@@xXxJakobxXx3 It's buildzoid. He probably killed it with a shunt mod and like 1000W
2 роки тому+35
Sounds like a good video! I did a EPROM swap on a PC NVIDIA Quadro K5000 to a bigger one so it would fit the Mac version of the bios, we burned the eprom first then I replaced the old eprom with the PC bios with the new chip with the Mac version and that was easy with some hot air and amtech flux. The Mac was happy as ever and worked as intended with the new bios and eprom. Would be cool with a colab between you an Jay and people can see some rework in action!
The bios chip actually worked fine, at least on 3 of it's connectors to the motherboard, you were able to successfully program it. The motherboard has a problem reading it, which is a different pin. You might be able to fix it with a reflow or redo the soldering properly but you'll have to find the correct one. Or you could find a donor card that's broken because of some other issue and swap chips. In either case, you'll have to resolder the bios chip to fix it.
QSPI nor flash devices use the same pins for both reading and writing. You have chip select, clock, and up to four data lines. I also reckon that the flash tool does a verify after programming so the fact that the flash process worked tells me that the flash device itself is fine
@@Numfuddle this is honestly a bit of a conundrum I’m assuming Jay doesn’t want to open it up and have to solder anything but aside from replacement of parts I don’t know what else could be done
@@angelicsailor1st QSPI devices can use 1, 2 or 4 data lines. The more data lines you have the higher the read speed. (there’s also octa flash which uses up to eight but they are not normally used for uefi applications). Quad devices (4 data lines) can also use dual or single SPI modes for backwards compatibility. So what could have happened is that the flash tool only uses single or dual mode to program and verify, while the actual bios load on power up uses quad mode for speed. In that case it could actually be the case that some of the data lines are not connected properly and reading the bios on boot fails.
@@Numfuddle Maybe the endpoint controller handing SPI program-verify runs at slower speeds than the 120-165MT/s while passing data to the CPU/RAM at boot and only has problems at those higher speeds.
Board level repair for a bios chip is pretty straight forward for anyone with board level repair skills. Usually the troubleshooting is the tuff part, trying to determine the failed component, and you've already done that.
"Adamant IT" on youtube has several videos showing where you can replace a bios chip if you have the gear (micro soldering, chip flashing device, etc), that would make a fun 'watch jay learn how to do this' video
I definitely get a hold of the manufacturer let him know what happened and just see if an RMA is an option it would at least be good info for anybody else running that card
Power colors rma service is terrible. They wouldn’t except mine when I tried because they kept saying thermals were within spec despite the 110+ degrees the card was running at stock. Trying to fix it myself only slightly improved the thermals but still made the card a terrible choice when it benched lower then a 6800xt thanks to the throttling.
@@BuckieJr I believe ya. PowerColor has been trash as long as I can remember. They've always been the cheapest cards and... suddenly they start throwing out these "SUPER DEVIL SATAN FIRE POWER" cards. I'm just like "what?" You've always been the bottom of the barrel brand. You've always had cards that overheat. You've always had cards that die. What are you doing, PowerColor? Edit: Had to add that even the name is stupid. Power.. color? What kind of Taiwanese mistranslation that never got fixed is this?
I would love to see what an RMA would look on this card. Would they send you a new one or just replace the BIOS chip? Have you thought about replacing the BIOS chip yourself could but an interesting experiment now that the card is not trustworthy.
Id run the RMA just for giggles n haha's. See what they do about it incase this happens / happened to other people out there. Be nice to have peace of mind knowing what the issue is as well as knowing if there is recourse for repair / replace.
Even if they bought it on release day its been 1 year, 4 months and 26 days and Red Devil's warranty is 2 years, so its still good, RMA is pretty solid option here.
Flash the quiet bios on the performance bios. You could have some stuck bits that aren’t taking. If it works then flash the performance bios back. Note: This is a problem I run into a lot with PLC’s.
I would LOVE an explanation here as to why. I find this topic very interesting. Do you think it is because NAND flash chips don't have there own "Firmware" to manage the 1's and 0's over time like say an SSD does? In my head that could cause "Bit Rot" or storage degradation as it were. As once the NAND gates start to fail in an SSD it tries to move those out. Why do you think that happens to these chips? Legit curious
@@Frizzy9000 SPI flash chips dont have firmware nor do they need one or have the capabilities to use one because they have no cpu on board.If you get a 128 MBit flash chip you get 128 megabits nothing more nothing less, there is no reserve and you dont need it because you write to it only once.Sometimes flash chips go bad on their own.
@@Frizzy9000 The gates tend to fail on an SSD through constantly being re-written though, not a problems on EPROM as generally its written once and never touched again. In theory that should last for at least 10 years, often lasts longer but of course once in a while will fail randomly. I had an ADSL router years ago that suddenly wiped its firmware, not sure if the recovery was in the ROM or I got lucky and only the OS partition failed, but it did let me re-flash it and came back to life. In fact its not entirely uncommon on consumer routers as there will be a "little" writing going on to store settings you can change. EPROM AFAIK is designed for retention over writes, so you can write it a lot fewer times than an SSD but the data stored on it should usually remain stable for a lot longer.
I ironically had an issue recently myself where my AMD drivers "uninstalled themselves", and when I ran a catlog of the entire system and just kept attempting to recreate the issue, I caught windows 10 itself deleting the drivers and replacing them with the microsoft release of the drivers, **and it would do this during startup!** So I only caught it when I hooked my backup/game server hosting computer to monitor the activity in this one from boot-time... Worse yet, windows defender was putting my screen capture software into "quarantine" there for a long while until I disabled defender and switched to another anti-v. It appears that microsoft caught my rant about it on twitter earlier this year, and hopefully they're still actually on it enough to make sure this isn't something in the OS that's tampering with the card and drivers again.
I'd claim that's a WINDOWS thing, not just AMD. I once(twice actually but that sounds weird) had the issue where there was a random "hardware disconnect" sound and the monitor would turn black, nothing helped. Threw in my old GPU as well and found the primary one SET TO DISABLED IN DEVICE MANAGER. Why?! How?! I feel like a number of issues on/with Windows could be avoided if it didn't touch things for no reason
this happened to my 6900xt red devil. i did a windows update and i say it downloading VGA drivers. it restarted and all my AMD stuff was not working and said i had the wrong version, i had to DDU and reinstall
Gotta get in that registry editor and start putting some 1's and0's in certain places, I wanna know when some company is going to make an OS that doesn't pull the Microsoft crap, I guess a good Linux Distro is our only real hope .. it's getting there maybe by the time my grandkids are on RDNA 22.5 we won't have Microsoft doing that crap
Bad ICs happen. This seems like a situation where you could just RMA the card, but failing that, it's totally possible to replace a BIOS ROM IC. You order a new one from Digi-Key, Mouser, etc. and then replace the dead one (gotta figure out which one is which) with hot air and a preheater (and flux, solder, et cetera, obviously). I'm pretty confident I could do this, if I had the card in my hands.
For you or me maybe it's an obvious choice, but not everybody watching Louis Rossmann's videos every day. Even Jay has a scary video from 2020 when he attempted to solder back a resistor to a Titan. Now imagine how would that technique work with a teeny-tiny SOIC-8 bios chip. :D
I have the Sapphire Nitro + SE and it's been powerful, quiet and efficient - the sales guy at MC recommended the Sapphire card over the PowerColor cards because of quality and reliability, I'm glad I took his advice.
Saphire is making only AMD cards and I believe they learned how to do it properly. I had same card from Saphire and XFX, both RX580 8gb and saphire had better framerates and 3-4 C cooler temperatures in same build. From that expreience I am always going to go with Saphire.
Tech? I am not sure.. I wish I could get a sapphire nitro rx6700xt but alas it’s still inflated pricing, I could buy a 3070 it for what it’s charged meanwhile the qick 6700xt is double digits off msrp.
Even if stuck on a single BIOS, you can use a flash programmer like an EZP 2019 to flash it onto the hardware chip. If it's not the BIOS chip, it's probably a component that initializes the chip, like power-on signal.
Damnit, Jay. As of October 1, 2019, Toshiba Memory is now KIOXIA. Unfortunately, our lineup of consumer flash and storage products are no longer available in the United States, but we will continue to offer support for Toshiba Memory and OCZ SSDs. Here you can research warranty information and get support for your legacy products. We thank you for being a loyal customer all these years. You can still look forward to seeing KIOXIA SSDs in laptops, PCs and data centers everywhere as we continue to uplift the world with “memory.”
It would be really cool to see you play around with 3D printing a new shroud for the cooler on that card. Or possibly just see what ridiculous mods you can do to it that you wouldn't normally do to a GPU for fear of breaking it.
Hey Jay, I own and running the red devil ultimate 6900xt that has literally done the same exact thing. I continued running the card in silent bios and has been about 6 months or so and the card seems to operate just fine from what I can tell.
