
@@RandomGaminginHD Yes. Maybe in a decade I’ll get an RTX 4090 when that is cheap. I never buy high end cards when they are new because of the price but years later you can get them for a fraction of the cost. Thanks for the channel btw. I like how you do random things and not just the same things every other tech RU-vidr does.
@@wertywerrtyson5529 funny enough. The 4090 will still be a capable card 10 years down the line. Hell If my 1080 Ti is still running everything I throw at it and it’s now a 6 year old card. I’m sure a 4090 will do great.
It died for our (especially scalpers’ and miners’) sins and rose back up on the third day only to die once again after performing some miracles. Render in Peace GTX 780...
The GTX 780 and 780 TI/Titan were in a class of their own within the Kepler architecture. They had a significant performance increase over the GTX 770, which itself typically beat the GTX 680. They're still good even to this day for a budget (except the Titan) gamer as long as the games don't require DX12 native.
Late in the gpus life there was still a noticeable jump from the 770 to 780ti but due to kelper being kelper it was alot less than I hoped after experiencing a r9 290. I beleve it allowed a 20% res increse at the same fps in bordlands 3 Both cards where clocked at 1200 ish mhz core.
I had a very similar problem with a gtx 670 2gb. Lines appeared on the screen and the driver did not install. I put the card in the oven for 5 minutes (it was heated to 200°C) it started working in the second attempt in the oven :P
I had a GTX 770 which died in a very similar fashion back in 2016. Perhaps it was just shoddy VRAM or VRM cooling. It was pretty sad to see regardless. On a side note, I have a GTX 465 which still runs with no issues so perhaps heat is not the issue but the 600-700 series design was flawed, as I have seen a lot of laptops with 600-700 series GPU's die 3-4 years into their life span.
My 660 Ti is still running perfectly and overclocked since I bought it in early 2013, so it must be something with the higher cards or the manufacturer.
If some of the solder balls break, you could reball it. Or well, the old oven trick. That basically reflows the solder on the cheap. Not a guarantee, but if the card is already dead, you can't make it worse. Revived my old 8800 GT twice that way. That the card worked again after being pulled out might be a good sign that after cooling and moving the PCB around the contacts might gotten better. I had a card like that, it would run perfectly fine without driver, but with driver installed it would display artifacts even at the lightest load.
Older cards have this issue where like in a DOS os everything is perfect but as soon as you use windows 10 it instantly starts dying. My friend who fixes graphics card, replaces vram and other chips he said he noticed this Aswell ive noticed if you use a GPU for a while and then put in in a box on a shelf for a while, after that whike it will brick itself, then again leave it and it fixes itshelf
It might be worth your while to undervolt and down clock the card. Perhaps flashing a lower OC'd BIOS would help it work better. Considering its age, it may also be prudent to check the capacitors and various solder joints to see if they need any fixing as well.
You should really try the oven/heat gun trick, i recently got a GTX 780 aswell for around 13-14$ and it had some artifacts aswell and wouldn't install drivers, i was lucky to fix it with a heat gun and it works fine now. Had the same problem with a HD7950 and the heat gun fixed that aswell, i know its more of a temporary fix but still. Love ur videos btw :)
Any tutorial on using the heat gun. I have a Radeon HD 7870 with artifacts and some other issues and would like to use the heat gun, just for fun and hopefully sucking a bit more of life out of it xd I bought used 4 months ago and it started failing a few weeks ago xd
@@LucasLopez-kk5ny Yes TechYesCity has a great video on it that i followed, it's called "How to Use a Heatgun to fix (MANY) dead Graphics Cards..." but basically you just strip the graphics card down to the pcb(remember thermal pads & thermal paste) and put your heat gun on the lowest setting, then just have it around the GPU with a circular motion for 5 minutes. Hope it makes sense :)
Same thing happened with my GTX 760TI except with no artefacts, it would randomly turn the screen a random solid colour when it reached max boost clock. I "fixed" it by underclocking the Core and Memory Clocks by -100MHz and the card works flawlessly now. Probably just degradation since it has been used for heavy gaming since 2013.
