Owner of this card, thanks again Tony! Highly recommend him to anyone needing a repair but if you are here reading this, you probably already knew that.
I think you let it sag that’s why it happened. Maybe vertical mount it or get an anti-sag bracket/brace please. Or like he said shoved a screwdriver where it wasn’t supposed to go.
@@timbersrcadventures hey man, he didn’t share in the video but the whole pc was knocked over pretty violently by a German shepherd. It has already been replaced by a 4080 super, this was just a repair so I could give the card to my younger sibling
@@northwestrepair you are welcome. i watched your rust repair and was flabbergasted why you would even take the work. if its for the video then i would love to see a historical gpu repair series. fixing a broken voodoo anyone?
@@patrickderp1044 He did once before fix (or try to fix) some very old AGP amd card Problem with those is no schematics/boardviews, no clue what's what, what's the voltage in some place supposed to be, no old vram chips for replacement, etc
Hey Tony, saw your x16 issue wioth the motherboard. To confirm its the CPU which has the x16 BUS controller codes coming off it to directly link the first PCI x16 slot. Means no chipset involved. If this slot is working x1 speed, then the x16 codes are not initiated from the CPU PCI Bus controller. MEans needs to try replacing the CPU to another one which support x16 bus controlling properly. Secondly, a matching motherboard to support such CPUs to facilitate the direct BUS code management of x16 PCI buses on the motherboard. Normally the first one or the first 2x PCI slots can be controllers directly by CPU without any chipset managing it. Hope this tip helps. I had this issue, and had to change CPU and motherboard to match and then the first PCI slot started working on x16. second PCI slot normally shares with the M2 NVME slot. In some motherboards, the first slot is shared and second slot is a dedicated x16. Hope this helps your diagnosis. Good luck 👍👍
i want so bad to do that work for my entire life, it's so cool to see a professionnal with this amount of knowledge work and showing what and how he repair Gpu, thanks man
AMD generally utilizes the same socket for A LONG TIME (6+ years). Because of this, newer components may be physically compatible, but the older one (generally the motherboard) needs a software (BIOS) upgrade to work. AMD uses AGESA and Infinity fabric to tune your specific hardware combination to the best and most stable configuration each time you change any hardware
This right here, depending on the CPU used on that B450 mobo, a BIOS update is strongly recommended. Especially if you are using a Ryzen 5000 series CPU. With AMD and how long they hold on to sockets, it's always best to update the BIOS as soon as you get it.
22:14 how ironic, I like watching that part at regular speed! Something very satisfying about the cleaning process. I bet I'm in the minority, but I wanted to put it out there. Great vid!
bought for a friend the same design gigabyte cooler but 6750xt amd and i opened the box to check if everything's ok. Removed the card from the antistatic bag and inspected a little bit around. Noticed a light shining through some area it shouldn't. Using a heat gun warmed up a bit the cooler to make the separation easier and voila. The thermal pad strip on the vrm in the middle of the card didn't have any contact at all with the pads. Zero imprint! Grabbed a 1mm pad i think, replaced the thermal paste that was completely dried out and re-assembled the card. Then removed it again to check for paste spread and if the pad was contacting. So far the card works perfectly fine. I don't know what they had for lunch that day but definitely they had a little too much to drink! I didn't rma the card for that. i knew they wouldn't had done better job than i did. Also the card was a gift and i didn't want to spend more time shipping it from Greece to whatever they'd supposed to idk
My motherboard has PCI-E Bifurcation. Where it would split the PCI-E lanes between the GPU and the M.2 slot and had to figure it out and find out where to turn it off in the BIOS. Once I turned it off, my GPU was then recognized as x16 speed instead of x8 speed. Your bord may have PCI-E Bifurcation enabled by default, or something similar, that will split PCI-E lanes.
This is done automatically in the BIOS. Most likely the motherboard needs a BIOS update to satisfy the latest AGESA requirements for the CPU + Socket combination, as AMD has historically supported older sockets for 5+ years and sometimes the BIOS needs be updated to allow it to work correctly. PCIe automatic mode detection is 1 of these parameters.
one tip i can give for oxidised pads is that you should add some solder paste onto it then use a iron. When i did it for a first time i was surprised how quick and easy it was instead of scraping
I’ve come across the oxidize pad before many times. So I always felt that it happens at edges a ground planes cause from that area of the substrate not being up to temp during the removal process. It’s mailable enough to release from the pads but not liquid enough to leave a bump. I understand that you’re using a probe, but I believe the probe was on the other side. I also think you tap your component to see if it springs back. I don’t know if that part was cut out. Also the die in these types of package require careful temp profiling. I have noticed magnitudes less issues if I wait a bit longer on heat.
With regards to the AMD motherboard, the first thing I'd try is re-socketing the CPU and possibly even try some deoxit on the socket. PGA is notorious for socket connection issues happening once a CPU installed in a board is shipped. I've run into AM4 systems flat out refusing to fan-spin a countless number of times and re-socketing the CPU has always been the fix. Also, a few videos ago you asked about a decent thermal putty that wasn't ridiculously expensive. Upsiren UTP-8 from Aliexpress is like $20/100G and performs at least as well as the thermalright pads so it's a good choice to use in scenarios where clearance is an issue with typical pad sizes. There's also U6Pro which is less viscous but that stuff is a little sticky. There's also some unbranded CX-H1300. Do not get K5-Pro or the knockoff U6Pro that looks like K5-Pro with pink coloring. It needs to at least be thick enough to mold with your hands into little balls and have a rough gritty texture, if it's runny it's garbage and probably worse than just using copious amounts of thermal paste.
