It was new information to me that a core can be dead without a short circuit. I'm glad you're around to help people even if the results are sometimes like this.
Anything electronic can be dead with a short circuit or no circuit at all except maybe an old fashioned light bulb that dies from no longer having a circuit and cannot short without being destroyed in the process.
There are many more failure modes for (complex integrated) circuits besides it having a short circuit. Calling this core 'dead' isn't actually true by the way, faulty maybe, but not dead.
@@hapskie as a gaming device it is dead. You might argue that it will function with the 1st GB of VRAM or something. Or like just display device, something like GT 710... Maybe for CUDA workloads? It is possible that you would find some use for it still. But is there a point doing it?
Don’t know if this comment will go thru but here goes. I notice when work is done with hot air practically noting is shielded. The pieces of a coke can over the memory chips are a good start but doesn’t prevent the hot air going under the chips. When I used to fix laptop video chips I always had a custom shield for the entire motherboard made from folded aluminum foil and aluminum tape around the cores. After many repairs I had quite a bit of shielding templates that prevented unwanted damage to other components and only focused the heat on the chip I had to work on. Would suggest getting into this practice as it’s not hard to do for GPU’s as the core size is always going to be the same, adjustments will need to be made for the smaller 70 series and below cores well as memory chips. Can start with one and adjust as needed but always using aluminum tape around the components as close as possible. This will prevent any errant hot air from damaging other components. Be glad to show pictures of the collection I have on the Discord or DM if you’d like.
Watch previous videos. He shields components with thin shields of cut coke cans or kapton tape or aluminum foil when needed. It's not always needed. His air control is very good and he is not likely to burn anything thanks to experience.
@@btwbrand I already said that about the coke cans. Doesn’t matter all of the board should be shielded except the component being worked on. Takes 10 minutes to make something up and saves hours of aggravation when something gets damaged trying to chase down the new problem. It takes very little deviation of heat say attaching a new memory chip or core to melt the solder of the components nearby especially the outer BGA solder balls. No reason to make more work for oneself when it can easily be avoided, remember the laptop case he melted cause he didn’t think the motherboard needed to be removed? 5 minutes of removing a few screws would have prevented that debacle and that’s what happens with shortcuts.
While a dead core leaves no one better, your diligence and fault finding is very superior. Thank you for the videos reminding us excellence is still a product in my country. In fact if ya'll ran for office, I would back you 100%. Wherever you are working, these people need to double your pay salary.
Thanks for the video! Just wonder, have you tried under-volting / over-volting the core on this one? Based on your other videos this could help, right?
I made a comment before this one and deleted it. I had jumped into a conclusion. But I still think the power delivery could be at fault even after a second and third thought. There could be a component that got degraded. I know how crazy this sounds, but considering how good you are at reballing chips this would be a great opportunity to learn something by taking that core off its board and putting it on anoher PCB. At least that's what I would do, so that I could test the core with a _known good_ power delivery circuit. If I'm wrong, I kind of waste a couple of hours of a learning session, but if I'm right, I could make thousands from buying "dead" 4090s.
I looked at the power delivery and everything looked fine according to sensors. I dont have another 4090 board to try the core on so i have to make an educated guess.
hey just wanted too say thank you for being around i literally would just push through life day after day and i stumbled across electronics repair and i found something i want to do in life a year and half later empty wallet from all this equipment i found your videos and just want too say how easy you make this look i second guess all my moves and what tools too get scared they will be terrible and not worth it definetly joining discord i would even pay money per question too ask you for help again thank you brother your awesome!!!