You can block the fan, turn on PC for few minutes or more (even when screen is not working) an then reboot. I was able to temporarily turn on some bricked GPU by this way.
hey! from new zealand, currently bidding on a blower style EVGA GTX 760 and a Nvidia Quadro 4000, auction closes in 2 days, so far its up to a dollar! $15 shipping. wish me luck!
be careful - there are scammers on ebay selling faulty parts as working. in the past month i have bought 2 gtx 980ti's - both broken. one has been relisted. this is uk ebay fyi. these accounts had 100% pos feedback. the cards were tested in 2 systems and with 2 different working psu's. 2 separate accounts ive noticed were the same person with identical listing forms.
What's likely causing the issue is failing BGA solder points on the memory modules; as the card heats and thermal expansion occurs, it's making better or completed contact with the pads, thus why everything cleared up when left running for a while. I'd bet real money that reflowing would fix the problem (though I certainly couldn't be bothered for some old Kepler card).
Probably a bad soldier joint. Over time the joints can weaken because of heat. the artifacts is likely memory related. If you pull it apart start there and look for brown or black smears on soldiered joints
I used to have a GTX 1060 back in 2019 that was given to me cause it was a faulty card. The HDMI didn't work, but the DVI did. That card was a frickin demon, sometimes it worked but didn't load the drivers, other times it worked flawlessly, and other times didn't even display image. I think it was a faulty memory, because with low vram demanding games like csgo didn't had any problems (with fps capped)
Anyone else suspect the dog of doing something..... Suspicious that he takes his good boy for a walk, and it starts working, then he takes him for another one and it stops.... Mighty suspicious.....
A bit of a coincidence. I decided to try my PNY GTX 750 ti again and update the drivers, it turned off again in the middle of the installation. Well, it lasted a good 7 years, it's time for an upgrade.
I was playing Lord of the Rings:Online today, 4K full ultra and wow what a gawjus looking game it is. It was bloody gawjus under 1080p but this Philips 4K monitor just rips my old monitor to pieces. You should give LotR:O a try, I was a alpha tester for that game and had to drop World of Warcraft for a time to devote many many long hours into aiding debugging, the Elven startup area is the original test area and the door you can't get in was for us testers as a "GM Island" is at the other end of the pathway.
could be the unsoldered VRAM, most probably still good but the solder has cracked. using a blower type solder to heat the VRAM can fix it, you might need some flux for job like this.
@@RandomGaminginHD I looked into it a bit and messed around with the card after the code 43 but to no avail. Interestingly the consensus online seems it might have been a bad Microsoft driver pushed on windows update, and in more cases than I could count the cards died in game while playing warzone.
This is called terminal lucidity. Very strange phenomenon that happens with humans, animals, machines ... Being or device that is dying suddenly gets better just hours before final collapse and death.
I remember buying 2 Zotac Amp Extreme edition Gtx 780s for $120 in 2018 and had a great time never did sli just sold the second card and used one for awhile fond memories
I'm still using an ASUS CUII GTX 780 ti on pretty much a daily basis! I just changed the thermal paste a few days ago once again after two years of constant use. It's still tickin' along fine! 👍👍
I remember my 750ti had a wonderful bug with windows 10, where installing the drivers would result in either a magenta screen, or black screen, when it came time for the installer to switch to the new drivers after install. It would essentially halt the install, and a reboot fixed it. The only PROPER fix i could find, was reinstalling Windows 7, installing the drivers, then upgrading to Win10. No idea if this is reproducible in current year.
I had a similar weird issue with an R9 270, green artifacts on screen, tried to "reflow" the solder, nothing happened after that. Then out of nowhere while searching for a new GPU, it just suddenly worked. I did every single stress test I could, repasted/cleaned it again and it worked great, I was able to play games no problem for almost a week, but then out of nowhere the artifacts suddenly came back and now it's dead again.
I have the same problem with my old rx470, works fine after installing the driver, but sometimes the driver uninstalls its self, and yeah the system goes crashes to black screen and following with a disconnected device sound, I'm giving up with that card, and bought Rx 6600xt and I still keep the 470 in my room (that keeps hoping I found the problem)
beast card for the day, I started with a GT 640 4gb and did me well then a R7 370 solely for gta V and csgo paired with i5 3470 and 4 then eventually 8 gigs ram. The closest I got to a 780 was a GTX 670 which I abused in warzone and forgot to replace thermal paste and only lasted me about 2 month =[
Speaking of zombies though, my hard drive was 0% in health due to some bad sectors and overheating. Lo and behold a day or two later its bsck to 95% no issues with it whatsoever.