You hit the nail on the head with the putties. UTP-8 is incredible. K5-Pro and their knockoff U6Pro is utter junk. I attempted to use K5-Pro on a laptop, 6 months later I cracked it open and discovered bubbling/pitting in the putty. It left bubble marks all over each memory chip. It literally boiled under the temps of the memory......
i get that prices change but it would be nice for you to put the price of the repairs in the video and a disclaimer that prices may not be current. But it makes it hard to tell when it's worth it for someone to send a card in vs just buying a new one or buying a used working card for less.
For reference, I replaced this card with a 4080 Super before I sent it to Tony. This repair cost me ~160 after shipping both ways, etc. which is worth it considering this is still a viable card that I can gift to someone else.
@@west9641 Thanks for the info, that's awesome. Was it a guaranteed price to fix up front? Or does it depend on what's wrong with it and he shoots you an email to find out if you want to pay to fix or just pay for 'no fix'? It's always hard to tell if there are big gaps in videos like this where the person stops and contacts someone and then comes back to it later.
@@zachansen8293 price changed based on repair success and parts needed. Baseline was like 120 plus shipping both ways if it was fixable, 40 for a failed attempt, and then parts on top of that. For me, I didn’t include the thermal pads because I tore them when taking it apart the first time so I had to pay for those too. You agree to this pricing upfront and then just wait. This video came out before I even knew he had the card in his possession which is impressive.
Tony usually on AMD cpu the failing progress of an overtemperature cpu is the progressive failing of pci express lines .. from 16x to 8x to 4x .. conpletely dead pci-x slot ( issue on the cpu .. not Mb ) .. switching to another pci express slot ( if available at full speed ) maybe can solve but probably the cpu is dying
Try updating the bios. I know my friend was having a similar issue as to what you were describing, and he went through this whole entire troubleshoot saga. In the end, he figured out that his motherboard (I think it’s asus as well) what shipped with a bios with a software bug in it. After he updated, the problem was completely solved. His system is also a AMD system btw. I also seem to remember reading up that a lot of motherboards were shipping with a bios with said bug in it.
Asus bios have been trash a long time now and msi to when a website has like 30 bios for 1 board you know it's constant revisions because they don't make the bios right
@@JohnChrysostom101 Silly, they delete intermediate and old beta versions or versions with known severe bugs, they unpublish most of them and just keep good milestones, or sometimes they unpublish most of them regardless after some time passes. Thing is there's constant AGESA updates from AMD and they're needed to fix AMD's bugs and introduce support for new processors that keep coming out, and AMD released new processors for old AM4 boards just months ago again, sooo... of course there's going to be a lot of BIOS revisions for an old board! Unless it got abandoned and is no longer getting new CPUs... other reasons for BIOS releases is if they have released a new hardware revision of the mainboard, say the ethernet chip or the soundchip had to be substituted for availability reasons and they'll add support for both into the same BIOS. So counting BIOS versions is a completely pointless exercise.
Regarding the motherboard related issues, I would apply factory defaults and update the bios of the motherboard(s) to the latest version first for sure, if not attempted yet. Just to rule out the bios being the culprit. Of course, the customer may or may not be comfortable doing that, so that can be a situation to solve. If that doesn't resolve the problem, maybe clear the cmos by removing the battery. After that it looks to be about testing with different motherboards and cards.
Yo.. Hey I use a fine scuff pad on the everything boards chips it works amazing just keep it lubricated using alcohol or something light on the pressure.. It's good to try it out on an old board or GPU obviously.. It's also good for knocking off any leftover solder on the pads before a recall.. Anyways keep up the good work God bless!
Closest to the PCIe connector with the hook and latch which often flexed and cracked if the GPU was installed in a typical tower case with no brace to keep the GPU level.
Who else is here watching because they have a 3070 and we never get any love on this channel? Much love for 4080 and 4090 owners but 3070 owners? no Love!
Tony, can you tell us how you got into gpu repair and how long it took you to get to where you are with your abilities in repairing gpus? I'm interested in knowing.
Just saw your comment on the Tech Powerup website regarding the 1x PCIe lane, that is very strange, may have been a flaw with that specific B450m board/the 2200g you’re using
Great repair as always. Since I got a 3070 8 gigabytes myself, the question is this. Would it make sense money wise, to upgrade the card to 16 gigabytes of RAM? Assuming of course it is at all possible and if it is would it be possible to upgrade it to a 3070 ti?
Fun fact Earth has a high degree of surface flatness. The tallest mountains and the deepest oceans are barely so much height variation that it's as smooth as a ball bearing.
on the AMD MB stuck in X1 PCIE mode, a quick google search shows that it seems to be a common issue. Some say it only happens on cold start, and will switch to X16 after a reboot. Some say it's related to the GPU BIOS Some say it's related to MB bios ... not clear ! Kudos for going that far in the troubleshooting, I'm sure you'll find the answer.
can't be related to GPU bios afaik, the slot lanes are mb. No vbios will choose to go 1x, also afaik there is no way to do this because there is no reason to be able to disable/enable that sort of thing on GPU.