Just practice on whatever you can find, even an old microwave has stuff you can remove and reinstall, $5 fm radios even have the nice teency parts to play with, I can't even start to tell you how many things I destroyed before I learned what to use and do. 😂
I have information. I looked into this case and made some interesting deductions. The first of which is the fact that the number "1" in the BRSFT column is obviously the norm, while number "2" can be ruled as the odd ones out. Also, the numbers "2" directly correlate with the low numbers ("8" and even "0") on the next column to the right, name "INTPL". Then you can also notice that those low numbers on INTPL show a tendency to occur with pins CA4 and CA5 on the column that goes to the left of the numbers 2.... Now here comes the interesting part; CA4 and CA5 are PRECISELY the pins at the center of the memory chips according to the GDDR6X ballout data. In other words, 4 chips show a low number of "INTPL" with an amazing coincidence in the fact that all 4 chips have troublesome numbers exactly at the center of their ballout. I think this is proof enough to show that the problem is NOT the core, but rather your heat delivery on the memory chips. Your heat is not properly melting the balls at the center.
@@odjsjaks Then explain why the pins at the center are coincidentally failing in 4 different chips. Not only CA4 and CA5, but also CABI which is ALSOOOOO at the center. :)
@@odjsjaks That would be a good excuse if we DIDN'T have documentation, but we do. Just look up GDDR6x ballout and you will get the documents I got which you did not :)
It's ok, you are thinking right, Kris found one just like that, also a 4090, with crazy shorts and things that did not make sense, something about that model I guess. Good video
The symthoms (crash, artifacts) looks similar to an over "overclocked" GPU (or memmory timming), maybe try to underclock the GPU boost and memmory settings in After Burner.
yeah but this is a FE, not sure why it would be overclocked, AT ALL. usually FE cards are clocked the lowest and undervolted, which is why they always use a low amount of fans. they have the best efficiency.
@@ruxandy my suggestion is for debugging purpose not for a fix. If all is ok with minimal speeds, maybe is good to look at memmory speed specs (i know that chines, rebrands chips with lower speed) or boost voltage control. Definetly on the video the card crash when it reach 2700+Mhz and above...
14:07. 🤔. You will send me all of your FIXED GPU's (in complete working order of course 😮) so I may sell them on the ebay for triple the MSRP of course 😮.
My advice is to not touch chemicals with your skin under no circumstance. Use a brush, do not use your finger. You're poisoning yourself bit by bit by touching it and allowing it to be absorbed through the skin. It is in tiny quantities but if you absorb 1/1000 of a gram for one memory, doing 1000 memory chips would give you a 1 full gram of absorbed poison. You need to re-think your work protocols. Your health is by far the most important thing.
@Dandan ... It seems trivial having a bit of silicone glue or thermal paste or cleaner get on you hands. I know a few older guys with long term skin issues on their fingers from contact with paints, thinners, silicone and PU sealants etc etc. Tiny amounts, over 30 years. One guy had to wear white cotton gloves laced with lotion for a long time to grow back his subdermal skin layers. It's no joke. Silicone products are the worst. Flux probably isn't very bad but not worth finding out in your retirement!
I wonder if the dead core's flaw is only in the graphic rasterizer part of the chip? If so, it might still be usable as a compute card for mining or AI or raytracing accelerator. Of course you'd need tests for those applications first.
I do a small percentage of the current cost of the product in the market (about 10%) + labor + parts + shipping, that's much better than doing all for the same price, because somebody can benefit from this using your time and knowledge.
Hi! Can I ask you where I can get one of those charts. I not a GPU fixer or anything I just want to have an idea about how to get one of those about other computer components. But I would love to know how you get those charts. So if it's okay please share.
I've had that same random and 2's with NVMT never meant anything imo lol.. When I got my 3090 a year ago it gave so called errors with 2's on some banks so I replaced with new modules didn't change a thing and it been working flawlessly for a year.. I'm in the same boat idk what it means but in my experience its nothing to worry about..
So the idea of this repair was like this? The memory chips appeared to pass but it was possible they have sporadic issues or under different loads had issues. 2) There could have been a connection problem of one of the pins of the core and the memory. But what was concluded is the core had developed physical damage so that was that?