Dunno, still doesn't sound good. I mean, I have one that decided to kill itself, after tons of overwrites and surface checks it looks fine, but still causes issues.
when the driver fails to load is often a sign of core damage maybe when you first installed it into the system and booted it from cold you kinda did thermally shock the core, temporairly fixing the damage in the core enough for doing the game testing thats why when you booted again it kinda held on but after you stopped the testing seems like the core degraded enough to crash the drivers and send it back to ded this card might be a good candidate for the heat gun, it might revive itself again, but once the core is damage like that is just making it limp along for a while longer, it will just die again sooner or later
I still have my old GTX 660 Ti here (also a Kepler card) and that card still runs perfectly fin and can play surprisingly many games for it's age. And Kepler cards are easy to bring onto 1202 MHz without much issue. If you get the card fixed might be worth looking into. When in doubt a modded BIOS will do the trick.
Gtx 770 2Gb sits as a paperweight on a shelf atm lol. Comparable to 1050's from what I remember, just with obvious vram limits. Hence it's a paperweight lol. Was good looking these up during the GPU wasteland, but some good cheap decent stuff available if know where to look
@@NaldzHobbySide Owch lol I was a bit stuck, but I found an rx580 8gb on eBay for £42 😳 Thrown it in an old 4690k system and it's a bedroom TV pc/gaming pc now. Some good deals are popping up as the prices are coming down. Kinda why the 770 just collects dust atm lol
I got a GTX 770 for 50euro recently, sort of just for the hell of it. It has worked flawlessly, apart from the strange issue of it making my main rig, and specifically only it, not POST, if and only if I have it connected to my 1440p monitor. 1080p monitor works fine. Just not the 1440p one
I had a 2080 Super that basically did the same thing. Just for it replaced by MSI with a 3080 Suprim X so I can't complain! The 2080 Super did me a favor dying during warranty lol.
@@RandomGaminginHD heck yeah, I waited a bit to RMA it because of shortages and used a 6700XT. Did the RMA when the overstock of 3000 cards came about and they replaced it with a free upgrade without thinking twice about it lol.
My 780Ti after some years started to give a lot of artifacts and sometimes crash the system. I guess 7xx generation has A LOT of VRAM issues. My older 4870x2 still in working condition (i usually give my old cards to my friends when i get new one on condition what when it dies or they get better card they return it to me for a collection so every card works for at least 5 years total) and of all the cards i ever had only 780Ti by Gigabyte is dead.
Try the 472.98 driver instead, much more stable for my PNY 780, I dare not update the driver since I baked the card (200C, 10 min), but it has been working fine for 6 months, I only get som random blackscreens when switching between 1 & 2 monitors. I'm not 100% it was the baking or the driver that fixed it though.
Try the oven trick, bake 7 min on 200 Celsius with chip facing up, than let it sloooowly cool dovn (just open oven door) and then reasemble and play :)
I still have laptop with GTX 980m 8gb graphics card and its performance matches GTX 780, i can still play many modern games on my laptop becoz it has 8 gb vram & still getting drivers support. Nvidia made a huge mistake by making Gtx 780 only 3 gb. If vram and driver support was still there, gtx 780 would still be performing well in modern games.
you could look up nvidia MATS (basically nvidia in-house tool for gpu problem testing and stuff) to try narrow down the issue, it can help you know exactly which vram chip is causing issues or if its a vram chip issue at all, i had similiar problems with gtx 1060 6gb on a laptop and one of the vram was bad and i had it replaced by someone since i dont solder (heatgun or oven treatments arent a reliable repair more like temporary or if u dont care if it has chance to fail again later solution)
Iv had a rx5500xt with strange problems if you install the driver's it would crash so flashed the bios driver's will install then it would over heat so re paste got better took the back plate of even cooler but u couldn't play games at all
My uncle gave me his entire pc years ago because he couldn't get it working. It had a gtx 780 3gb that was burned out, being the cause of why it wouldn't boot up. One of the r230 transistor chips looked super brown compared to the others. I'm assuming it was burned. Its a shame because that card would still outperform my 1050 ti by a small margin according to benchmarks.
Woah we have the same thought Even though I can't fix it YET(by yet i mean i will self study on how to fix gpu and buy my stuff) I would still buy it Especially FE cards Because even new, my country doesn't have those cards. But i don't plan on buying them lol Currently own a 750ti used when bought