In my experience what can cause PCI E errors: Cpu being clamped too hard/incorrectly on socket. Maybe over tightened screws. Issue with Power supply. It could be the power supply is damaged or something is wrong with the electrical outlet in his room/house. That could also come from a bad device being plugged into the outlet. Based on yourvideos I want to assume something on the card is shorting? Or that issue with power has corrupted something on the card.
What CPU you using Tony? Specs for the board: 3rd/2nd/1st Gen AMD Ryzen™ Processors 1 x PCIe 3.0/2.0 x16 (x16 mode) 2nd and 1st Gen AMD Ryzen™ with Radeon™ Vega Graphics Processors 1 x PCIe 3.0/2.0 x16 (x8 mode) *1 AMD Athlon™ with Radeon™ Vega Graphics Processors 1 x PCIe 3.0/2.0 x16 (max at x4 mode) AMD B450 chipset 2 x PCIe 2.0 x1
Hey Tony, what CPU do you have in that AM4 motherboard? I reported a BIOS/UEFI issue on Asus Prime A320M-E which they (supposedly) confirmed with AMD that it's an AGESA bug, up to my understanding it exists on CPUs with PCIe x8 bus (2200G/3200G etc.) Try a bios for your Mobo with AGESA ComboV2PI
Hello🤝Tony, you probably know this, but older AMD socket AMD4 processors such as Athlon 200GE do not support the entire PCI-e bus. I didn't know it before and I thought that I had a damaged card or PCI-e slot. It's best to read the processor specifications. Best regards.
I've got the same problem with PCIE port and Ryzen. In my case it's a CPU. 5600x works just fine at x16 and 3700x only does x1. X570 board with newest bios installed.
could also be settings in bios.. if i remember correctly some m.2 and sata drives were tied together with pcie and if a certain combo is used it reduces speed of pcie
oho i just replaced exact same card for msi 4080. Great card served me very well, but its old, never opened and i don't dont to repaste it myself neither send it somethere. Seen way to much failed repairs or cards damaged in delivery
that motherboard is at the best pcie 3 x16 with the right CPU, that motherboard can use an old "APU" CPU + integrated GPU and I recall some limitation with that kind of processor
Have you tried to use ammonium chloride as a flux ? when I have a lot oxidation that cannot be fixed by a regular flux, I use it. It has to be washed with DDW and then with alcohol after use.
Ive got that 1x before on a intel board when I cleaned my gpu and inserted it but I cleaned the slot and contacts which seemed to fix it. I still check gpuz every time though just to make sure. For the AMD boards I think quite a few of their gpus only get 8x mostly if there is the igpu like the vega 8.
Have 3070 TI from Colorful, don't know why the card even crash with default setting what ever the driver version is. Luckyly MSI afterburner is there, just undervolt it and down the TDP to 75% to make it work fine. I don't know what the manufacturer thinking when the card even can't run with their default setting.
I have an old hd6670 from amd which is stuck on 4x mode but it works perfectly fine and gets similar score from other hd6670, except the random no display few times a year. I gave it to a repair shop they couldn't fix it, cos they didn't understand what is causing it to run at only 4x.
Is the Ryzen CPU on that motherboard one with integrated graphics? AM4 CPUs with integrated graphics or those with fused off integrated graphics are limited to a x8 PCIe slot for the GPU. AM4 CPUs without integrated graphics have a x16 PCIe slot. If you want a second set of graphics outputs on an AM4 CPU with a x16 slot it means you need a second graphics card. You seem to have at least one mining riser, so connecting an old GT710 or equivalent AMD card to a x1 slot with such a setup should be easy enough. FYI, the AM5 Ryzen 8000 series are also limited to a x8 PCIe slot for the GPU.
First off, the Asus B450 only has a 3.0 pcie controller, and I would bet the 4080 is a Pcie 4.0 or even a 5.0, which the 4080 won't perform to it's full potential.
There might be some kind of voltage feedback from GPU to PCIE that intel board can handle but AMD can't. That break the AMD board when y insert that GPU. Y Should try to read PCIE lanes from that GPU when powered using oscilloscope to see if there is some weird spikes when compared to known Good GPU. If Bad GPU breaks PCIE slot it do not matter what GPU y but in the slot.
I think you bought a wrong platform. Should have gotten a a cheap B550 board. What you need is PCI-E 4.0 and preferbly BIOS flashback button feature. After so many AGESA updates from AMD, weird things do happen, like wonky USB controllers from the beginning of Ryzen 3000 series , so you need to be able to test multiple BIOS revisions to troubleshoot as well. B450 is abit too old of a chipset to test accurately. By the way, riser cables always cause problems with PCI-E lanes not detected and stuff. Some risers only work with PCI-E 3.0 and you need to set in BIOS before u plug in the GPU , or else it won't POST .
Lol, Gigabyte: That Rube Goldberg power connect solution! No way it cost less than just enlarging the PCB and using a proper "dual single-Connect" method. (Every plug is a possible failure point!) The connector vendor was happy though, his P.O. value for this card run more than tripled, he, he!