I didn't see any physical damage anywhere. The memory errors were misleading. Information on the internet about the use of NVMT is vague at best and I actually never really bothered to investigate anymore than I already have prior to this point. But when I did, I started to doubt my self and I became suspicious about 4 memory chips which I shouldn't have, replacement of which had no change whatsoever.
my rtx 2070 just decided to up and quit working one day. left it alone for like a month then got a hair up my rear. stuck it in the oven for like four hours at 320 degrees and viola it functions again. I know Improper as hell but not gonna knock it. have not had a issue yet. and surprisingly it achieves higher clock speeds now.
@@tonycstech interesting. maybe a in-game test could work. I tried to undervolt my laptop's 3060 in Phasmophobia (got 2 laser projectors to increase the stress on core), purple dot artifacts had shown, after setting the curve on default it returned back to normal instantly.
Might be worth the money to ask a company to make a mold with an lga socket for memory chips that has the pins going into the sides where the value sheets you have are printed onto the mold.
I guess replacing the memory chip wouldn't have done anything either that shorted previously, i would've assumed it was memory artifacting looking like this but i guess this is not always the case.
How do you even have superb magnifying vision without using microscope, i barely can see large object at that distance let alone micro size component. dang amazing!
Wait so u heated a chip which shorted the chip but u probably tested after that u maybe killed core by shorting that chip while Replacing the one next to it ?
My laptop repair and gpu shorted, one phase was remove from 7, can gpu laptop runing without one phase remove, but in bios and gpu detected but not show on task manager, by the way my laptop gpu is rtx2060
is there any way to know WHICH core on that die is bad? could you leave some balls off and disable power to that core? run it in a slower configuration? do they use the same die on a 4080 as a 4090?
what software is it that you can boot into and get the pass or fail thing. I saw it before on a tour of evga's repair facility but figured it was just something they made
no shorts try different hdmi/dp for mads next time, show it nvida cards does not does sleep off the monitor when two cables are connected, even if a second cable with off monitor is inserted- maybe, maybe this does nothing. keep up your good youtube videos.
you should buy those Graphite Thermal Pad instead of paste(sice they can be reused over and over again) Man i follow you since you had like 2k sub and i still can't stop being amazed at how good you are and what you can do
I think we would all love to know something. What is YOUR favorite GPU brand? I imagine that answer is based on quality which we could all benefit from.
I am serious when I say STOP overclocking graphics cards. Rather underclock them. I underclocked my 4080. I lost about 3 to 5fps but my wattage is about 90w to 100w, games fully playable, also temperature is about 60 degrees. If I dont underclock, card easily pulls 180w to 250w, and runs close to 80 degrees. Also, use the support on the back of your card. If you dont have a support, let your case rather lay flat, the stress on the board with this heavy cards causes problems with the PCI Express slot, and card cracking. If the card pulls on the slot, it can short itself per accident, and you lose your core or memory.
Well at the end of the day the gpu core is expensive for itself not the pcb or any other components no wonder why people buying the 4090 and not willing to pay for the repair to just replace the chip. This where I draw the line in repairs
It was new information to me that a core can be dead without a short circuit. I'm glad you're around to help people even if the results are sometimes like this.
Anything electronic can be dead with a short circuit or no circuit at all except maybe an old fashioned light bulb that dies from no longer having a circuit and cannot short without being destroyed in the process.
There are many more failure modes for (complex integrated) circuits besides it having a short circuit. Calling this core 'dead' isn't actually true by the way, faulty maybe, but not dead.
@@hapskie The core can't be repaired. If it doesn't function to spec it's functionally dead.
@@btwbrand
Only put some atoms in place.
@@hapskie as a gaming device it is dead. You might argue that it will function with the 1st GB of VRAM or something. Or like just display device, something like GT 710... Maybe for CUDA workloads? It is possible that you would find some use for it still. But is there a point doing it?
Don’t know if this comment will go thru but here goes.
I notice when work is done with hot air practically noting is shielded. The pieces of a coke can over the memory chips are a good start but doesn’t prevent the hot air going under the chips.