You should encourage him to try the performance bios copied onto the silent bio side, there could be something else in the card that's only accessed by the performance bios that could be the faulty culprit. He should also see if either of the fans are drawing too much current if the windings in the motor short out they could cause the card to not function when they are attempted to ramp up high
@@dustineverhart4028 he even put out a video saying none of those are ever him. thats a bot or scammer. unless its JUST that persons name, with a check by it (usually) ITS NOT THEM. this is the internet bro everyone is trying to troll you or rob you. trust nothing
Since you flashed the OC BIOS to the Silent switch, I'd be interested to see what would happen if you flashed the Silent BIOS to the OC switch...and then if that worked, possibly flashing them back to the correct switches.
I had the uninstalled driver issue as well. For me it was right after a Windows 11 update and I thankfully haven’t noticed any issues after that. 6900 XTs are wierd.
Flipping the BIOS toggle switch so you can flash the bad BIOS reminds me of when I had a powerfailure updating the flash on an old motherboard. I had another motherboard of the same kind (same CPU socket for an Althlon XP) but back then the BIOS chips were in tiny sockets. So i was able to boot to the other motherboard into DOS, pop out the BIOS chip while the PC was still running, put the corrupted BIOS chip into that motherboard and run a flash utility with a bunch of switches to ignore the fact the bios image wasn't meant for this motherboard. Bing bang boom, I recovered the bios for my old mobo, and both mobos still worked.
When I was a shop tech in the late '90s, early '00s, I did this exact same thing to recover a mainboard from a bad flash. Worked a treat, saved us an RMA.
I had an PCI SATA controller card with boot ROM (square PLCC type). I noticed you could pass pretty much any image to the firmware update utility. It being like £15 worth of card, I removed the chip, soldered a socket on and put the chip back in. Still worked... Used the update utility for that card to restore BIOS chips for years.
@@michaelmiller3012 Yep! It blew my mind when i was reading about what I could do to recover my bios. Can't do this anymore unless you de-solder the chip, or use a clamp and connect an eporm burner to it.
I just finished my first full custom build on Sunday and went to Micro Center to do it. Wouldn't've even gone there if it wasn't for some of your videos!
I never even consider single bios anymore. I'm not a serious overclocker, but I occasionally update bios for my motherboard. Dual bios has saved me at least 3 times in the last 5 years on my motherboard and once on my Sapphire Pulse 5700 XT.
@@mtx33 the first 2 times it just stopped updating. Let it sit for a full 24 hours and it didn't move. The 3rd time a power outage. The Sapphire Pulse, it stopped entirely, but I was less patient and only waited 30 mins before I bricked it on purpose, flipped to the other bios and forced the new bios through dos commands.
Don't overwrite the good BIOS switch setting! I think you should flash the working BIOS image over the broken BIOS ROM and see if that works. I suspect the card itself has a problem and can only operate with one of the BIOS images and not the other one.
@@linusevosonic I think the OC BIOS is perhaps too overly aggressive for some of the chips variences. They didn't seem to have repaired anything on the replacements so it looks like maybe just a vBIOS change.
@@UnhingedSystems I hope the store exchange it for a sapphire or Asus 6900XT. Don’t want a maybe unreliable card which i payed 1300€ for. Never had such problems with any Graphics Card i ever owned.
Having had an expensive GPU die before has made me keep the video card I'm replacing instead of selling it ever since. After buying the 3080 last year, I kept the gtx 1080 even knowing I could have recouped possibly half of the cost of the 3080 by selling it. These cards run so hot that I don't trust their longevity.
Time for me to shill for EVGA like Jay! But seriously, extended warranty, giant cooler. For what they cost the past two years I wasn't going to buy one without both of those.
My Asus Strix LC 6900 xt runs a cool 50-60c in games all maxed out 4k @ 100hz…gotta love liquid cooling. I dropped Red Devil because they run too hot and i had issues like this before as well. I put my trust on Asus Strix this time and so far so good, and with 5 year warranty i got less of a worry
@@TheDaveJoosten Straight up... STRIX is the best. The go overkill on everything. Usually custom PCB's with custom memory layouts and overkill VRMs that can support WAY more power than the card can possibly draw through the 8-pin(s) and pci-e slot. It's the only brand I buy. Never had one.. single.. issue.. I had 6x RX480 8GB (STRIX) mining ETH for years until I sold them near the end of last year. Those VRMs could support around 300 watts. The card can only pull 225 watts max through the pci-e power connector and slot.
Radeon drivers randomly uninstalled themselves previous to the last update on my son's rig with a(n) RX580 8GB - DDU'd and installed the newer driver set and it's been fine. Also Dual/Multi bios is a life saver and I look for it when buying hardware for myself.
There has to be something strange in the universe, because the same thing happened to me and my friend in the same week (two weeks ago). Either was Win Update or Win Update + Graphics Driver messing something up. The driver uninstalled itself and I did DDU and installed it again and I haven't got an issue again. Windows usually install "something" for the GPU. This time there was nothing.
Same. And I'm running an ancient video card. Windows update screwed up GPU big time about 2 weeks ago. Had to reinstall driver and tweak MB bios to make system stable again.
Windows updates are the devil. I have auto update turned off, can unplug my ethernet cord at night and it will still find a way to connect the Wifi(all settings disabled) and download the update while I'm sleeping.
Had this happen recently with my asus TUF 3080ti. I updated the driver to most recent one and did a custom install as per geforce app. After the system reset, it would not POST with mobo showing VGA error led. Such a lucky you posted this video a day after I experienced this problem. I'll try flicking the switch to the other bios on the card. Not sure if I should attempt to reflash the faulty bios...
My initial thought after hearing the bios chip might be the problem I would say to try and RMA it. If they dont want to then I would see if its possible to get the chip replaced and send it in to a reputable electronics repair shop like Luis Rossmans but that they specialize in GPU repair.
Is there still a POST failure if the "Silent" version of the BIOS is loaded onto the "Failed" EPROM instead of the OC version? If it does then possibly something wrong when the OC code increases the clock.
@@japaneserequired6314 depending on lead time and needs, sometimes it might be worth trying to fix things. but if you have no idea what you're doing, at the least find a repair guy that does.
Not really... what is happening is that they are using low-lead or no-lead solder for the boards now and that stuff is worthless to be mild on the subject. They keep on saying "We have fixed the issues with the no lead solder!" but the people I know in the industry say "No, they haven't... they are being bullied by the government to say they have in order to not have to repeal the laws mandating that poor quality solder."
The thing is during a CPU sometimes more than 8 years in service, so all elements should also hold up. My last cpu before the update to a Ryzen worked for 10 years. And then comes today's GPU or mainboard generation that doesn't even last through an update or higher load for a short or long time. Of course 10 years ago the control software wasn't that complex or the structure of the components. However, I expect that a graphics card can operate for at least 5 years without major maintenance work, the mainboard at least as long as there is no update path or the CPU is completely lagging.
11:59 That is actually a serviceable part. Check out AdamantIT here on youtube. ITs a bit more involved and requires soldering and reprogramming or completely replacement of said chip, but it CAN be done.
@@klansimpson Well if there is still warranty - absolutely. but if not? Reflashing or replacing the BIOS chip by a competent repair shop is not that much of a complex operation. Adamant IT does have indeed some very instructive and interesting videos on how to do it. Any well equipped and competent hardware repair shop can do this. The point is, that this car is far from being bricked, it's actually a rather minor fault.
@@catriona_drummond all true. I would hope a 6900XT still has warranty though as its like what sub 2 years old. I would rather run on the silent profile with an overclock, than screw with it honestly as it doesn't stop the card functioning
I have always been an Intel fan boy. Built a gaming PC for my partner and did a full amd system to give it a go. Have had nothing but problems with it. Weird driver issues, stability is all over the show. The problems I've had to fault find and fix on her PC is like nothing I've ever had to do on Intel. Amd have good hardware but seem to lack in software.
When flashing it usually verifies what was flashed is what should be there based on the file (block checksum). This isn't an eeprom issue, the overclock killed the card. Same as some chips reach X Ghz, others don't, you just now have one that doesn't anymore. Only ways to prove this wrong would be to overclock the card on the regular bios or flash the non-OC BIOS on to the OC slot. I will guarantee it won't make the same clock speed it did and be stable anymore.
Be careful when downloading BIOS for your card, even if it's for the same exact card you have, because there are differences in memory chips ( Samsung, Micron, Hynix etc. ) and they use different memory timings that CAN cause problems if for example you have a Micron VRAM GPU but you flash a Hynix VRAM BIOS on it.
@@ryomario90 Yeah - this depends on whether the card was made cheap or not. Because those that have multiple versions of BIOS for different memories, usually have small EPROM. So they cannot fit ALL the data. They often can barely fit the data they need for their own configuration and substantial amount of time is spent to make sure that BIOS is small enough to fit on it. Happened a lot to AMD - given that they were seen as "cheap option". Especially during this period of Bulldozer/Zen/Zen+. Many of X370 mobos have the same problem. Manufacturers cheaped out on chip and quality, because who would buy from AMD? Lots of people as we found out and newer generations of the boards were better. Same happened to Polaris - depending on vendor you might have a version that was 50 cents cheaper to make, by having small EPROM.
