• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Time for a new gpu?

Joined
Jan 25, 2014
Messages
2,098 (0.50/day)
System Name Ryzen 2023
Processor AMD Ryzen 7 7700
Motherboard Asrock B650E Steel Legend Wifi
Cooling Noctua NH-D15
Memory G Skill Flare X5 2x16gb cl32@6000 MHz
Video Card(s) Sapphire Radeon RX 6950 XT Nitro + gaming Oc
Storage WESTERN DIGITAL 1TB 64MB 7k SATA600 Blue WD10EZEX, WD Black SN850X 1Tb nvme
Display(s) LG 27GP850P-B
Case Corsair 5000D airflow tempered glass
Power Supply Seasonic Prime GX-850W
Mouse A4Tech V7M bloody
Keyboard Genius KB-G255
Software Windows 10 64bit
A few weeks ago this happened to me with the only difference that the lines were pink. It kept boot looping after a few white screens of death. After reinstalling the windows it seemed to work for about two weeks but today it happened again. Is it time to change the gpu? Or could the reason be something else? Windows wont load so I cant check the event log. It starts loading windows and the restarts and like that in circles.
 

Attachments

  • IMG_20200730_094923.jpg
    IMG_20200730_094923.jpg
    3.9 MB · Views: 289
  • IMG_20200730_094937.jpg
    IMG_20200730_094937.jpg
    6.2 MB · Views: 303
  • IMG_20200730_094943.jpg
    IMG_20200730_094943.jpg
    5.2 MB · Views: 288
Have you check your cable between the graphics card and monitor and tried another one? if yes tried reseating the graphics card? if this is also a yes then it might be time for a new nice graphics card.

I can see you have a AMD Radeon 280X what does you use your graphics card for?
 
Have you check your cable between the graphics card and monitor and tried another one? if yes tried reseating the graphics card? if this is also a yes then it might be time for a new nice graphics card.

I can see you have a AMD Radeon 280X what does you use your graphics card for?
Sometimes I use it for gaming, but lately less and less. If I were to get a new one probably would not get anything too expensive. Maybe something like a 1050ti
 
Looks like vram is dying or something.
 
Since everything inside your computer case depends on good, clean, stable power, I would swap in a spare power supply to see what happens before spending money on any other hardware. If necessary, borrow one from another computer from a trusting friend or relative.

Also, it looks like the motherboard came out in 2011. It might be time to budget for a whole new computer - especially if the remaining components are about the same age. If the budget does not allow for that any time soon, and you get the same results with a spare PSU, I would not spend a lot of money on a new card. I recommend getting a budget card for now - something just to get the computer productive again for important tasks like work, school, paying bills, etc.

If the spare PSU clears your problems, buy a quality replacement, at least 80 PLUS Bronze, preferably Gold. I like EVGA and Seasonic.
 
Maybe try lowering clocks on kt to buy you time... but if cables and a fresh os install didn't do it... from looks of it, than thing is on the way out.
 
I would follow the previous troubleshooting advice before trying such a thing, but if all else fails...

I had a similar problem to this with my Radeon 8500 back in the day and I baked it in the oven at 350 for a 1/2 hour (which I'm told was way too long, should have only been ten minutes or so lol). I can't remember how much more time I got out it but it did work for a while before it finally died completely.

Note: If you try this, I would recommend not touching it/taking it out of the oven until it is completely cooled.

That poor card, I abused it so bad even before that... lol
 
I would also say to you to test, clean, check fans, fans speeds.
But, I also think that VRAM is dying, or there's some capacitor on the GPU dying. There's some technicians that fixes GPUs, but I can't recommend on for you, based on location;
I would do a cleaning, and thermal paste changing, and cross my fingers.

Good luck sir!
 
I baked it in the oven at 350 for a 1/2 hour (which I'm told was way too long, should have only been ten minutes or so lol).
I wish I knew where this idea originated and I had a TARDIS because I would go back in time and slap the original person to suggest this upside the head!

There are so many reasons NOT to try this.
  • Note that 350°F = 176.67°C.
  • As seen here, many solders are designed to melt at 180 - 190°C (356°F - 374°F) or even less.
  • Ovens typically have a temperature cycle range of ±25°. That is, when set to 350°F, the thermostat takes the oven up to 375°F (190.56°C) then it kicks off until it cools to 325°F, then it kicks on again taking it back up to 375°F.
  • Oven thermostats are notoriously imprecise.
  • The heat radiating off the red-hot heating element when it is cycled on can easily be over 1000°F (a common self-cleaning oven temp).
  • Many components, including the GPU, RAM and other devices were never ever designed to tolerate those temperatures.
  • The PCB substrates (layers) can separate.
  • Circuit traces can lift off the board.
  • Caustic fumes can develop.
  • Oh, and then there's fire.
In ancient times when vacuum tubes and crystal ovens and other excessive heat generating devices were used in electronics, this oven trick might have made sense for a last ditch effort. But with high-density, heat sensitive microelectronics, it doesn't. Please don't try this.
 