Update the BIOS on motherboard. Guy on FB Marketplace listed a 4080 bc it didnt work on first motherboard slot but worked on 2nd at 1x. Told him to update bios and it worked in first slot at 16x
After seeing a lot of your video am starting to not recommend gigabyte ever, the amount of gigabyte card you fixed is insane. are you suggest MSI over gigabyte
Make sure you're not using a "g" part or a part that is a g part without the igpu dissable (such as a 5500) if you do change out the cpu. Also make sure bioses are all updated and both MB and GPU are set to same pcie generation and boot from sata. Though I suspect you've already updated the bioses and checked pci setting which I why I started elsewhere. Also used has higher likelihood of being defective in some subtle way. last though, a scary one though, is a contagious hardware problem. Last one of those I know of was for 1.44MB micro floppy drives and the Mobo side controllers used for them.
Why not G? I mean he isn't after performance, and memory test and the like without a working video out is unwieldy. There's a full 16x bus coming out of the G CPU anyway. Cezanne has Gen3 instead of Gen4 but that's OK.
@@SianaGearz Not sure what I was thinking that long ago, I do know g parts and their relatives such as the 5500 don't have pcie-4 that you need for lower end gpu's that only have pciex8 interfaces as they can and will bottleneck on pcie3. Also there are smaller caches on those cpus, space being needed for the igpu, also you loose pcie lanes , 4 lanes IIRC. The g parts are all around weaker prior to zen4. Again I don't recall what prompted me to make that observations, but those are the weaknesses of the g parts.
@@kaseyboles30 See it's not a computer, it's a test rig for GPUs, to see whether they have hardware issues. It'll do fine with a 2200G, which is not a CPU one would recommend for any other use except light office use. You can not bottleneck Furmark. FWIW my actual computer is a Cezanne, and i know i'm not going to have x4 or x8 narrow bus GPUs in here, the price difference between 5500 and 5600 was immense, it was almost half the price, and cheaper than 3600 and usually faster, and it's still perfectly usable several years later. You do not lose any PCIe lanes.
Bro, let me just tell you bro... You are good at what you do. Your nonchalant defiance against - N "big daddy" vidia (Nvidia 4 u slows out-there) is indicative of a degree of confidence that portray those whom have mastered there craft or atleast close to it. Like you are Tony Montana and Nvidia are the cockroach...'s. You seldom go into technical explanation of what it is that you are doing on video and move on to the next clip/cut/shot/angle (what ever you call it) and then queue music montage, and I get it, out of respect for the people you do not want to be that guy who preaches at the people, a "mr know it all". But bro, let me tell you bro... This is the first video in which I see you, not only go into a technical discussion on what it is you are doing, but you are also fkn dope azz fk at it!!! I am oneof those who can barely identify a resistor and never even seen what it is that you do before, and I found myself hanging on what you were saying and I realized I was fkn learning... . ¡¡¡Fkn ey, great video!!! .
Rewatching this video gave me both doubts on if I can fix one of the cards I have sitting aorund and a question. I've got a 3070 that gives error 43 (AMD build), but it also has a memory error so hearing the error 43 on AMD builds makes me wonder if fixing the memory issue would still have the same result.
i had a problem on one intel board after replacing graphic card slot. stuck in x1 after. never figure out if that came from changing a slot or something else in a process gets damaged.
Owner of this card, thanks again Tony! Highly recommend him to anyone needing a repair but if you are here reading this, you probably already knew that.
I think you let it sag that’s why it happened. Maybe vertical mount it or get an anti-sag bracket/brace please. Or like he said shoved a screwdriver where it wasn’t supposed to go.
🙏
Could you please share how mush did it cost ?
@@timbersrcadventures hey man, he didn’t share in the video but the whole pc was knocked over pretty violently by a German shepherd. It has already been replaced by a 4080 super, this was just a repair so I could give the card to my younger sibling
@@m0neez 160ish shipped there and back
HE UPLOADED, WE SHALL WATCH
Lol thanks.
@@northwestrepair you are welcome. i watched your rust repair and was flabbergasted why you would even take the work. if its for the video then i would love to see a historical gpu repair series. fixing a broken voodoo anyone?
@@patrickderp1044 He did once before fix (or try to fix) some very old AGP amd card
Problem with those is no schematics/boardviews, no clue what's what, what's the voltage in some place supposed to be, no old vram chips for replacement, etc
🙌
Tony makes core reballing look like child's play. Respect
it mostly is if you know what youre doing :D
Hey Tony, saw your x16 issue wioth the motherboard. To confirm its the CPU which has the x16 BUS controller codes coming off it to directly link the first PCI x16 slot. Means no chipset involved. If this slot is working x1 speed, then the x16 codes are not initiated from the CPU PCI Bus controller. MEans needs to try replacing the CPU to another one which support x16 bus controlling properly. Secondly, a matching motherboard to support such CPUs to facilitate the direct BUS code management of x16 PCI buses on the motherboard. Normally the first one or the first 2x PCI slots can be controllers directly by CPU without any chipset managing it. Hope this tip helps. I had this issue, and had to change CPU and motherboard to match and then the first PCI slot started working on x16. second PCI slot normally shares with the M2 NVME slot. In some motherboards, the first slot is shared and second slot is a dedicated x16. Hope this helps your diagnosis. Good luck 👍👍
Check the bios settings, had the sake issue with my RTX4070Ti with 3900x processor. It was the setting in the bios regarding the memory (xmp).