@@ryomario90 That's the same exact card I soft bricked on my brother trying to update his bios profile. Downloaded the correct manufacturer model but it bricked, had to do like Jay and switch to the other bios on the card. Never knew why it failed, even tried a second time and same result
This problem plus the issue with your 12900k system makes me think your studio equipment could be adding noise to your power circuits, maybe try the computers on a UPS and give your studio lights, amplifiers etc a dedicated ground.
A failing EEPROM can cause all manner of strangeness. When boards come back from the field here at work, the weirdest failures can be resolved by replacing the E² and re-finalizing the software.
I'm going to play devils advocate here: What if it's the bios switching circuit that is bad and if it only had one bios, then you wouldn't of had this problem. Also i'm curious of if a new eprom chip could be put in the card, components like that are typically pretty cheap.
it allowed him to switch to silent, boot, switch back to the OC, flash and reboot, then sat there drooling. There is a possibility a new bios chip could be a solution but it could have popped a trace somewhere from said chip. Either way RMA would be what I would do with this personally
even if it's a single bios card you can still remove the bios chip and put it into one of them little readers/programming adapters , just need to exercise soldering skills 😂
Either the chip is dying, or the trace(s) going from the chip to the output cracked. Since you were having fps problems and had to juggle the settings to get it to stabilize, but it lost that info on game shutdown or system shutdown, probably is the chip. I used to see problems with my output when the system heated up (traces expanded and the crack shorted out), but in this instance, it doesn't seem to be the problem.
Send the card to Buildzoid so he can replace the eprom (if he can) so he'll have a crazy card to mod until it explodes because he thought it would be cool to run 800 watts through it on liquid nitrogen.
@@Jeroensgambling Well yeah, but if they're pushing out advanced tech that can't take the beating, then it's essentially not mature enough to distribute to the general public. It's pretty absurd that 30 series and 6000 series cards aren't all that common yet there's quite an alarming amount of failures being reported that never happened on the previous generation. To be completely honest, I had a suspicion that failures were going to be pretty common when I heard that 90 skew cards were going to be a thing.
The chip is a GD25Q80CTIGR, available at Digikey for around 50 cents. It's a fairly large 8-SOP package which is relatively easy to solder with a proper hot air station and some flux (I'd recommend Chipquik NC191). Maybe give it a shot? Don't use a hot air gun, though. That'll just kill the board. You need focused heat from a hot air soldering station - and it'd also be good to protect the surrounding area with kapton or aluminium tape. One thing that's a little suspicious to me though is that this particular chip is discontinued... Possibly due to reliability issues?
Replace the chip. I guess it's time for you guys to learn some soldering. On the bright side, that means you get to buy (and play with) a bunch of new toys.. I mean tools.
I finally got an RTX 3080 on Saturday and after less than 24 hours it already blackscreened twice and crashed my system in the end. I don't know exactly what happend, but it's just dead, so I send it back today. Has nothing really to do with this video apart from a failed GPU, but interesting timing still
Depending on what you were doing at the time, could be a PSU issue... My 8 - 10 year old AX1200 (which had previously been overdrawn on a crossfire 7970 setup, pulling 500 watts per card, and made the LED light bulbs in my room flicker, lol) bricked my 3090 because it couldn't handle the voltage spikes going from idle to 1000 FPS in the Dyson Sphere Program menu, lol. Replaced the PSU and the 3090 I got from RMA has been rock solid since. EVGA Warranty FTW, lol
I'd like to see how long this lasts with a re-flashed silent bios, in theory the card is still good besides that one chip, so how long would it last, after drivers and the bios flash would it still run same speed, same performance, for just as long, did you actually add a second life by flipping the switch or is it only a half life?
@@funkysmokee8764 Yeah, that's what I assume will happen, so I don't see a reason to waste the GPU or change it out, just keep using it, I want to see how it lasts or if something else is broken.
I'd like to think that with a card that pricey, and in todays market, they'd still RMA it for you. We've got stock now, but they can't afford to throw away GPU's for one faulty chip. Would be nice to see the RMA route, to see what'd happen and what they have to say about this issue. As others have said. This could be informative, for people that are facing similar issues. Either way, love the content. I've always looked to you, Steve and Linus for info on the best upgrade path for my personal rig, and you always deliver. Safe to say, if you three don't have positive things to say about something, my wallet stays where it is! Keep it coming, you do great work.
I had my MSI 6800xt Gaming Trio X hooked to my 50" Sony X90J and woke it up one day 2 months ago and had all kinds of artifacting. I immediately took the 6800xt out and installed my 5700xt. Upon installing the 5700xt, I had the same problem. Changed the HDMI header on the x90j from port 1 to port 2 and it cured the artifacts. I still have a fear that the 6800xt might have fried the HDMI port of the TV. Luckily I have 4 ports on this TV. I'm also using an 8K HDMI cable, so I didn't cheap out so much on that part either.
It could have been caused by a BitFlip caused by ionising radiation hitting the memory cell. Ionising radiation is omnipresent, and if one ray hits exactly a memory cell, Its value gets changed. EDIT: ok it still not working after flashing it, eradicates the possibility of physical software error. ... yes you've Red this correctly, Physical Software. Because Software is getting stored in physical values of 1 and 0
I've seen this happen on a friends. It was shutting down Adrenaline software every time he started Battlefield 2042. It made the driver not update properly saying that the Windows version was a newer one when it wasn't. Did a fresh install of Windows 10 and drivers. Everything worked until Battlefield 2042 started and it would all go nuts again.
Great content...well as crisis keep rising, one needs to have different streams of income, a well detailed diversified investment portfolio in the financial markets is needed to survive,as well as secure a profitable investment future
You can still recover the corrupted bios even without having an Igpu or a secondary gpu, all you have to do is to swtich the bios back to the corrupted one while you're inside windows and run the vflash utility, then restart and as long as the new bios you put into the eprom is not also corrupted it'll post. At least that worked on the old amd cards back to like r9 280x 290x 380x era. Maybe even the hd 7970 did work the same but I can't remember, I used to flash the bios of amd cards so often back then. Now it's not even worth it because we have more power tool of which we can directly modify the power tables from.
Certain Mobos used to have removeable EEPROM chips, which was super handy if your chip died or your BIOS got corrupted during a bad flash. I got a couple of Asus AM3 boards that were dead from eBay and revived them with EEPROM chips for less than 8 quid, also from eBay.
I can confirm the importance of dual BIOS anything if you can afford it. I shelled a little more for a dual BIOS MB and it has saved me multiple times where I bricked one of the BIOS.
Jay, they don't have multiple BIOS chips, they just connect one of ROM the address lines to the switch to select top or bottom half of the chip. The corruption if it was a single bit could have only knocked out one BIOS image.
I had a similar thing happen, my 3090 and 3060 in the same pc somehow had the Graphics drivers corrupted, it refused to show video when any nvidia card was in the system. To fix this i needed video out, so I plugged in an RX 6600, used DDU to get rid of Nvidia drivers, plugged in 3060, reinstalled nvidia driver, plugged in 3090, everything worked fine afterwards.
Whenever the program says "update" it will check for a newer version, but to put the same bios on you need to find a "restore" option, or command line "write" option. Use the software that came with the vcard. There is warranty that covers the repair then use it. Knowing what happened is important so that when it is repaired and installed the PSU doesn't do the same thing again because you fixed the PSU.
I have that exact card I got mine a year ago at Microcenter open box for $1500, no big issues so far, just Radeon resets to default settings quite alot, so I had to make a save settings file so I can easily recover my custom settings.
RMA/return it and see what they do to fix it. Most people think returning parts result in two outcomes. You get a new part as replacement, or they repair your part and return it to you. The latter gives a more "inferior" impression as a solution due to the possiblity of human error or other factors during repair. On a different note: what mobo is that that has a display screen for boot errors?
Makes me nervous now. Running the Red Devil 6900xt Ultimate. Have had issues from the beginning with random monitors not working ( all 3 are DP ). Had no issues for years with my 1070ti. At one point I couldn't get any video out, hard reset and power cycle didn't help. Swapped back for 1070ti and came right up. Threw the 6900 back in and so far it is working. My first AMD card and so far it isn't instilling any confidence in me.
I remember when I had a powercolor 9800 pro back like 20 years ago, and it had problems too - I don't really have faith in powercolor as a brand, and I'm surprised they're still around.
If you figure out which chip it is, you should be able to get a replacement pretty easily. It's the soldering that would be tricky. It's probably not EEPROM either, most likely flash. EEPROM is pretty tough, FLASH not so much.
RMA no point having a faulty graphics card with a non-working feature would be the best course of action. When I had the EVGA Geforce 690 back in the day it was starting to fail and I had the 10 year extended warranty so was perfect to RMA it and they sent me back a 980 as a replacement.