When I used to fix laptop video chips I always had a custom shield for the entire motherboard made from folded aluminum foil and aluminum tape around the cores. After many repairs I had quite a bit of shielding templates that prevented unwanted damage to other components and only focused the heat on the chip I had to work on.
Would suggest getting into this practice as it’s not hard to do for GPU’s as the core size is always going to be the same, adjustments will need to be made for the smaller 70 series and below cores well as memory chips.
Can start with one and adjust as needed but always using aluminum tape around the components as close as possible.
This will prevent any errant hot air from damaging other components.
Be glad to show pictures of the collection I have on the Discord or DM if you’d like.
Watch previous videos. He shields components with thin shields of cut coke cans or kapton tape or aluminum foil when needed. It's not always needed. His air control is very good and he is not likely to burn anything thanks to experience.
@@btwbrand I already said that about the coke cans. Doesn’t matter all of the board should be shielded except the component being worked on.
Takes 10 minutes to make something up and saves hours of aggravation when something gets damaged trying to chase down the new problem.
It takes very little deviation of heat say attaching a new memory chip or core to melt the solder of the components nearby especially the outer BGA solder balls.
No reason to make more work for oneself when it can easily be avoided, remember the laptop case he melted cause he didn’t think the motherboard needed to be removed? 5 minutes of removing a few screws would have prevented that debacle and that’s what happens with shortcuts.
Impeccable knowledge and skill.
Lol I wish 🤣
@@northwestrepairdon’t be so modest haha.
@@northwestrepair Have you fired up the oscilloscope and checked the phases during initial load.. Might have one acting up you never know..
@@BanksRacing11 Exactly my suspicion too!
@@northwestrepairyou wish? You are so f****** knowledgeable way more than 99.999% of people
and now everyone wants this 4090 for you're signature .
Tony, you are a LEGEND! No one can beat your GPU repair prices!!!
19:12 oh trust me, we can see the giant fly
And i for one welcome our new flying overlords
Sir, you are one of a kind. No one on youtube can do what you do, or even atrempt it. Well done!!!!!
So satisfying to see repairs on this channel.
I can't wait for the next 4090 video where you open it and see your writing :D
That would be epic!
While a dead core leaves no one better, your diligence and fault finding is very superior. Thank you for the videos reminding us excellence is still a product in my country. In fact if ya'll ran for office, I would back you 100%. Wherever you are working, these people need to double your pay salary.
i like your videos, youre a good human with a good sense of humor ^^
that is great you have this much work people trust you I wish I could build up a cliente as you have congrats!
make videos of your repairs
Damn 4090s just came out not that long ago. Why isnt this RMA'ed? Its most def covered under warranty still.
19:12 Fly graphics test passed, slight blurring though perhaps change DLSS to quality setting.
Thanks for the video! Just wonder, have you tried under-volting / over-volting the core on this one? Based on your other videos this could help, right?
I know the person who sent it in got it for free. But I'm confused as to why the original owner didn't try to claim warranty.
I made a comment before this one and deleted it. I had jumped into a conclusion. But I still think the power delivery could be at fault even after a second and third thought. There could be a component that got degraded.
I know how crazy this sounds, but considering how good you are at reballing chips this would be a great opportunity to learn something by taking that core off its board and putting it on anoher PCB. At least that's what I would do, so that I could test the core with a _known good_ power delivery circuit. If I'm wrong, I kind of waste a couple of hours of a learning session, but if I'm right, I could make thousands from buying "dead" 4090s.
I looked at the power delivery and everything looked fine according to sensors.
I dont have another 4090 board to try the core on so i have to make an educated guess.
your videos are so entertaining - well edited, too. easy to follow and i feel like i’m learning. cheers
I feel like people love pushing their cards past the limit when it's not needed. Overclocking isn't needed on a card that already has great enough fps
hey just wanted too say thank you for being around i literally would just push through life day after day and i stumbled across electronics repair and i found something i want to do in life a year and half later empty wallet from all this equipment i found your videos and just want too say how easy you make this look
i second guess all my moves and what tools too get scared they will be terrible and not worth it definetly joining discord i would even pay money per question too ask you for help
again thank you brother your awesome!!!