@@Q96500how can XMP be related to PCIe 🤔
Morning coffee + reballing cores = perfect start of my day
i want so bad to do that work for my entire life, it's so cool to see a professionnal with this amount of knowledge work and showing what and how he repair Gpu, thanks man
AMD generally utilizes the same socket for A LONG TIME (6+ years). Because of this, newer components may be physically compatible, but the older one (generally the motherboard) needs a software (BIOS) upgrade to work. AMD uses AGESA and Infinity fabric to tune your specific hardware combination to the best and most stable configuration each time you change any hardware
This right here, depending on the CPU used on that B450 mobo, a BIOS update is strongly recommended. Especially if you are using a Ryzen 5000 series CPU. With AMD and how long they hold on to sockets, it's always best to update the BIOS as soon as you get it.
22:14 how ironic, I like watching that part at regular speed! Something very satisfying about the cleaning process. I bet I'm in the minority, but I wanted to put it out there. Great vid!
I truly respect your craftsman ship
28:57 “I don’t need the fast and best, I just need enough to test”
I need this on a shirt 😁
bought for a friend the same design gigabyte cooler but 6750xt amd and i opened the box to check if everything's ok. Removed the card from the antistatic bag and inspected a little bit around.
Noticed a light shining through some area it shouldn't.
Using a heat gun warmed up a bit the cooler to make the separation easier and voila. The thermal pad strip on the vrm in the middle of the card didn't have any contact at all with the pads. Zero imprint!
Grabbed a 1mm pad i think, replaced the thermal paste that was completely dried out and re-assembled the card. Then removed it again to check for paste spread and if the pad was contacting. So far the card works perfectly fine.
I don't know what they had for lunch that day but definitely they had a little too much to drink!
I didn't rma the card for that. i knew they wouldn't had done better job than i did. Also the card was a gift and i didn't want to spend more time shipping it from Greece to whatever they'd supposed to idk
if you are aware of Tony's work at all, just watch and like his videos, his work is outstanding and deserves all the views and subs possible.
Slap The Like Button!
Thanks for no annoying background music
My motherboard has PCI-E Bifurcation. Where it would split the PCI-E lanes between the GPU and the M.2 slot and had to figure it out and find out where to turn it off in the BIOS.
Once I turned it off, my GPU was then recognized as x16 speed instead of x8 speed.
Your bord may have PCI-E Bifurcation enabled by default, or something similar, that will split PCI-E lanes.
This is done automatically in the BIOS. Most likely the motherboard needs a BIOS update to satisfy the latest AGESA requirements for the CPU + Socket combination, as AMD has historically supported older sockets for 5+ years and sometimes the BIOS needs be updated to allow it to work correctly. PCIe automatic mode detection is 1 of these parameters.
one tip i can give for oxidised pads is that you should add some solder paste onto it then use a iron. When i did it for a first time i was surprised how quick and easy it was instead of scraping
'Like' for the explanation on memory chips at the corner of the GPU vs. chips on the middle of the side of the GPU.
I’ve come across the oxidize pad before many times. So I always felt that it happens at edges a ground planes cause from that area of the substrate not being up to temp during the removal process. It’s mailable enough to release from the pads but not liquid enough to leave a bump. I understand that you’re using a probe, but I believe the probe was on the other side. I also think you tap your component to see if it springs back. I don’t know if that part was cut out. Also the die in these types of package require careful temp profiling. I have noticed magnitudes less issues if I wait a bit longer on heat.
I learned about GPU from this video, thank you man!
With regards to the AMD motherboard, the first thing I'd try is re-socketing the CPU and possibly even try some deoxit on the socket. PGA is notorious for socket connection issues happening once a CPU installed in a board is shipped. I've run into AM4 systems flat out refusing to fan-spin a countless number of times and re-socketing the CPU has always been the fix.
Also, a few videos ago you asked about a decent thermal putty that wasn't ridiculously expensive. Upsiren UTP-8 from Aliexpress is like $20/100G and performs at least as well as the thermalright pads so it's a good choice to use in scenarios where clearance is an issue with typical pad sizes. There's also U6Pro which is less viscous but that stuff is a little sticky. There's also some unbranded CX-H1300. Do not get K5-Pro or the knockoff U6Pro that looks like K5-Pro with pink coloring. It needs to at least be thick enough to mold with your hands into little balls and have a rough gritty texture, if it's runny it's garbage and probably worse than just using copious amounts of thermal paste.
You hit the nail on the head with the putties. UTP-8 is incredible. K5-Pro and their knockoff U6Pro is utter junk. I attempted to use K5-Pro on a laptop, 6 months later I cracked it open and discovered bubbling/pitting in the putty. It left bubble marks all over each memory chip. It literally boiled under the temps of the memory......
Your videos are really entertaining and very knowledgeable 👏 Thanks and keep em coming. Greetings from Germany 🍻
This knowledge and lesson are free🗣️
The byte of all gigas, that just cracked me up 😂
You're definitely getting a card with toothpaste instead of thermal paste now 🤣
but which brand of toothpaste is the best - we deserve benchmarks and testing!!!!1
@@Hotkife better be sensodyne
Well, at least their GPU will have a fresh, minty meltdown!
OK i wonder now when's the first person who puts on cum instead of thermal paste.
i get that prices change but it would be nice for you to put the price of the repairs in the video and a disclaimer that prices may not be current.