The one thing about a ROM chip is that a unregulated voltage spike can corrupt the memory and ruin it. You should be able to replace it inexpensively either though a warranty claim or even sending the card back for a replacement. But first see if you can find out why there was a spike (or not) otherwise it will happen again.
Ideally contact support and have them send out a replacement subject to bill you for it if they don't receive the card with issue so the end user doesn't have to deal with down time ranging weeks to months. I'm sure many will not send a card out in advance which sucks and I'd make note of that for future purchases.
Dude, check this out: Yeah the first bios may be fragged, but why? Is it the bios chip it’s self or is it possibly shorting to ground somewhere? Check the caps or other components connected to that chip. If there is a blown cap or even another chip not pushing current like it’s supposed to, it would cause the chip to fail. If there is no such short or faulty components, then it’s definitely the chip it’s self. Either way, you need to disassemble the card and dust off your multimeter and soldering iron.
Been running this same rig. 5950x - 6900xt red devil ultimate edition. for about 8 months. 0 issues, fantastic performance. Both cpu and gpu ran overclocked constantly. AMD ecosystem is fantastic. Adrenalin tool is very helpful. DEFINITELY RMA this card. Would be nice to see PowerColor stand behind their flagship card.
Only thing I can think of to fix it would involve some soldering work, which doesn't quite seem to be up your guys' alley. (Electronic soldering is fairly tricky even when you're working with amateur grade PCBs - the ones where you acid etch off a single layer of copper for your traces...this is....rather significantly more complex than that.)
I had this exact issue with my 6800 XT, where it eventually wouldn't boot regardless of BIOS. Had to RMA it after only 3 weeks of use. Thankfully the company i bought it from got me a new one straight away.
I also have a Red Devil, but the 6700xt... And it's drivers crash like every other day... No matter what BIOS I use. The card is five months old. I contacted PowerColor, they told me to contact the retailer. The retailer told me to contact Powercolor because they couldn't find an issue during their short test that cost me €145... 😮💨 I'm keeping it until I saved up for a new card, because it still kind of works... Half of the time. Funny to see I'm not the only one with this issue
There are tools for flashing BIOS chips outside of a system. I have a CH314a that I got for less than $20 and it saved a motherboard that I thought was just dead. Idk if GPU's use the same BIOS chips as motherboards and sometimes you have to desolder and remove the chip to flash it, but it's worth a look. If you plan on using it regularly, definitely get a better clip, the one it comes with is only good for a use or 2 before it becomes a real pain
Dual-BIOS.... I remember my old-ass Socket 754 Gigabyte motherboard made that a big selling point (I guess they pioneered it at least in the consumer market?). And it did save me hassle a few times... especially with BIOS updates.
An alternative possibly could be malware in the OC bios, maybe crypto, It can mask its location in BIOS by misreporting memory usage, but if the OC bios is close to the size limitation physically of the CMOS, It would get corrupted at the end because it couldn't fit. Thereby locking up your video card when it doesn't find the end of the file. Could be interesting to try a hex dump of the BIOS
It's actually been a long time since I heard of a bios failure short of a flash gone wrong. Back in the day doing PC repairs/troubleshooting my first thought would've been a possible hack or virus affecting initializing or ' boot ' stages. Try another of the exact cards in the affected system without making changes and if it quits you'll know something more than the card is at fault. Also try a quick new build on a test bench with all budget parts. Sometimes I've found some unknown rare glitch in the more expensive hardware is to blame whereas the ' cheap ' system with the suspect parts works fine, with all drivers updated of course. 👍
If GUI doesn't work then command line it. If that fails, RMA it. Do a lotto for the RMA card for the person that chooses the correct date it is returned. Just create a website with dates. First person who fills out information to hold date slot gets that slot.
If you have contacts at AMD, it would be interesting if they could take a look at it and tell you what the error is. They might be interested in finding this out as well for their internal QA.
You could try to swap BIOS chips. Preferably from a dead card but potentially just swap the 2 chips on that card itself. Just to verify that it is the chip and not some traces or something else. Thankfully the chips are usually a bit larger so should be doable to solder it.
Try flashing the same version of the VBIOS that currently works onto the non-working EEPROM. If that doesn't work then I think you're correct that something about the EEPROM is dead and I'd just RMA it. It happens sometimes so might as well let them eat it.
I would try writing the silent BIOS onto the OC chip first. There could be some weird thing happening when the OC bios is on the OC chip. I have seen that happen before. If it doesn't help, I would remove the heatsink and clean the card. I once had an issue on my MacBook where it would work just fine and boot fine but every now and then when it went to sleep, the screen wouldn't wake up properly. Guess what, there was some dust that landed in a spot near a controller chip and was causing the issue. A good clean later worked perfectly 100% of the time.
Yeah Jay I think you should just try to replace the Eprom on that or if you don’t want to risk it I wouldn’t mind trying it myself they sell brand new EPROMs that you can solder in place
![[UA] NAVI vs MOUZ | BO5 | IEM Rio 2024](http://i.ytimg.com/vi/CMBYk0cbGoA/mqdefault.jpg)
I'd probably just replace the BIOS chip. I have a dead 6900XT PCB that should have 2 perfectly good BIOS chips on it.
The man, the myth, the legend.
great idea but don't let Jay do any soldering.
May I ask why you have a dead PCB of this card? Every single one of them is under two years old, why didn’t you ask for a RMA under warranty?
@@xXxJakobxXx3 It's buildzoid. He probably killed it with a shunt mod and like 1000W
Sounds like a good video! I did a EPROM swap on a PC NVIDIA Quadro K5000 to a bigger one so it would fit the Mac version of the bios, we burned the eprom first then I replaced the old eprom with the PC bios with the new chip with the Mac version and that was easy with some hot air and amtech flux.
The Mac was happy as ever and worked as intended with the new bios and eprom.
Would be cool with a colab between you an Jay and people can see some rework in action!
RIP....just in time for the 6950. Coincidence? I THINK NOT! (well played Phil)
Why wait for the 6950 when the 3090Tie is already waiting? 🙃
😂😂😂😂😂😂😂😂😂😂😂😂😂😂😂😂😂
3090ti is an overpriced waste of money
Mined it and next GPU as the electric is free
@@SkateClipsAndTips overpriced waste of energy bills 😆
The bios chip actually worked fine, at least on 3 of it's connectors to the motherboard, you were able to successfully program it. The motherboard has a problem reading it, which is a different pin. You might be able to fix it with a reflow or redo the soldering properly but you'll have to find the correct one. Or you could find a donor card that's broken because of some other issue and swap chips. In either case, you'll have to resolder the bios chip to fix it.
QSPI nor flash devices use the same pins for both reading and writing. You have chip select, clock, and up to four data lines. I also reckon that the flash tool does a verify after programming so the fact that the flash process worked tells me that the flash device itself is fine
@@Numfuddle Correct, they are not using dual/triple ported flash chips, if such things even exist.
@@Numfuddle this is honestly a bit of a conundrum I’m assuming Jay doesn’t want to open it up and have to solder anything but aside from replacement of parts I don’t know what else could be done
@@angelicsailor1st QSPI devices can use 1, 2 or 4 data lines. The more data lines you have the higher the read speed. (there’s also octa flash which uses up to eight but they are not normally used for uefi applications). Quad devices (4 data lines) can also use dual or single SPI modes for backwards compatibility. So what could have happened is that the flash tool only uses single or dual mode to program and verify, while the actual bios load on power up uses quad mode for speed. In that case it could actually be the case that some of the data lines are not connected properly and reading the bios on boot fails.
@@Numfuddle Maybe the endpoint controller handing SPI program-verify runs at slower speeds than the 120-165MT/s while passing data to the CPU/RAM at boot and only has problems at those higher speeds.
Board level repair for a bios chip is pretty straight forward for anyone with board level repair skills. Usually the troubleshooting is the tuff part, trying to determine the failed component, and you've already done that.
"Adamant IT" on youtube has several videos showing where you can replace a bios chip if you have the gear (micro soldering, chip flashing device, etc), that would make a fun 'watch jay learn how to do this' video
I definitely get a hold of the manufacturer let him know what happened and just see if an RMA is an option it would at least be good info for anybody else running that card
Power colors rma service is terrible. They wouldn’t except mine when I tried because they kept saying thermals were within spec despite the 110+ degrees the card was running at stock. Trying to fix it myself only slightly improved the thermals but still made the card a terrible choice when it benched lower then a 6800xt thanks to the throttling.
@@BuckieJr I believe ya. PowerColor has been trash as long as I can remember. They've always been the cheapest cards and... suddenly they start throwing out these "SUPER DEVIL SATAN FIRE POWER" cards. I'm just like "what?" You've always been the bottom of the barrel brand. You've always had cards that overheat. You've always had cards that die. What are you doing, PowerColor?
Edit: Had to add that even the name is stupid. Power.. color? What kind of Taiwanese mistranslation that never got fixed is this?
@@JustAGuy85 my red Devil 5700xt was a champ when I had it, was the whole reason I grabbed the 6900xt version. :(
I would love to see what an RMA would look on this card. Would they send you a new one or just replace the BIOS chip? Have you thought about replacing the BIOS chip yourself could but an interesting experiment now that the card is not trustworthy.