Just practice on whatever you can find, even an old microwave has stuff you can remove and reinstall, $5 fm radios even have the nice teency parts to play with, I can't even start to tell you how many things I destroyed before I learned what to use and do. 😂
Best opening line ever 'its balls day today". That it is
Love the frequent uploads. Keep it up
I have information.
I looked into this case and made some interesting deductions. The first of which is the fact that the number "1" in the BRSFT column is obviously the norm, while number "2" can be ruled as the odd ones out. Also, the numbers "2" directly correlate with the low numbers ("8" and even "0") on the next column to the right, name "INTPL". Then you can also notice that those low numbers on INTPL show a tendency to occur with pins CA4 and CA5 on the column that goes to the left of the numbers 2.... Now here comes the interesting part; CA4 and CA5 are PRECISELY the pins at the center of the memory chips according to the GDDR6X ballout data.
In other words, 4 chips show a low number of "INTPL" with an amazing coincidence in the fact that all 4 chips have troublesome numbers exactly at the center of their ballout. I think this is proof enough to show that the problem is NOT the core, but rather your heat delivery on the memory chips. Your heat is not properly melting the balls at the center.
I'd run this info past someone else that repairs gpu's
@@odjsjaks Then explain why the pins at the center are coincidentally failing in 4 different chips. Not only CA4 and CA5, but also CABI which is ALSOOOOO at the center. :)
@@odjsjaks That would be a good excuse if we DIDN'T have documentation, but we do. Just look up GDDR6x ballout and you will get the documents I got which you did not :)
because the issue is not the ram but the core it's self.@@EDOD_EseDelOtroDia
It's ok, you are thinking right, Kris found one just like that, also a 4090, with crazy shorts and things that did not make sense, something about that model I guess.
Good video
The symthoms (crash, artifacts) looks similar to an over "overclocked" GPU (or memmory timming), maybe try to underclock the GPU boost and memmory settings in After Burner.
while that might work, it's not a 'repair', it's just a band-aid (and a temporary one at that).
yeah but this is a FE, not sure why it would be overclocked, AT ALL. usually FE cards are clocked the lowest and undervolted, which is why they always use a low amount of fans. they have the best efficiency.
@@ruxandy my suggestion is for debugging purpose not for a fix. If all is ok with minimal speeds, maybe is good to look at memmory speed specs (i know that chines, rebrands chips with lower speed) or boost voltage control. Definetly on the video the card crash when it reach 2700+Mhz and above...
@@users-bibi50 Ha yes, i have not looked this video yet but i see what you mean :)
such a shame, but the dedication is out of this world.
Very nice detective work there.....(experience)
Love your vids. Love what you do!
"Hello North West Repair!" there I done it. dang feel so good! ❤
NWR, that surgeon that resurrects tech necrons 😮 its Alive!!
14:07. 🤔. You will send me all of your FIXED GPU's (in complete working order of course 😮) so I may sell them on the ebay for triple the MSRP of course 😮.
where do I find this nice color chart around 8:00 ? thanks!!!
20:45 Did you wrote "Deab" core? 🤔
My advice is to not touch chemicals with your skin under no circumstance. Use a brush, do not use your finger. You're poisoning yourself bit by bit by touching it and allowing it to be absorbed through the skin. It is in tiny quantities but if you absorb 1/1000 of a gram for one memory, doing 1000 memory chips would give you a 1 full gram of absorbed poison. You need to re-think your work protocols. Your health is by far the most important thing.
Health protocol? You clearly don't know this guy.
🤓
@Dandan ... It seems trivial having a bit of silicone glue or thermal paste or cleaner get on you hands. I know a few older guys with long term skin issues on their fingers from contact with paints, thinners, silicone and PU sealants etc etc. Tiny amounts, over 30 years. One guy had to wear white cotton gloves laced with lotion for a long time to grow back his subdermal skin layers. It's no joke. Silicone products are the worst. Flux probably isn't very bad but not worth finding out in your retirement!