I wish I knew where this idea originated and I had a TARDIS because I would go back in time and slap the original person to suggest this upside the head!

There are so many reasons NOT to try this.
  • Note that 350°F = 176.67°C.
  • As seen here, many solders are designed to melt at 180 - 190°C (356°F - 374°F) or even less.
  • Ovens typically have a temperature cycle range of ±25°. That is, when set to 350°F, the thermostat takes the oven up to 375°F (190.56°C) then it kicks off until it cools to 325°F, then it kicks on again taking it back up to 375°F.
  • Oven thermostats are notoriously imprecise.
  • The heat radiating off the red-hot heating element when it is cycled on can easily be over 1000°F (a common self-cleaning oven temp).
  • Many components, including the GPU, RAM and other devices were never ever designed to tolerate those temperatures.
  • The PCB substrates (layers) can separate.
  • Circuit traces can lift off the board.
  • Caustic fumes can develop.
  • Oh, and then there's fire.
In ancient times when vacuum tubes and crystal ovens and other excessive heat generating devices were used in electronics, this oven trick might have made sense for a last ditch effort. But with high-density, heat sensitive microelectronics, it doesn't. Please don't try this.
it works in some cases, despite your objections believe it or not o_O unless myself and countless othe TPU members as well as 100's of other such cases noted on the web are all part of some big conspiracy to lie about such a thing
 
I wish I knew where this idea originated and I had a TARDIS because I would go back in time and slap the original person to suggest this upside the head!

There are so many reasons NOT to try this.
  • Note that 350°F = 176.67°C.
  • As seen here, many solders are designed to melt at 180 - 190°C (356°F - 374°F) or even less.
  • Ovens typically have a temperature cycle range of ±25°. That is, when set to 350°F, the thermostat takes the oven up to 375°F (190.56°C) then it kicks off until it cools to 325°F, then it kicks on again taking it back up to 375°F.
  • Oven thermostats are notoriously imprecise.
  • The heat radiating off the red-hot heating element when it is cycled on can easily be over 1000°F (a common self-cleaning oven temp).
  • Many components, including the GPU, RAM and other devices were never ever designed to tolerate those temperatures.
  • The PCB substrates (layers) can separate.
  • Circuit traces can lift off the board.
  • Caustic fumes can develop.
  • Oh, and then there's fire.
In ancient times when vacuum tubes and crystal ovens and other excessive heat generating devices were used in electronics, this oven trick might have made sense for a last ditch effort. But with high-density, heat sensitive microelectronics, it doesn't. Please don't try this.
Yada... it isn't for you. Understood. It is higher risk, indeed. But if you have a dead/dieing card it makes sense today as a last ditch effort for those who want to try it.

That said, when done properly, this has saved COUNTLESS video cards from dieing (though of course it doesn't always work).......... or should I say extended their lives longer. It's all over the place and there is no way I'd tell people not to... just be honest with the warning. I'd pick apart some of your reasons, but really don't want to go down a Bill Bright hole. Just know I'd slap you right back (with a white glove, like a gentleman) for not sharing and trying to hide this information...lol.
 
I wish I knew where this idea originated and I had a TARDIS because I would go back in time and slap the original person to suggest this upside the head!

There are so many reasons NOT to try this.
  • Note that 350°F = 176.67°C.
  • As seen here, many solders are designed to melt at 180 - 190°C (356°F - 374°F) or even less.
  • Ovens typically have a temperature cycle range of ±25°. That is, when set to 350°F, the thermostat takes the oven up to 375°F (190.56°C) then it kicks off until it cools to 325°F, then it kicks on again taking it back up to 375°F.
  • Oven thermostats are notoriously imprecise.
  • The heat radiating off the red-hot heating element when it is cycled on can easily be over 1000°F (a common self-cleaning oven temp).
  • Many components, including the GPU, RAM and other devices were never ever designed to tolerate those temperatures.
  • The PCB substrates (layers) can separate.
  • Circuit traces can lift off the board.
  • Caustic fumes can develop.
  • Oh, and then there's fire.
In ancient times when vacuum tubes and crystal ovens and other excessive heat generating devices were used in electronics, this oven trick might have made sense for a last ditch effort. But with high-density, heat sensitive microelectronics, it doesn't. Please don't try this.
Well I certainly wouldn't try it unless the only alternative is the trash/recycle bin, that's for sure.
 
I never said it has not worked or been successful. But I will say "countless times" is a bit of an exaggeration.

