I recently pulled a broken graphics card from an HP gaming PC which had some serious issues. After uploading the video, a subscriber reached out to ask if they could try to fix it. Here's how it went... Thanks for watching :)
Except that the GPU die is completely cooked and it is gradually dying completely, finally this will be the case when the silicon has reached its maximum entropy by constantly overheating to succeed in functioning and then the GPU will refuse definitely restart...
There are people who can repair them on the component level That's expensive though, and people who do that usually advertise it and the guy would have told RGHD about his eletronics repair business
@@phen96 In the video he said the paste is fresh so I assumed the guy took the cooler off. For the sticker part, idk maybe the sticker was yummy or the actual plan of the guy was taking the sticker and we just don't expect it 🤣🤣
In that case, there are numerous cases where GPU repair technician somehow managed to get it working but also, at the same time, breaks its own clock speed scanner. So it's incorrectly reported in MSI Afterburner. I've seen someone fixing RTX 3080 and after it was fixed, everything runs normally as if it's brand new. Except for 1 thing: the Clock speed of 3080 shows at 0 MHz. Lol Imagine playing Cyberpunk 2077 with RTX 3080 on max settings, averaging 90 FPS, but clock speed reported 0 MHz. That happened
@@thesmashtvnetwork also, don't forget that RTX 3080 chip Pin layout is incompatible with any other card below it. Like RTX 3070, or 3060 or 4070 or 4060
@@IvanOoze1990 I did say it's about RTX 3080 repair, not 3090 or 4090. So it's far less likely to use fake chip for the repair since RTX 3080 already at the bottom of performance of its Compatible Pins layout. No other weaker card has same pins layout as 3080.
He probably did the oven method by cooking it in the oven for 10 minutes at 200 Celsius, or used a heat-gun on it for the soldering lanes to relink. Unfortunately this is just a temporary quick fix, because what caused the issue with the card in the first place was heat, hence why he probably adjusted the cards clock specification to prevent future high temperatures. Oven method has never been lasting repair.
Was about to post pretty much the same. I think people need to be more aware that putting a GPU in a Oven can cause harm, as there's lead and possibly traces of it landing in the food later on.
I am very intimate with that kind of artifacting you got. It was definitely a blown cap - I've had two HD 7850s that behaved the same way when their power delivery blew up.
Wonder if he had one that worked, swapped em over and sent you the one that worked back whilst he worked on the broken one? Strange he didn't message you or anything though.
Andrew, thank you for trying to fix the card and hang some success with that. And for giving us some RU-vid content from this enjoyable RU-vidr. Would be interesting if you could let us know what you did. Maybe through the RU-vidr, or in a comment, or some other way. I'm guessing you're a tech interested person who's learning to fix electronics, and probably most of us viewers here would enjoy seeing that.
He might of pulled the sticker off to look to see if there was a way to lubricate the fan possibly? but like others said, I think he took it apart and baked it in his kitchen oven.
The green screen means that the card's firmware is gone. It happens that the EPROM with the firmware becomes damaged over time. He reflashed it with lower timings, but poor thing it was time to die 😸He removed the sticker from the fan to see if there was a hole to put the oil back in, but they are made of paper with strong glue and are difficult to reattach. Maybe it can still be fixed, we need to reflow the gpu chip, ram seems ok, no artifacts. Bought rx580 £40 refurbished new on aliexpress
He's edited the BIOS and limited the clocks. He raised the voltages but that would normally have made it run a lot hotter, so he lowered the max clock to keep temperatures in check.
@RandomGamingHD Please send this card to someone who has the ability to test it thoroughly so we can know what the hell is happening with this weird card
Maybe the core and memory are faulty / damaged. So I think he could have flashed a new bios, that has limits and limits the MHz to be as low as possible, in order to make it boot and stable. Maybe previously the bios was modded or flashed and had it forced at a MHz that it couldn’t handle. Maybe?
The 580 s a great card , my granddaughter is still using my old one in the PC I built for her. Glad to see you mention a 4060 I get fed up when I see reviewers slatting it eg. Hardware Unboxed it time they realised that not everybody plays triple A games I bought a 4060Ti and it plays all my games, no problem keep the great videos coming
Faulty core or power delivery compensated for by limiting the power draw, probably through a firmware hack is my guess. That would explain the high temps at low speeds.
If he really did try to reflow it by baking then there are a few things to consider: this method doesn’t fix the problem of the solder itself which became bad and brittle. Another issue the card may suffer are bad ceramic caps which regenerate a little due to the heat but fail soon again.
Probably bios modded to lower mem clocks since that is what was failing. Potentially with some voltage tweaks as well. --- A nice trick to try if you can get a dead/dying mining card for free. --- Though I would not sell a card after bios modding it can be handy if you're putting together a zero-budget build. RX 470/480/570/580 were used a lot for mining so it's fairly common to find them in a similar state.
Try a bios re-flash maybe? It's worked out well on 2 cards I purchased. I even managed to flash one that produced no image using the Intel graphics on my I7-4790K. Might be worth a shot man. The missing sticker will probably forever remain a mystery lol.
Maybe Andrew is made up just to hide the fact that this bloke was drunk as a kite last night and had such a hunger (we're hungry when we are drunk, aren't we?) that he put his Gpu in the microwave and ate the AMD sticker afterwards. Confess!!