This also works the other way around. Skin chemicals are corrosive for electrical contacts.
everything is chemicals
14:42 Respect!
Dead RTX4090 core is the saddest most heartbreaking thing in the world.
its a ballsday today -northwestrepair
man you scard the crap out of me with that fly XD
We could see the fly lol -- that was interesting
Scope the power phases, that error log looks odd.
I wonder if the dead core's flaw is only in the graphic rasterizer part of the chip? If so, it might still be usable as a compute card for mining or AI or raytracing accelerator. Of course you'd need tests for those applications first.
Mining is absolutely worthless in every sense.. maybe money but please people stop wasting energy for it.
I do a small percentage of the current cost of the product in the market (about 10%) + labor + parts + shipping, that's much better than doing all for the same price, because somebody can benefit from this using your time and knowledge.
I was doing flat cost for over a year. Now it's little more like a percentage too.
Low enough to keep repair as the #1 option.
Awesome video doc! 😂
Hi! Can I ask you where I can get one of those charts. I not a GPU fixer or anything I just want to have an idea about how to get one of those about other computer components. But I would love to know how you get those charts. So if it's okay please share.
I've had that same random and 2's with NVMT never meant anything imo lol.. When I got my 3090 a year ago it gave so called errors with 2's on some banks so I replaced with new modules didn't change a thing and it been working flawlessly for a year.. I'm in the same boat idk what it means but in my experience its nothing to worry about..
So the idea of this repair was like this? The memory chips appeared to pass but it was possible they have sporadic issues or under different loads had issues. 2) There could have been a connection problem of one of the pins of the core and the memory. But what was concluded is the core had developed physical damage so that was that?
I didn't see any physical damage anywhere.
The memory errors were misleading.
Information on the internet about the use of NVMT is vague at best and I actually never really bothered to investigate anymore than I already have prior to this point.
But when I did, I started to doubt my self and I became suspicious about 4 memory chips which I shouldn't have, replacement of which had no change whatsoever.
Always great vids . Be wonderful if you would do a solder vid to help learners like me have a good day keep up the awesome work my friend .
my rtx 2070 just decided to up and quit working one day. left it alone for like a month then got a hair up my rear. stuck it in the oven for like four hours at 320 degrees and viola it functions again. I know Improper as hell but not gonna knock it. have not had a issue yet. and surprisingly it achieves higher clock speeds now.
Does your 2070 still work today
@@mistabrown830 Sadly no burnt the vrm phases out. *i was up to no good and pushing it too hard.* got a rtx 3060 now.*
Undervolting - underclocking wouldn't help? I mean is there anyway for that core to work?
why don't you lower core frequences to test it like in a previous video? it would cut mem chips reball part.
I did. It changed nothing.
@@tonycstech interesting. maybe a in-game test could work. I tried to undervolt my laptop's 3060 in Phasmophobia (got 2 laser projectors to increase the stress on core), purple dot artifacts had shown, after setting the curve on default it returned back to normal instantly.
You have it your all
Can u replace the core from another cpu or will it not work just curious Cuse I never see u replace them ❤❤ love the vids
4090 cards was released less than 1 years ago. No warranty?
Might be worth the money to ask a company to make a mold with an lga socket for memory chips that has the pins going into the sides where the value sheets you have are printed onto the mold.
They all work fine until they don't, and then you come in:-)
Awesome sauce video… 😊
I guess replacing the memory chip wouldn't have done anything either that shorted previously, i would've assumed it was memory artifacting looking like this but i guess this is not always the case.
Great video, but what I don't understand is that these cards must still be in warranty, so why do they end up in your shop ?
Warranty kick backs and 2nd hand owners.
All 12v are 12.5 to 12.7 under load , my 4090 is 11.9 to 12.2 underload , idle 12.3 to 12.4
Nothing to complain. You did your best. But I wish you got more spare parts.