The problem is, there are too many unknowns here - some of which could lead to hazardous situations.

"IF" I knew the solder composition/melting point,
"IF" I had a precision oven,
"IF" I was in a well ventilated place,
And "IF" all other attempts to repair it had failed, then I might try it too for Ss & Gs. But I sure would not start out at 350°F.

Drastic measures are fine, for a last resort and you know what you are doing. Heck, we even hosed out a UHF air traffic control receiver then stuck it in the Phoenix sun for 2 days. It worked, btw, but that was not a recommended repair. But it didn't work for its transmitter counterpart. :(
 
Drastic measures are fine, for a last resort
Most people who advise "baking" advise so on this basis, "last resort" and honestly, there's too many if's in that reply for a otherwise dead card. 175c is what is generaly recommended temperature for 10-15 minutes only and what I have had success in the past so who are you to argue if you haven't done it and "fixed" (even if temporarily) a GPU by baking?

In OP's case, it defo looks like one of the vram chips have gone/going, so lower clocks and voltage would be my first resort and being as it's a 280x which are about 7-8 years old? with a used value on ebay of about $30-40? so if it's definitely DOA then he has nothing to lose baking as a last resort.
 
so who are you to argue
LOL So you can post your opinion on this but I can't? I never said it couldn't work. I never said anyone was wrong. I only showed how there are many reasons not to try it. And I said "Please don't try it".

But most importantly, I never degraded the thread by resorting to personal affronts against you or ED or Prime2515102 when I expressed a different viewpoint.
 
LOL So you can post your opinion on this but I can't? I never said it couldn't work. I never said anyone was wrong. I only showed how there are many reasons not to try it. And I said "Please don't try it".

But most importantly, I never degraded the thread by resorting to personal affronts against you or ED or Prime2515102 when I expressed a different viewpoint.
I said who are you to argue that it doesn't work in some cases. Why was it personal against you lmao. And you can have all the opinions you want, doesnt mean people have to agree with you buddy :cool: it works in some cases was the point, but if you want to turn it into you being personally affronted..... then I see you are one of the always offended types, I'm sorry if I hurt your feelings :p
 
But I will say "countless times" is a bit of an exaggeration.
I watched this work countless times. I honestly cannot count how many times I have seen this work (you know the definition of countless) over the years across the forums I frequent. Again that doesn't mean it always does, it certainly doesn't... in fact it may fail more often than not (depends on what the problem is). But I have, in fact, seen this work (as a last ditch effort) countless times - it is NOT an exaggeration. Dozens of times at least... (so maybe I can count it, lolol)

Taking the blue pill and getting the fuck out of this conversation though..........
 
I tried checking the cable. Didn't fix it.
Tried taking the gpu out and cleaning the gold contact pins with alcohol, and put it back in. Didn't fix it.
Took it out again,replaced the thermal paste. Didn't fix it.
Tried putting in my old HD 7850 that was broken, and it kind of worked for a few hours.
When I say kind of worked. I mean it worked, and let me get into windows, but did not let me install GPU drivers and when I opened GPU-Z there were no stats.
All the tabs were empty.
Then the PC randomly rebooted and it wont load Windows again, it just boot loops.
Will try a new PSU, but before that I will try to get my hands on a used HD 5450 or something really low end and see if that fixes the issue.
Thank you all for the feedback. Will report back how I managed to fix it, and if I did.
 
but before that I will try to get my hands on a used HD 5450 or something really low end and see if that fixes the issue.
But you already tried a different card and it didn't work right either. Are you not certain the HD 7850 was good? You say you will try a "new" PSU. Do you mean you don't have access to another you can try and that you would have to buy a new PSU?
 
Are you not certain the HD 7850 was good? You say you will try a "new" PSU. Do you mean you don't have access to another you can try and that you would have to buy a new PSU?
The 7850 was black screening a few months ago so it got replaced with the 280x. But I thought I give that one a try, but device manager does not seem to recognize it .
I will not buy a new PSU unless I certainly know that it is the culprit. I have a spare psu from an older build. I just have to find it first. :laugh:
 
That spare PSU better be good, the R9 280X sucks/blows 300W when running fully loaded (furmark or newer game engines) - I know I have MSI HD7970 OC BE (1050 Boost clock) -> cold (max 67°C) but thirsty.
 
Tried the spare psu. The issue is the same so I can rule out the psu. Trying to get myself a used hd 7770.
 
Sorry that failed but at least now you know before spending money on a new PSU you didn't need. Good luck. Keep us posted.
 
Got my new gpu today. Well not really new, but you know. It's a HD 7770 GHz edition and the Pc works fine for now. I just need it to hold out until Christmas. Then I plan to build a new PC. This one is long in the tooth. Thank you all for the suggestions.
 
Back
Top