But it makes it hard to tell when it's worth it for someone to send a card in vs just buying a new one or buying a used working card for less.
For reference, I replaced this card with a 4080 Super before I sent it to Tony. This repair cost me ~160 after shipping both ways, etc. which is worth it considering this is still a viable card that I can gift to someone else.
@@west9641 Thanks for the info, that's awesome. Was it a guaranteed price to fix up front? Or does it depend on what's wrong with it and he shoots you an email to find out if you want to pay to fix or just pay for 'no fix'? It's always hard to tell if there are big gaps in videos like this where the person stops and contacts someone and then comes back to it later.
@@zachansen8293 price changed based on repair success and parts needed. Baseline was like 120 plus shipping both ways if it was fixable, 40 for a failed attempt, and then parts on top of that. For me, I didn’t include the thermal pads because I tore them when taking it apart the first time so I had to pay for those too. You agree to this pricing upfront and then just wait. This video came out before I even knew he had the card in his possession which is impressive.
Tony usually on AMD cpu the failing progress of an overtemperature cpu is the progressive failing of pci express lines .. from 16x to 8x to 4x .. conpletely dead pci-x slot ( issue on the cpu .. not Mb ) .. switching to another pci express slot ( if available at full speed ) maybe can solve but probably the cpu is dying
Try updating the bios. I know my friend was having a similar issue as to what you were describing, and he went through this whole entire troubleshoot saga. In the end, he figured out that his motherboard (I think it’s asus as well) what shipped with a bios with a software bug in it. After he updated, the problem was completely solved. His system is also a AMD system btw. I also seem to remember reading up that a lot of motherboards were shipping with a bios with said bug in it.
Asus bios have been trash a long time now and msi to when a website has like 30 bios for 1 board you know it's constant revisions because they don't make the bios right
@@JohnChrysostom101 Silly, they delete intermediate and old beta versions or versions with known severe bugs, they unpublish most of them and just keep good milestones, or sometimes they unpublish most of them regardless after some time passes.
Thing is there's constant AGESA updates from AMD and they're needed to fix AMD's bugs and introduce support for new processors that keep coming out, and AMD released new processors for old AM4 boards just months ago again, sooo... of course there's going to be a lot of BIOS revisions for an old board! Unless it got abandoned and is no longer getting new CPUs... other reasons for BIOS releases is if they have released a new hardware revision of the mainboard, say the ethernet chip or the soundchip had to be substituted for availability reasons and they'll add support for both into the same BIOS.
So counting BIOS versions is a completely pointless exercise.
Try a second probe on the other side of the substrate to ensure you have equal heat temps in all areas.
BEST! content on YT...thank u sir for the entertainment 👍🏼😆👍🏼
I’m still very puzzled on how you are able to remove a lead free soldered core from the pcb with 20s of hot air. 😮
You heard the man don’t ship em dry, Spit On Dat Thang (im sorry)
Regarding the motherboard related issues, I would apply factory defaults and update the bios of the motherboard(s) to the latest version first for sure, if not attempted yet. Just to rule out the bios being the culprit. Of course, the customer may or may not be comfortable doing that, so that can be a situation to solve. If that doesn't resolve the problem, maybe clear the cmos by removing the battery. After that it looks to be about testing with different motherboards and cards.
Yo.. Hey I use a fine scuff pad on the everything boards chips it works amazing just keep it lubricated using alcohol or something light on the pressure.. It's good to try it out on an old board or GPU obviously.. It's also good for knocking off any leftover solder on the pads before a recall.. Anyways keep up the good work God bless!
Depends on the Processor i think, the am4 G Processors with intergrated Graphics have a reduced pci bandwith.
I recently repair the same model of 3070. It's always the B1 chip with ripped pads on the memory side 😂😂. Same video done by Krisfix Germany too.
Closest to the PCIe connector with the hook and latch which often flexed and cracked if the GPU was installed in a typical tower case with no brace to keep the GPU level.
Its bracked pressure mixed with almost no cooler pressurw which lets chip/solder bend, thats why you should dissemble cards often
Who else is here watching because they have a 3070 and we never get any love on this channel? Much love for 4080 and 4090 owners but 3070 owners? no Love!
Hey, did you check the bifurcation settings in the bios? Please Make sure to use the latest bios, also if you have integrated graphics, turn them off.
Tony, can you tell us how you got into gpu repair and how long it took you to get to where you are with your abilities in repairing gpus? I'm interested in knowing.
Just saw your comment on the Tech Powerup website regarding the 1x PCIe lane, that is very strange, may have been a flaw with that specific B450m board/the 2200g you’re using
Great repair as always. Since I got a 3070 8 gigabytes myself, the question is this. Would it make sense money wise, to upgrade the card to 16 gigabytes of RAM? Assuming of course it is at all possible and if it is would it be possible to upgrade it to a 3070 ti?
Good work 👍
You always do a great job looks simple but 💪
Fun fact Earth has a high degree of surface flatness. The tallest mountains and the deepest oceans are barely so much height variation that it's as smooth as a ball bearing.
Remember kids, no flex in the champagne room
Always plug in the 8 pin as well as the 24 pin, some boards require both plugged in.
For the pcie x1 issue try a bios update and also try to DDU and reinstall the latest AMD chipset drivers.