Moisture or there suddenly happened bi metal effect as old thermostats bi metal does bend when get hot.
ouch! this one hurt..
How do you even have superb magnifying vision without using microscope, i barely can see large object at that distance let alone micro size component. dang amazing!
was it a costumers repair or one you bought to repair and sell?
what hot air station are u using??
Is the wax dead like that ring around the chip is it blackened
4090 is revatively new card, is there no warranty?
How about a new firmware? Or did I miss that? May have, I kinda skimmed through
would love if you could bring your voice at a bit higher volume...
Wait so u heated a chip which shorted the chip but u probably tested after that u maybe killed core by shorting that chip while Replacing the one next to it ?
My laptop repair and gpu shorted, one phase was remove from 7, can gpu laptop runing without one phase remove, but in bios and gpu detected but not show on task manager, by the way my laptop gpu is rtx2060
is there any way to know WHICH core on that die is bad? could you leave some balls off and disable power to that core? run it in a slower configuration? do they use the same die on a 4080 as a 4090?
cant imagine 2000€ card without warranty and nowhere to rma.... never ever any expensive gpu without warranty ... golden rule for life... ☝☝☝
what happened to the nvidia warranty?
No idea. Depending on where they get it, it may come void of all warranty
those balls seem so tedious to fondle around.
What can kill a core like that?
what software is it that you can boot into and get the pass or fail thing. I saw it before on a tour of evga's repair facility but figured it was just something they made
NVMT
This card is still new, does't it have warrranty.
no shorts
try different hdmi/dp for mads next time, show it
nvida cards does not does sleep off the monitor when two cables are connected, even if a second cable with off monitor is inserted-
maybe, maybe this does nothing.
keep up your good youtube videos.
What i dont get is , if the card is so new why wasnt it RMA"d , or was it 2nd hand?
Another one to the kingdom of dead computers.
anyone with 4000 series should still have warranty, so why you need to fix it if they can just send it to the company you bought it from
you should buy those Graphite Thermal Pad instead of paste(sice they can be reused over and over again) Man i follow you since you had like 2k sub and i still can't stop being amazed at how good you are and what you can do
What is flux exactly and is it harmful to skin? I mean, you rub those chips every day with this stuff.
I think we would all love to know something. What is YOUR favorite GPU brand? I imagine that answer is based on quality which we could all benefit from.
3080, 3090, 6800XT, & 6900XT have the highest all time failure rates.
Anyone know what it would cost to replace the core? I.e. the cost of a new chip without labor?
4090 chip will cost more that a broken 4090 on eBay which can have a working chip 50/50 chance.
Right hard one
Yes I saw a Deab core 😄
Good effort on your end but a dead core is too expensive to fix.
Why don't you charge more if you have too many orders? Just let the market sort it out.
You sound exhausted on this one. Too many balls? Rest my son ... rest. But keep them vids coming! (😉🖖)
Thank you!
Why isn't the 4090 under warranty?
probably was bought used if i had to guess
RIP 4090
Goddamn how is it possible that new cards like that one go bad? Unless you used it improperly
Well, if you can get hold of a few of those broken PCB 4090s from Northbridge, you can fix that one ;-)
its balsday everyday for you
it's ballsday baby ⚽⚾🏀
If I saw that on ebay, I'd demand my $50 back
I am serious when I say STOP overclocking graphics cards. Rather underclock them. I underclocked my 4080. I lost about 3 to 5fps but my wattage is about 90w to 100w, games fully playable, also temperature is about 60 degrees. If I dont underclock, card easily pulls 180w to 250w, and runs close to 80 degrees. Also, use the support on the back of your card. If you dont have a support, let your case rather lay flat, the stress on the board with this heavy cards causes problems with the PCI Express slot, and card cracking. If the card pulls on the slot, it can short itself per accident, and you lose your core or memory.
Well at the end of the day the gpu core is expensive for itself not the pcb or any other components no wonder why people buying the 4090 and not willing to pay for the repair to just replace the chip. This where I draw the line in repairs