They'll most likely replace it outright and fix the bad one to sell later.
Hard to RMA something you didn’t buy
@@xbhollandx alot review card DOA or die soon after power up lol
They don't waste time to replace chips or whatever, they have some cards for rma and replace it with a new for sure
It would be interesting to actually see Power Color's RMA process, like if they'll actually stand behind their products, or try to pass the buck.
Id run the RMA just for giggles n haha's. See what they do about it incase this happens / happened to other people out there. Be nice to have peace of mind knowing what the issue is as well as knowing if there is recourse for repair / replace.
Even if they bought it on release day its been 1 year, 4 months and 26 days and Red Devil's warranty is 2 years, so its still good, RMA is pretty solid option here.
Oh man! He could do a collab with steve on that!
Would be cool if they actually told him what was wrong if he chooses to send it in so we have closure. I can't live like this not knowing.
@@markthestark1 last I checked since they repasted the card it's voided the Warrenty
@@Alucard-gt1zf even if you do, sometimes they're nice and still take them. Most of them know the paste they use is trash and needs to be replaced.
Flash the quiet bios on the performance bios. You could have some stuck bits that aren’t taking. If it works then flash the performance bios back.
Note: This is a problem I run into a lot with PLC’s.
I would LOVE an explanation here as to why. I find this topic very interesting.
Do you think it is because NAND flash chips don't have there own "Firmware" to manage the 1's and 0's over time like say an SSD does?
In my head that could cause "Bit Rot" or storage degradation as it were. As once the NAND gates start to fail in an SSD it tries to move those out.
Why do you think that happens to these chips? Legit curious
I had to do this on a motherboard
Shouldn't a VBIOS flasher do a checksum against the provided ROM and the copied ROM? A stuck bit would give a different result.
@@Frizzy9000 SPI flash chips dont have firmware nor do they need one or have the capabilities to use one because they have no cpu on board.If you get a 128 MBit flash chip you get 128 megabits nothing more nothing less, there is no reserve and you dont need it because you write to it only once.Sometimes flash chips go bad on their own.
@@Frizzy9000 The gates tend to fail on an SSD through constantly being re-written though, not a problems on EPROM as generally its written once and never touched again. In theory that should last for at least 10 years, often lasts longer but of course once in a while will fail randomly.
I had an ADSL router years ago that suddenly wiped its firmware, not sure if the recovery was in the ROM or I got lucky and only the OS partition failed, but it did let me re-flash it and came back to life. In fact its not entirely uncommon on consumer routers as there will be a "little" writing going on to store settings you can change.
EPROM AFAIK is designed for retention over writes, so you can write it a lot fewer times than an SSD but the data stored on it should usually remain stable for a lot longer.
I ironically had an issue recently myself where my AMD drivers "uninstalled themselves", and when I ran a catlog of the entire system and just kept attempting to recreate the issue, I caught windows 10 itself deleting the drivers and replacing them with the microsoft release of the drivers, **and it would do this during startup!** So I only caught it when I hooked my backup/game server hosting computer to monitor the activity in this one from boot-time... Worse yet, windows defender was putting my screen capture software into "quarantine" there for a long while until I disabled defender and switched to another anti-v.
It appears that microsoft caught my rant about it on twitter earlier this year, and hopefully they're still actually on it enough to make sure this isn't something in the OS that's tampering with the card and drivers again.
That's a known Microshaft thing.
You have to go into the optionals in winblows update and turn the option to update the driver for the card off.
I'd claim that's a WINDOWS thing, not just AMD.
I once(twice actually but that sounds weird) had the issue where there was a random "hardware disconnect" sound and the monitor would turn black, nothing helped.
Threw in my old GPU as well and found the primary one SET TO DISABLED IN DEVICE MANAGER. Why?! How?!
I feel like a number of issues on/with Windows could be avoided if it didn't touch things for no reason
this happened to my 6900xt red devil. i did a windows update and i say it downloading VGA drivers. it restarted and all my AMD stuff was not working and said i had the wrong version, i had to DDU and reinstall
Oh you too? I almost went back to Linux after that one
Gotta get in that registry editor and start putting some 1's and0's in certain places, I wanna know when some company is going to make an OS that doesn't pull the Microsoft crap, I guess a good Linux Distro is our only real hope .. it's getting there maybe by the time my grandkids are on RDNA 22.5 we won't have Microsoft doing that crap
Bad ICs happen. This seems like a situation where you could just RMA the card, but failing that, it's totally possible to replace a BIOS ROM IC. You order a new one from Digi-Key, Mouser, etc. and then replace the dead one (gotta figure out which one is which) with hot air and a preheater (and flux, solder, et cetera, obviously).
I'm pretty confident I could do this, if I had the card in my hands.
whats ic
@@dreamxxx333mmm2 IC = integrated circuit, or a "chip" in common parlance.
For you or me maybe it's an obvious choice, but not everybody watching Louis Rossmann's videos every day. Even Jay has a scary video from 2020 when he attempted to solder back a resistor to a Titan. Now imagine how would that technique work with a teeny-tiny SOIC-8 bios chip. :D
If the utility doesn't work, use a ch341a programmer. Brought back my bricked 5700xt, sooooooo good.
I have the Sapphire Nitro + SE and it's been powerful, quiet and efficient - the sales guy at MC recommended the Sapphire card over the PowerColor cards because of quality and reliability, I'm glad I took his advice.
Sapphire apparently stepped up their game. Their R390X tended to burn itself to death close to warranty end.
Saphire is making only AMD cards and I believe they learned how to do it properly. I had same card from Saphire and XFX, both RX580 8gb and saphire had better framerates and 3-4 C cooler temperatures in same build. From that expreience I am always going to go with Saphire.
I still have a 5700XT Sapphire Nitro + and it's really good and had no problems with it ! Upgrading to a 6800XT by end of year :)
Tech? I am not sure.. I wish I could get a sapphire nitro rx6700xt but alas it’s still inflated pricing, I could buy a 3070 it for what it’s charged meanwhile the qick 6700xt is double digits off msrp.
Even if stuck on a single BIOS, you can use a flash programmer like an EZP 2019 to flash it onto the hardware chip. If it's not the BIOS chip, it's probably a component that initializes the chip, like power-on signal.
Damnit, Jay.
As of October 1, 2019, Toshiba Memory is now KIOXIA. Unfortunately, our lineup of consumer flash and storage products are no longer available in the United States, but we will continue to offer support for Toshiba Memory and OCZ SSDs. Here you can research warranty information and get support for your legacy products.
We thank you for being a loyal customer all these years. You can still look forward to seeing KIOXIA SSDs in laptops, PCs and data centers everywhere as we continue to uplift the world with “memory.”
It would be really cool to see you play around with 3D printing a new shroud for the cooler on that card. Or possibly just see what ridiculous mods you can do to it that you wouldn't normally do to a GPU for fear of breaking it.
Oh no! Phil killed his system while editing! Even computer systems could handle Phil's editing skills and demands.
Hey Jay, I own and running the red devil ultimate 6900xt that has literally done the same exact thing. I continued running the card in silent bios and has been about 6 months or so and the card seems to operate just fine from what I can tell.
You should encourage him to try the performance bios copied onto the silent bio side, there could be something else in the card that's only accessed by the performance bios that could be the faulty culprit. He should also see if either of the fans are drawing too much current if the windings in the motor short out they could cause the card to not function when they are attempted to ramp up high
@♜ Pìnned by JayzTwoCents I hope this wasn't really you?
@@dustineverhart4028 he even put out a video saying none of those are ever him. thats a bot or scammer. unless its JUST that persons name, with a check by it (usually) ITS NOT THEM. this is the internet bro everyone is trying to troll you or rob you. trust nothing
@@dustineverhart4028 That is scam, they are all over the place. Report and ignore.
@@dustineverhart4028 it's a spam account.. report it
Since you flashed the OC BIOS to the Silent switch, I'd be interested to see what would happen if you flashed the Silent BIOS to the OC switch...and then if that worked, possibly flashing them back to the correct switches.
I was going to suggest flashing the Silent BIOS to the faulty one and see if it makes a difference.
I had the uninstalled driver issue as well. For me it was right after a Windows 11 update and I thankfully haven’t noticed any issues after that.
6900 XTs are wierd.
Flipping the BIOS toggle switch so you can flash the bad BIOS reminds me of when I had a powerfailure updating the flash on an old motherboard. I had another motherboard of the same kind (same CPU socket for an Althlon XP) but back then the BIOS chips were in tiny sockets. So i was able to boot to the other motherboard into DOS, pop out the BIOS chip while the PC was still running, put the corrupted BIOS chip into that motherboard and run a flash utility with a bunch of switches to ignore the fact the bios image wasn't meant for this motherboard. Bing bang boom, I recovered the bios for my old mobo, and both mobos still worked.