People really need to make sure the bios are updated. Make sure that's not the issue.
16:37 Baller move right there. Watch and learn kids!
on the AMD MB stuck in X1 PCIE mode, a quick google search shows that it seems to be a common issue.
Some say it only happens on cold start, and will switch to X16 after a reboot.
Some say it's related to the GPU BIOS
Some say it's related to MB bios
... not clear !
Kudos for going that far in the troubleshooting, I'm sure you'll find the answer.
can't be related to GPU bios afaik, the slot lanes are mb. No vbios will choose to go 1x, also afaik there is no way to do this because there is no reason to be able to disable/enable that sort of thing on GPU.
In my experience what can cause PCI E errors:
Cpu being clamped too hard/incorrectly on socket. Maybe over tightened screws.
Issue with Power supply. It could be the power supply is damaged or something is wrong with the electrical outlet in his room/house. That could also come from a bad device being plugged into the outlet.
Based on yourvideos I want to assume something on the card is shorting? Or that issue with power has corrupted something on the card.
What CPU you using Tony?
Specs for the board:
3rd/2nd/1st Gen AMD Ryzen™ Processors
1 x PCIe 3.0/2.0 x16 (x16 mode)
2nd and 1st Gen AMD Ryzen™ with Radeon™ Vega Graphics Processors
1 x PCIe 3.0/2.0 x16 (x8 mode) *1
AMD Athlon™ with Radeon™ Vega Graphics Processors
1 x PCIe 3.0/2.0 x16 (max at x4 mode)
AMD B450 chipset
2 x PCIe 2.0 x1
Hey Tony, what CPU do you have in that AM4 motherboard? I reported a BIOS/UEFI issue on Asus Prime A320M-E which they (supposedly) confirmed with AMD that it's an AGESA bug, up to my understanding it exists on CPUs with PCIe x8 bus (2200G/3200G etc.) Try a bios for your Mobo with AGESA ComboV2PI
We are here for the process though
Awesome video
Hello🤝Tony, you probably know this, but older AMD socket AMD4 processors such as Athlon 200GE do not support the entire PCI-e bus. I didn't know it before and I thought that I had a damaged card or PCI-e slot. It's best to read the processor specifications. Best regards.
Might be a bios issue or driver for the 3080
Weight of video card twisting from top end of the card. We need a support rail across
I've got the same problem with PCIE port and Ryzen. In my case it's a CPU. 5600x works just fine at x16 and 3700x only does x1. X570 board with newest bios installed.
could also be settings in bios.. if i remember correctly some m.2 and sata drives were tied together with pcie and if a certain combo is used it reduces speed of pcie
oho i just replaced exact same card for msi 4080.
Great card served me very well, but its old, never opened and i don't dont to repaste it myself neither send it somethere.
Seen way to much failed repairs or cards damaged in delivery
that motherboard is at the best pcie 3 x16 with the right CPU, that motherboard can use an old "APU" CPU + integrated GPU and I recall some limitation with that kind of processor
I use a 5900X on a b350 board (original ryzen gen). In the last bios works flawlessly. Had some hangs before.
Have you tried to use ammonium chloride as a flux ? when I have a lot oxidation that cannot be fixed by a regular flux, I use it. It has to be washed with DDW and then with alcohol after use.
Ive got that 1x before on a intel board when I cleaned my gpu and inserted it but I cleaned the slot and contacts which seemed to fix it. I still check gpuz every time though just to make sure. For the AMD boards I think quite a few of their gpus only get 8x mostly if there is the igpu like the vega 8.
Update the board bios.
i love it doing pad restoration :D
he may have bent cpu pins on the amd platform, could affect pcie speeds/signals
Have 3070 TI from Colorful, don't know why the card even crash with default setting what ever the driver version is. Luckyly MSI afterburner is there, just undervolt it and down the TDP to 75% to make it work fine.
I don't know what the manufacturer thinking when the card even can't run with their default setting.
I have an old hd6670 from amd which is stuck on 4x mode but it works perfectly fine and gets similar score from other hd6670, except the random no display few times a year. I gave it to a repair shop they couldn't fix it, cos they didn't understand what is causing it to run at only 4x.
Is the Ryzen CPU on that motherboard one with integrated graphics?
AM4 CPUs with integrated graphics or those with fused off integrated graphics are limited to a x8 PCIe slot for the GPU. AM4 CPUs without integrated graphics have a x16 PCIe slot.
If you want a second set of graphics outputs on an AM4 CPU with a x16 slot it means you need a second graphics card. You seem to have at least one mining riser, so connecting an old GT710 or equivalent AMD card to a x1 slot with such a setup should be easy enough.
FYI, the AM5 Ryzen 8000 series are also limited to a x8 PCIe slot for the GPU.
Heavy Power requirements. X1 if insufficient power ? Possibly due to ASUS DIGI VRM feature. Are Fans installed correctly?
Thank you!
Weird question but what's the most beautiful pcb you've seen
hey Tony so what did you do with these missing pads on the memory chip ?
First off, the Asus B450 only has a 3.0 pcie controller, and I would bet the 4080 is a Pcie 4.0 or even a 5.0, which the 4080 won't perform to it's full potential.
+1 for Gigabyte having power connectors not directly on pcb
20:18 Check out the corrosion on the swab!