Seems like a reflow job via the heat gun or oven method. This sometimes works, and mostly doesn't - and when it does work, it's only temporarily. This GPU is completely cooked.
@@Yy-ig6fm Possibly, but good luck taking the GPU die off in one piece, removing the old solder balls, adding fresh solder balls, and then getting the GPU back on the board again in one piece.
I had similar low clocks when I used thermal pads that were a quarter of a mm too thick. I fixed it by using a rolling pin on the pads because .5mm lower was too low. The die just needed better mounting pressure.
what a good history, probably andrew fell bad about just doing a mess and that´s the reason of the silence, he had their best intentions and even resurrected it a bit
Bios on Polaris gpus is so easy to change, tweak, do almost everything you like with the values. Idk if there was a reflow of core and memory (lines should be fixed with reflow or total changing of chips both gpu or mem), but bios reprogramming is easy to do and make a gpu work on different values...like higher voltage maybe, power phases, mem clock OC, core downclock, fan speed...idk what he did there, I'm just saying what can be done with Polaris gpus. They are really made great for tweaking with bios values changes.
Check the bios version maybe flashed a bios for a different model but more than likely there is a damaged resistor somewhere from the possible reflow/oven attempt. Most cards are fixable for someone like northwestrepair if the core is not dead so it can still be fixed I would even have a go but i'm overseas from you. Clean the pcie strips on the card
Looks like a bad power phase, It looks like he tried to replace the mosfet but it kept shorting so he may have removed the mosfet all together, hence it is missing a power phase. It may explain the Core frequency being so low. Or he could have disabled the phase by using the I2C interface.
That he did these things isn't weird (just kind of expected when the average punter says they can "fix" your card) but the fact that he completely ghosted you, that's weird.
One explanation he might have been a repair guy and labeled all his jobs with the return address and after he repaired it passed away and his family decided to mail back all of his jobs that had addresses.
Imagine if that graphics card could talk. As soon as it got put into the envelope for the return mail, it’d be like *’Ah s***, here we go again.’* like in GTA
Removed the Radeon Sticker and got surprised temps went up!, Try removing the heatsink and/or fan from your cpu, and see what happens to the temperatures! I mean stickers are a vital part of the systems cooling, why you think they used to stick them all over the place on old laptops/prebuilts? Just to advertise it was an Intel Inside? no of course not, to keep things running cool.
Ah clever. He either baked the chip with a heat gun or the actual oven and then set a custom bios so it doesn't boost to normal frequencies, hoping to prolong the already dead chip.
Since you flipped from amd drivers to nvidia drivers for your cards, were you doing a driver cleaner before you installed the nvidia drivers? If you don't, that can cause the problems that you are seeing. Just a thought.
Not really. I used different cards in my system and it is not how it works. Each card pick up its own drivers and nothing more (unless you force it to, but it is a BAD BAD idea even if you technically can do it).
I don't know if you have access to a GTX 1650, but if you do, it could be a interesting video to test FSR 3.1 in supported games to compare if it makes them playable on low end hardware.
Not really. Frame gen doesnt work good on old cards, and if you didnt have good FPS to begin with - it will still feel like shit even if you cant see powerpoint presentation effect anymore.
@@alexturnbackthearmy1907 Sure, that's the impression we all have, but, if you don't see the new tech in use, do you really know? If everyone just assumes that's the case, and no one tests it, what is gained? Science is about testing assumptions, not just using what we know in most cases and going with that.
Yep, he baked it. It reflowed the GPU a little BUT after you tested it and let it cool down again it started again. It is either shaders in the core itself or balls are not properly touching making the gpu starve and run it in slow mode. Whenever a sticker is down = GPU was baked. Also sometimes, if it is baked more or multiple times, the PCIE gets darker shade of yellow. EDIT: And flashed lower clocked BIOS to compensate for the power.
Some huge assumptions there, none of which you can substantiate. You have no evidence, other than axmissing sticker. Removing a sticker could just as easily mean the fan was being cleaned and serviced. It might not even be the same fan.
@@another3997 the sticker has nothing below. There is no reason to take it off if you are cleaning the fan. Few times I experienced situation like this with RX 4xx and 5xx GPUs these were the reasons why. Balls and (or) partially dead part of GPU
yup, you don't take very good care of your cards. j/k idk i've had my fair share of issues with nvidia and amd. every single one of the 10s of 8800gts / 9600gts i had died out early. my gtx 970 was a pile of trash that crashed at stock clocks and could manage only lower and lower clocks over time (my RMA was only accepted on the fourth try!), and my used 6900xt didn't coil whine, it coil screeched. it was literally loud! unbearable. meanwhile i have a working geforce 2 GTS, geforce 2 MX, diamond monster 3d II (voodoo2), and radeon 9500pro
@@GraveUypo Yeah, both has their fair share of issues. But only AMD gets dunked on (and because of drivers from all things! I hated nvidia guts by the time i actually managed to install drivers correctly, meanwhile ancient HD6750 that i found in some random box of equally ancient tech worked first try and installed its drivers right away without some "cant install drivers" BS).
My own RX 570 runs extremely hot under load, shoots up to 90°C in seconds on furmark or a demanding game, i had to undervolt it a little bit and add a lot of fans around the GPU to have it run fine lmao