When I was a shop tech in the late '90s, early '00s, I did this exact same thing to recover a mainboard from a bad flash. Worked a treat, saved us an RMA.
I had an PCI SATA controller card with boot ROM (square PLCC type). I noticed you could pass pretty much any image to the firmware update utility.
It being like £15 worth of card, I removed the chip, soldered a socket on and put the chip back in. Still worked...
Used the update utility for that card to restore BIOS chips for years.
@@michaelmiller3012 Yep! It blew my mind when i was reading about what I could do to recover my bios. Can't do this anymore unless you de-solder the chip, or use a clamp and connect an eporm burner to it.
I just finished my first full custom build on Sunday and went to Micro Center to do it. Wouldn't've even gone there if it wasn't for some of your videos!
I never even consider single bios anymore. I'm not a serious overclocker, but I occasionally update bios for my motherboard. Dual bios has saved me at least 3 times in the last 5 years on my motherboard and once on my Sapphire Pulse 5700 XT.
Honestly it's a scary number. But how exactly gone wrong that many update? did you tweaked the bios files or how?
@@mtx33 the first 2 times it just stopped updating. Let it sit for a full 24 hours and it didn't move. The 3rd time a power outage.
The Sapphire Pulse, it stopped entirely, but I was less patient and only waited 30 mins before I bricked it on purpose, flipped to the other bios and forced the new bios through dos commands.
Don't overwrite the good BIOS switch setting!
I think you should flash the working BIOS image over the broken BIOS ROM and see if that works. I suspect the card itself has a problem and can only operate with one of the BIOS images and not the other one.
I've had a few issues with their OC BIOS/Dual BIOS. Two different units sent for RMA where it would crash mid-game. The replacements were good so far.
Same problems here, i sent it back tomorrow.
@@linusevosonic I think the OC BIOS is perhaps too overly aggressive for some of the chips variences. They didn't seem to have repaired anything on the replacements so it looks like maybe just a vBIOS change.
@@UnhingedSystems I hope the store exchange it for a sapphire or Asus 6900XT. Don’t want a maybe unreliable card which i payed 1300€ for. Never had such problems with any Graphics Card i ever owned.
@@UnhingedSystems They didnt’t exchange the card and now it even doesn’t switch on the minidisplayports.
Having had an expensive GPU die before has made me keep the video card I'm replacing instead of selling it ever since. After buying the 3080 last year, I kept the gtx 1080 even knowing I could have recouped possibly half of the cost of the 3080 by selling it. These cards run so hot that I don't trust their longevity.
Time for me to shill for EVGA like Jay!
But seriously, extended warranty, giant cooler. For what they cost the past two years I wasn't going to buy one without both of those.
(what model do you have that runs hot and have you considered undervolting?)
My Asus Strix LC 6900 xt runs a cool 50-60c in games all maxed out 4k @ 100hz…gotta love liquid cooling. I dropped Red Devil because they run too hot and i had issues like this before as well. I put my trust on Asus Strix this time and so far so good, and with 5 year warranty i got less of a worry
I have the same 6900XT (ultimate edition). Max 65C under 100% load on stock fan curve. They aren't hot. This is bad luck
@@TheDaveJoosten Straight up... STRIX is the best. The go overkill on everything. Usually custom PCB's with custom memory layouts and overkill VRMs that can support WAY more power than the card can possibly draw through the 8-pin(s) and pci-e slot. It's the only brand I buy. Never had one.. single.. issue.. I had 6x RX480 8GB (STRIX) mining ETH for years until I sold them near the end of last year. Those VRMs could support around 300 watts. The card can only pull 225 watts max through the pci-e power connector and slot.
Radeon drivers randomly uninstalled themselves previous to the last update on my son's rig with a(n) RX580 8GB - DDU'd and installed the newer driver set and it's been fine. Also Dual/Multi bios is a life saver and I look for it when buying hardware for myself.
There has to be something strange in the universe, because the same thing happened to me and my friend in the same week (two weeks ago). Either was Win Update or Win Update + Graphics Driver messing something up. The driver uninstalled itself and I did DDU and installed it again and I haven't got an issue again. Windows usually install "something" for the GPU. This time there was nothing.
Same. And I'm running an ancient video card. Windows update screwed up GPU big time about 2 weeks ago. Had to reinstall driver and tweak MB bios to make system stable again.
i had the same thing happen at the same time as you humm
Windows updates are the devil. I have auto update turned off, can unplug my ethernet cord at night and it will still find a way to connect the Wifi(all settings disabled) and download the update while I'm sleeping.
@@gotdangedcommiesitellyahwa6298 set your wifi as a metered connection, it wont try to sneak as much by
@@bradhaines3142 Thanks! I'll try that. Would just take the card out, but then I'd lose bluetooth.
Had this happen recently with my asus TUF 3080ti. I updated the driver to most recent one and did a custom install as per geforce app.
After the system reset, it would not POST with mobo showing VGA error led.
Such a lucky you posted this video a day after I experienced this problem. I'll try flicking the switch to the other bios on the card. Not sure if I should attempt to reflash the faulty bios...
My initial thought after hearing the bios chip might be the problem I would say to try and RMA it. If they dont want to then I would see if its possible to get the chip replaced and send it in to a reputable electronics repair shop like Luis Rossmans but that they specialize in GPU repair.
Love the videos, Jay and Phil. Thanks for being around and keeping the videos going!
With Phil's editing wizardry, I'm surprised the 6900XT lasted this long. 🤷
Phil's editing is more like the weirding way. Chops up a video by had, then splices in the spice.
Is there still a POST failure if the "Silent" version of the BIOS is loaded onto the "Failed" EPROM instead of the OC version? If it does then possibly something wrong when the OC code increases the clock.
Good idea. He should not overwrite the good ROM in case it's the image that's the problem not the ROM.
@@wayland7150 smart! - i'd try that first
Idk if I want to start messing with reloading files in case it is something else that is the issue because it works fine now
The glow accent was beautiful. It's like glowing hot pipes....
Also if it's a physical chip problem, maybe try resoldering/cook it?
Probably replacing the faulty BIOS-Chip might be an idea.
@@japaneserequired6314 depending on lead time and needs, sometimes it might be worth trying to fix things. but if you have no idea what you're doing, at the least find a repair guy that does.
RMA and let manufacturer fix, might take some time but Jay has a shelf of spare graphics cards.
@@japaneserequired6314 lol it really is a DIY thing though
Modern GPUs, CPUs, motherboards and other parts in PC have become so fragile.
My Corsair VIII Dark hero came faulty in the mail twice.
Not really... what is happening is that they are using low-lead or no-lead solder for the boards now and that stuff is worthless to be mild on the subject.
They keep on saying "We have fixed the issues with the no lead solder!" but the people I know in the industry say "No, they haven't... they are being bullied by the government to say they have in order to not have to repeal the laws mandating that poor quality solder."
The thing is during a CPU sometimes more than 8 years in service, so all elements should also hold up.
My last cpu before the update to a Ryzen worked for 10 years.
And then comes today's GPU or mainboard generation that doesn't even last through an update or higher load for a short or long time.
Of course 10 years ago the control software wasn't that complex or the structure of the components. However, I expect that a graphics card can operate for at least 5 years without major maintenance work, the mainboard at least as long as there is no update path or the CPU is completely lagging.
11:59 That is actually a serviceable part. Check out AdamantIT here on youtube. ITs a bit more involved and requires soldering and reprogramming or completely replacement of said chip, but it CAN be done.
would rather warranty return personally than kill your warranty rights. I would hope Jay has it sent back too
@@klansimpson Well if there is still warranty - absolutely. but if not? Reflashing or replacing the BIOS chip by a competent repair shop is not that much of a complex operation. Adamant IT does have indeed some very instructive and interesting videos on how to do it. Any well equipped and competent hardware repair shop can do this.
The point is, that this car is far from being bricked, it's actually a rather minor fault.
@@catriona_drummond all true. I would hope a 6900XT still has warranty though as its like what sub 2 years old. I would rather run on the silent profile with an overclock, than screw with it honestly as it doesn't stop the card functioning
thank god they made multiple bios switches, now I understand. I was always thinking that that switch just adjusts fan curve and that's it
@@denissh9478 depends on the specific card, some of those are just that.
I have always been an Intel fan boy. Built a gaming PC for my partner and did a full amd system to give it a go. Have had nothing but problems with it. Weird driver issues, stability is all over the show. The problems I've had to fault find and fix on her PC is like nothing I've ever had to do on Intel. Amd have good hardware but seem to lack in software.
Another strong possibility is the original BIOS is corrupt and the new image is also incorrect.
When flashing it usually verifies what was flashed is what should be there based on the file (block checksum). This isn't an eeprom issue, the overclock killed the card. Same as some chips reach X Ghz, others don't, you just now have one that doesn't anymore. Only ways to prove this wrong would be to overclock the card on the regular bios or flash the non-OC BIOS on to the OC slot. I will guarantee it won't make the same clock speed it did and be stable anymore.