23:40 box has been upgraded to blue
There might be some kind of voltage feedback from GPU to PCIE that intel board can handle but AMD can't. That break the AMD board when y insert that GPU. Y Should try to read PCIE lanes from that GPU when powered using oscilloscope to see if there is some weird spikes when compared to known Good GPU. If Bad GPU breaks PCIE slot it do not matter what GPU y but in the slot.
Are the ripped pads caused by heating/cooling cycles?
Sir top and bottom temp?
I think you bought a wrong platform. Should have gotten a a cheap B550 board. What you need is PCI-E 4.0 and preferbly BIOS flashback button feature. After so many AGESA updates from AMD, weird things do happen, like wonky USB controllers from the beginning of Ryzen 3000 series , so you need to be able to test multiple BIOS revisions to troubleshoot as well. B450 is abit too old of a chipset to test accurately. By the way, riser cables always cause problems with PCI-E lanes not detected and stuff. Some risers only work with PCI-E 3.0 and you need to set in BIOS before u plug in the GPU , or else it won't POST .
I would say ripped pads are manufacturer defect since the pads are not meant to pop off so easy. Or am i wrong?
Good work!
Lol, Gigabyte: That Rube Goldberg power connect solution! No way it cost less than just enlarging the PCB and using a proper "dual single-Connect" method. (Every plug is a possible failure point!) The connector vendor was happy though, his P.O. value for this card run more than tripled, he, he!
try to update the bios
Update the BIOS on motherboard. Guy on FB Marketplace listed a 4080 bc it didnt work on first motherboard slot but worked on 2nd at 1x. Told him to update bios and it worked in first slot at 16x
After seeing a lot of your video am starting to not recommend gigabyte ever, the amount of gigabyte card you fixed is insane.
are you suggest MSI over gigabyte
Hey, probably the problem is with the CPU because all motherboards support x16
and only the APUs from AMD work in x8 or even x1
Make sure you're not using a "g" part or a part that is a g part without the igpu dissable (such as a 5500) if you do change out the cpu. Also make sure bioses are all updated and both MB and GPU are set to same pcie generation and boot from sata. Though I suspect you've already updated the bioses and checked pci setting which I why I started elsewhere. Also used has higher likelihood of being defective in some subtle way.
last though, a scary one though, is a contagious hardware problem. Last one of those I know of was for 1.44MB micro floppy drives and the Mobo side controllers used for them.
Why not G? I mean he isn't after performance, and memory test and the like without a working video out is unwieldy. There's a full 16x bus coming out of the G CPU anyway. Cezanne has Gen3 instead of Gen4 but that's OK.
@@SianaGearz Not sure what I was thinking that long ago, I do know g parts and their relatives such as the 5500 don't have pcie-4 that you need for lower end gpu's that only have pciex8 interfaces as they can and will bottleneck on pcie3. Also there are smaller caches on those cpus, space being needed for the igpu, also you loose pcie lanes , 4 lanes IIRC. The g parts are all around weaker prior to zen4. Again I don't recall what prompted me to make that observations, but those are the weaknesses of the g parts.
@@kaseyboles30 See it's not a computer, it's a test rig for GPUs, to see whether they have hardware issues. It'll do fine with a 2200G, which is not a CPU one would recommend for any other use except light office use.
You can not bottleneck Furmark.
FWIW my actual computer is a Cezanne, and i know i'm not going to have x4 or x8 narrow bus GPUs in here, the price difference between 5500 and 5600 was immense, it was almost half the price, and cheaper than 3600 and usually faster, and it's still perfectly usable several years later. You do not lose any PCIe lanes.
Bro, let me just tell you bro... You are good at what you do. Your nonchalant defiance against - N "big daddy" vidia (Nvidia 4 u slows out-there) is indicative of a degree of confidence that portray those whom have mastered there craft or atleast close to it. Like you are Tony Montana and Nvidia are the cockroach...'s.
You seldom go into technical explanation of what it is that you are doing on video and move on to the next clip/cut/shot/angle (what ever you call it) and then queue music montage, and I get it, out of respect for the people you do not want to be that guy who preaches at the people, a "mr know it all".
But bro, let me tell you bro... This is the first video in which I see you, not only go into a technical discussion on what it is you are doing, but you are also fkn dope azz fk at it!!! I am oneof those who can barely identify a resistor and never even seen what it is that you do before, and I found myself hanging on what you were saying and I realized I was fkn learning...
. ¡¡¡Fkn ey, great video!!! .
Try the amd board with less memory slot populated.
is this the first time we've seen sunlight in this room?😎😎
videos r always fire ❤
0:46 just do hawk tuah on that thing lol
legend. thats all
Rewatching this video gave me both doubts on if I can fix one of the cards I have sitting aorund and a question.
I've got a 3070 that gives error 43 (AMD build), but it also has a memory error so hearing the error 43 on AMD builds makes me wonder if fixing the memory issue would still have the same result.
Spit??
i had a problem on one intel board after replacing graphic card slot. stuck in x1 after. never figure out if that came from changing a slot or something else in a process gets damaged.
I have a Vega 56 that's artifacting when the VRAM is over 740MHz (stock value is 800MHz). Is it fixable somehow?
I have the exact same card and after cleaning the card and re-pasting mosfet 6354 just catched on fire, any ideas why?