Be careful when downloading BIOS for your card, even if it's for the same exact card you have, because there are differences in memory chips ( Samsung, Micron, Hynix etc. ) and they use different memory timings that CAN cause problems if for example you have a Micron VRAM GPU but you flash a Hynix VRAM BIOS on it.
Modern vBIOSes support all currently available VRAM vendors. It's not like there's a vBIOS for each one.
@@-eMpTy- My GPU has 3 different VRAM BIOS all 3 of them with different VRAM timings, but then again not sure if the RX 580 can be called a modern GPU
@@ryomario90 Yeah - this depends on whether the card was made cheap or not. Because those that have multiple versions of BIOS for different memories, usually have small EPROM. So they cannot fit ALL the data. They often can barely fit the data they need for their own configuration and substantial amount of time is spent to make sure that BIOS is small enough to fit on it. Happened a lot to AMD - given that they were seen as "cheap option". Especially during this period of Bulldozer/Zen/Zen+. Many of X370 mobos have the same problem. Manufacturers cheaped out on chip and quality, because who would buy from AMD? Lots of people as we found out and newer generations of the boards were better. Same happened to Polaris - depending on vendor you might have a version that was 50 cents cheaper to make, by having small EPROM.
@@ryomario90 That's the same exact card I soft bricked on my brother trying to update his bios profile. Downloaded the correct manufacturer model but it bricked, had to do like Jay and switch to the other bios on the card.
Never knew why it failed, even tried a second time and same result
My Xfx speedster 6900xt Zero edition has been doing great. Knock on wood.
This problem plus the issue with your 12900k system makes me think your studio equipment could be adding noise to your power circuits, maybe try the computers on a UPS and give your studio lights, amplifiers etc a dedicated ground.
A failing EEPROM can cause all manner of strangeness. When boards come back from the field here at work, the weirdest failures can be resolved by replacing the E² and re-finalizing the software.
I'm going to play devils advocate here: What if it's the bios switching circuit that is bad and if it only had one bios, then you wouldn't of had this problem.
Also i'm curious of if a new eprom chip could be put in the card, components like that are typically pretty cheap.
It's just simple rerouting on which chip to read, it's not anything fancy.
@@Mysteoa The bios switching circuitry fails a lot actually, bios basically never corrupt by itself unless you messed with it.
@@Mysteoa yes, but if for some reason the switch isn't properly making connection, you could have problems with bios?
it allowed him to switch to silent, boot, switch back to the OC, flash and reboot, then sat there drooling. There is a possibility a new bios chip could be a solution but it could have popped a trace somewhere from said chip. Either way RMA would be what I would do with this personally
even if it's a single bios card you can still remove the bios chip and put it into one of them little readers/programming adapters , just need to exercise soldering skills 😂
It should work. But vgas normally use bga format chips, so without an adapter I doubt that he will be reprograming the chip again.
Either the chip is dying, or the trace(s) going from the chip to the output cracked. Since you were having fps problems and had to juggle the settings to get it to stabilize, but it lost that info on game shutdown or system shutdown, probably is the chip. I used to see problems with my output when the system heated up (traces expanded and the crack shorted out), but in this instance, it doesn't seem to be the problem.
"F*ck it, we'll do it live!"
That is an EPIC reference. LMFAO!!!
Trouble shoot the problem until you find the fault, the suspense is great and nice to know what the REAL problem is.
My 6900XT has given me nothing but problems for the past 8ish months. I'm on my 4th RMA now.
Yikes sorry to hear that
Sounds like 6900XTs are the faulty ones,my 6800XT never gave me any problems and i never heard them causing problems either,its always the 6900XT
Some friends had problem with ther 6900 XT...
I own a 6700 XT, never had any issues
Bullshit dude.
Which model? I haven’t even had a sniff from mine (that wasn’t entirely my fault that is lol)
Send the card to Buildzoid so he can replace the eprom (if he can) so he'll have a crazy card to mod until it explodes because he thought it would be cool to run 800 watts through it on liquid nitrogen.
I'm getting concerned at the death rate of all the latest GPU's.
Mind you it is a high-end GPU with proberly quite some power flowing through it. Lots of things can happen. They just get more complex over the years.
@@Jeroensgambling Well yeah, but if they're pushing out advanced tech that can't take the beating, then it's essentially not mature enough to distribute to the general public. It's pretty absurd that 30 series and 6000 series cards aren't all that common yet there's quite an alarming amount of failures being reported that never happened on the previous generation. To be completely honest, I had a suspicion that failures were going to be pretty common when I heard that 90 skew cards were going to be a thing.
@@thunderbolt10031 what does 90 skew mean?
@@kire929 The 90 skew, so RTX 3090 or RX 6900.
The chip is a GD25Q80CTIGR, available at Digikey for around 50 cents. It's a fairly large 8-SOP package which is relatively easy to solder with a proper hot air station and some flux (I'd recommend Chipquik NC191). Maybe give it a shot? Don't use a hot air gun, though. That'll just kill the board. You need focused heat from a hot air soldering station - and it'd also be good to protect the surrounding area with kapton or aluminium tape.
One thing that's a little suspicious to me though is that this particular chip is discontinued... Possibly due to reliability issues?
Replace the chip. I guess it's time for you guys to learn some soldering. On the bright side, that means you get to buy (and play with) a bunch of new toys.. I mean tools.
I finally got an RTX 3080 on Saturday and after less than 24 hours it already blackscreened twice and crashed my system in the end. I don't know exactly what happend, but it's just dead, so I send it back today. Has nothing really to do with this video apart from a failed GPU, but interesting timing still
Depending on what you were doing at the time, could be a PSU issue... My 8 - 10 year old AX1200 (which had previously been overdrawn on a crossfire 7970 setup, pulling 500 watts per card, and made the LED light bulbs in my room flicker, lol) bricked my 3090 because it couldn't handle the voltage spikes going from idle to 1000 FPS in the Dyson Sphere Program menu, lol. Replaced the PSU and the 3090 I got from RMA has been rock solid since. EVGA Warranty FTW, lol
I'd like to see how long this lasts with a re-flashed silent bios, in theory the card is still good besides that one chip, so how long would it last, after drivers and the bios flash would it still run same speed, same performance, for just as long, did you actually add a second life by flipping the switch or is it only a half life?
it is literally the same, since he flashed a normal bios on the silent bios eeprom. so it runs like it ran on the original eeprom that died.
@@funkysmokee8764 Yeah, that's what I assume will happen, so I don't see a reason to waste the GPU or change it out, just keep using it, I want to see how it lasts or if something else is broken.
I'd like to think that with a card that pricey, and in todays market, they'd still RMA it for you. We've got stock now, but they can't afford to throw away GPU's for one faulty chip.
Would be nice to see the RMA route, to see what'd happen and what they have to say about this issue.
As others have said. This could be informative, for people that are facing similar issues.
Either way, love the content. I've always looked to you, Steve and Linus for info on the best upgrade path for my personal rig, and you always deliver. Safe to say, if you three don't have positive things to say about something, my wallet stays where it is!
Keep it coming, you do great work.
Thanks Jay you have been a really good inspiration, and watching your vids has taught me a lot. Please keep posting and thank you.
I had my MSI 6800xt Gaming Trio X hooked to my 50" Sony X90J and woke it up one day 2 months ago and had all kinds of artifacting. I immediately took the 6800xt out and installed my 5700xt. Upon installing the 5700xt, I had the same problem. Changed the HDMI header on the x90j from port 1 to port 2 and it cured the artifacts.
I still have a fear that the 6800xt might have fried the HDMI port of the TV. Luckily I have 4 ports on this TV. I'm also using an 8K HDMI cable, so I didn't cheap out so much on that part either.
It could have been caused by a BitFlip caused by ionising radiation hitting the memory cell.
Ionising radiation is omnipresent, and if one ray hits exactly a memory cell, Its value gets changed.
EDIT: ok it still not working after flashing it, eradicates the possibility of physical software error.
... yes you've Red this correctly, Physical Software. Because Software is getting stored in physical values of 1 and 0
Then programming BIOS would fix it, but it didn't.
@@duskyechelon did you read the edit ?
Has Phil run a malware scan? The missing driver makes me ask. Also, maybe check his boot drive for errors?
I've seen this happen on a friends. It was shutting down Adrenaline software every time he started Battlefield 2042. It made the driver not update properly saying that the Windows version was a newer one when it wasn't. Did a fresh install of Windows 10 and drivers. Everything worked until Battlefield 2042 started and it would all go nuts again.
Great content...well as crisis keep rising, one needs to have different streams of income, a well detailed diversified investment portfolio in the financial markets is needed to survive,as well as secure a profitable investment future
Expert Amanda Hernandez never seems to surprise me i also trade with her, Her strategies are top notch I swear
my life is totally changed because i've been earning $12,250 returns from my $2,000 investment with Amanda
Please how can i contact Mrs. Amanda Hernandez? I want to invest with her too.
@Bill Schecter Thanks a lot, I'll send her a message right away...
i started with 2 grands and after 2week i received a returns of 6 grands then i continue with her ever since She has been delivering☺️☺️
Luv ya videos man got recommended to this channel haven’t looked back!!
You can still recover the corrupted bios even without having an Igpu or a secondary gpu, all you have to do is to swtich the bios back to the corrupted one while you're inside windows and run the vflash utility, then restart and as long as the new bios you put into the eprom is not also corrupted it'll post. At least that worked on the old amd cards back to like r9 280x 290x 380x era. Maybe even the hd 7970 did work the same but I can't remember, I used to flash the bios of amd cards so often back then. Now it's not even worth it because we have more power tool of which we can directly modify the power tables from.
He did that in the video. It didn't fix the problem, indicating a hardware fault instead of a corrupt BIOS.
Certain Mobos used to have removeable EEPROM chips, which was super handy if your chip died or your BIOS got corrupted during a bad flash. I got a couple of Asus AM3 boards that were dead from eBay and revived them with EEPROM chips for less than 8 quid, also from eBay.
I can confirm the importance of dual BIOS anything if you can afford it. I shelled a little more for a dual BIOS MB and it has saved me multiple times where I bricked one of the BIOS.
Jay, they don't have multiple BIOS chips, they just connect one of ROM the address lines to the switch to select top or bottom half of the chip. The corruption if it was a single bit could have only knocked out one BIOS image.
I had a similar thing happen, my 3090 and 3060 in the same pc somehow had the Graphics drivers corrupted, it refused to show video when any nvidia card was in the system. To fix this i needed video out, so I plugged in an RX 6600, used DDU to get rid of Nvidia drivers, plugged in 3060, reinstalled nvidia driver, plugged in 3090, everything worked fine afterwards.
Whenever the program says "update" it will check for a newer version, but to put the same bios on you need to find a "restore" option, or command line "write" option.
Use the software that came with the vcard.
There is warranty that covers the repair then use it.
Knowing what happened is important so that when it is repaired and installed the PSU doesn't do the same thing again because you fixed the PSU.
I had no idea you could flip the BIOS switch while the system is running, makes sense for flashing, and very cool.
I have that exact card I got mine a year ago at Microcenter open box for $1500, no big issues so far, just Radeon resets to default settings quite alot, so I had to make a save settings file so I can easily recover my custom settings.
Love it when you make these kinds of videos. always learning. thanks.
RMA/return it and see what they do to fix it. Most people think returning parts result in two outcomes. You get a new part as replacement, or they repair your part and return it to you. The latter gives a more "inferior" impression as a solution due to the possiblity of human error or other factors during repair.
On a different note: what mobo is that that has a display screen for boot errors?
Makes me nervous now. Running the Red Devil 6900xt Ultimate. Have had issues from the beginning with random monitors not working ( all 3 are DP ). Had no issues for years with my 1070ti. At one point I couldn't get any video out, hard reset and power cycle didn't help. Swapped back for 1070ti and came right up. Threw the 6900 back in and so far it is working. My first AMD card and so far it isn't instilling any confidence in me.
I remember when I had a powercolor 9800 pro back like 20 years ago, and it had problems too - I don't really have faith in powercolor as a brand, and I'm surprised they're still around.
If you figure out which chip it is, you should be able to get a replacement pretty easily. It's the soldering that would be tricky. It's probably not EEPROM either, most likely flash. EEPROM is pretty tough, FLASH not so much.
Maybe a fan short. Silent mode doesn't heat up as much and the middle fan doesnt seem to be spinning. The short would do all those things
RMA no point having a faulty graphics card with a non-working feature would be the best course of action. When I had the EVGA Geforce 690 back in the day it was starting to fail and I had the 10 year extended warranty so was perfect to RMA it and they sent me back a 980 as a replacement.
My Power Color Red Dragon 6800 had the same exact issue, I am currently using the silent bios so I can continue to use my card.
The one thing about a ROM chip is that a unregulated voltage spike can corrupt the memory and ruin it. You should be able to replace it inexpensively either though a warranty claim or even sending the card back for a replacement. But first see if you can find out why there was a spike (or not) otherwise it will happen again.
6600XT user and in elden ring some times the fans do not spin, that's how cool the card runs. It's all the card u need for 1080p.
Ideally contact support and have them send out a replacement subject to bill you for it if they don't receive the card with issue so the end user doesn't have to deal with down time ranging weeks to months. I'm sure many will not send a card out in advance which sucks and I'd make note of that for future purchases.
Dude, check this out: Yeah the first bios may be fragged, but why? Is it the bios chip it’s self or is it possibly shorting to ground somewhere? Check the caps or other components connected to that chip. If there is a blown cap or even another chip not pushing current like it’s supposed to, it would cause the chip to fail. If there is no such short or faulty components, then it’s definitely the chip it’s self. Either way, you need to disassemble the card and dust off your multimeter and soldering iron.
Been running this same rig. 5950x - 6900xt red devil ultimate edition. for about 8 months.
0 issues, fantastic performance. Both cpu and gpu ran overclocked constantly.
AMD ecosystem is fantastic. Adrenalin tool is very helpful.
DEFINITELY RMA this card. Would be nice to see PowerColor stand behind their flagship card.
Only thing I can think of to fix it would involve some soldering work, which doesn't quite seem to be up your guys' alley. (Electronic soldering is fairly tricky even when you're working with amateur grade PCBs - the ones where you acid etch off a single layer of copper for your traces...this is....rather significantly more complex than that.)
I had this exact issue with my 6800 XT, where it eventually wouldn't boot regardless of BIOS. Had to RMA it after only 3 weeks of use. Thankfully the company i bought it from got me a new one straight away.
I also have a Red Devil, but the 6700xt... And it's drivers crash like every other day... No matter what BIOS I use. The card is five months old.
I contacted PowerColor, they told me to contact the retailer. The retailer told me to contact Powercolor because they couldn't find an issue during their short test that cost me €145... 😮💨
I'm keeping it until I saved up for a new card, because it still kind of works... Half of the time.
Funny to see I'm not the only one with this issue
bro this video is amazing ! Jay you are a big hitter with great informative videos,
You “three” need to go Office Space on it, no joke. I think it would make good content!
There are tools for flashing BIOS chips outside of a system. I have a CH314a that I got for less than $20 and it saved a motherboard that I thought was just dead. Idk if GPU's use the same BIOS chips as motherboards and sometimes you have to desolder and remove the chip to flash it, but it's worth a look. If you plan on using it regularly, definitely get a better clip, the one it comes with is only good for a use or 2 before it becomes a real pain
Dual-BIOS.... I remember my old-ass Socket 754 Gigabyte motherboard made that a big selling point (I guess they pioneered it at least in the consumer market?). And it did save me hassle a few times... especially with BIOS updates.
An alternative possibly could be malware in the OC bios, maybe crypto, It can mask its location in BIOS by misreporting memory usage, but if the OC bios is close to the size limitation physically of the CMOS, It would get corrupted at the end because it couldn't fit. Thereby locking up your video card when it doesn't find the end of the file.
Could be interesting to try a hex dump of the BIOS
It's actually been a long time since I heard of a bios failure short of a flash gone wrong. Back in the day doing PC repairs/troubleshooting my first thought would've been a possible hack or virus affecting initializing or ' boot ' stages. Try another of the exact cards in the affected system without making changes and if it quits you'll know something more than the card is at fault. Also try a quick new build on a test bench with all budget parts. Sometimes I've found some unknown rare glitch in the more expensive hardware is to blame whereas the ' cheap ' system with the suspect parts works fine, with all drivers updated of course. 👍
If GUI doesn't work then command line it. If that fails, RMA it. Do a lotto for the RMA card for the person that chooses the correct date it is returned. Just create a website with dates. First person who fills out information to hold date slot gets that slot.
If you have contacts at AMD, it would be interesting if they could take a look at it and tell you what the error is. They might be interested in finding this out as well for their internal QA.
You could try to swap BIOS chips. Preferably from a dead card but potentially just swap the 2 chips on that card itself. Just to verify that it is the chip and not some traces or something else. Thankfully the chips are usually a bit larger so should be doable to solder it.
Try flashing the same version of the VBIOS that currently works onto the non-working EEPROM. If that doesn't work then I think you're correct that something about the EEPROM is dead and I'd just RMA it. It happens sometimes so might as well let them eat it.
I would try writing the silent BIOS onto the OC chip first. There could be some weird thing happening when the OC bios is on the OC chip. I have seen that happen before. If it doesn't help, I would remove the heatsink and clean the card. I once had an issue on my MacBook where it would work just fine and boot fine but every now and then when it went to sleep, the screen wouldn't wake up properly. Guess what, there was some dust that landed in a spot near a controller chip and was causing the issue. A good clean later worked perfectly 100% of the time.
(bvvrrt)
"ow my finger"
-immediately sticks finger back in almost the same spot...
xD
Yeah Jay I think you should just try to replace the Eprom on that or if you don’t want to risk it I wouldn’t mind trying it myself they sell brand new EPROMs that you can solder in place