Overclocking Gone Wrong

[Maplez]

I was overclocking my old Radeon R9 290X RoyalAce 8GB in an attempt to get the maximum performance this card can offer, after increasing the voltage in msi afterburner wasn't enough I tried modding the vBios to increase the voltage further. It seemed like the modded 1.3v wasn't applying according to MSI afterburner and HWINFO, after a while of gaming around 74C my gaming PC suddenly turned off and refused to turn back on. Removing the R9 allowed the system to boot again so I decided to take it apart and figure out what broke, since it kept tripping the PSU I assumed it was a problem with on of the VRMs. I inspected the card thoroughly but couldn't find any signs of damage, I assume used a thermal camera to see what was getting hot but since it would immediately trip the power supply I couldn't get any useful information.

 

I'm kind of confused here on what broke, if I broke the GPU it wouldn't instantly trip the PSU and if it was any of the power delivery I would be able to see it.

Also according to some online sources, it is considered safe [as long as temps are in check] to do 1.35v with air cooled cards and 1.45 with water cooled ones, and at the time of the card's death it was only pulling 1.275v @ 74C pulling 330w.

 

I'm really just curious if anyone has any ideas on what broke here, I ran numerous benchmarks and it was stable, and had been using these overclocked settings for around 3 weeks at the time.

[aisle9]

Well, for one, that's about a 10-ish year old card at this point. I'm not saying you shouldn't overclock old cards, but I will ask when the last time you changed the thermal paste and pads was.

 

It's also entirely possible that the power delivery of a 10-ish year old card couldn't handle the non-stock voltage. I would imagine those 1.35v figures came around several years ago, if not right around the time of release, and that number is very different on a card that's seen plenty of use in the last decade.

 

Basically, anytime you screw around with VBIOS, you risk breaking something. To be clear, I don't think it's anything you did wrong. I think it's more likely that the 10-year-old power delivery couldn't keep up, or that the card had thermal issues and finally just gave up on life. IIRC, the R9 GPUs were kind of a hot mess on thermals, showing readings as much as 20C below what was actually going on (like FX CPUs did). If that's the case, your GPU might actually have been running closer to 90-95C the whole time.

[Haswellx86]

That's 10 years of hard work. Someday, something has to die. Rip .

 

1.3s and 1.4s are some high voltages to run at. The 290X also consumes a ton of power and run hot. They have low lifespan because of this.

 

Drama aside, I think the GPU is fine, and what is causing problems is probably some power supply mechanism on the PCB. You should take it to a repair center, where they can take a look of each component using their tools, and identify which component is failing here.

[WereCat]

Check if there are any fuses on the card.

My Sapphire Pulse 6800 XT has fuses next to each 8-pin limiting each to 240W power before the fuse blows to protect the card.

 

Though it should not be tripping your PSU...

[Maplez]

2 hours ago, aisle9 said:

Well, for one, that's about a 10-ish year old card at this point. I'm not saying you shouldn't overclock old cards, but I will ask when the last time you changed the thermal paste and pads

I believe 3 weeks before it broke I changed the thermal paste with some thermal Grizzly Kyronaunt Extreme, I changed the thermal pads 3 days before it broke and I matched the thickness.

Thermals on this card never hit the 90s, and after the thermal paste change it stayed in the low 70s. This 290X is a 3 fan model which helped it run much cooler then the reference design.

2 hours ago, aisle9 said:

IIRC, the R9 GPUs were kind of a hot mess on thermals, showing readings as much as 20C below what was actually going on (like FX CPUs did). If that's the case, your GPU might actually have been running closer to 90-95C the whole time.

That kinda reminds me of my FX3800M thermal issue, put new thermal pads on and didn't put the right thickness, half of the gpu die never made contact with the heatsink but the thermal sensor said everything was a-ok. It survived 10 minutes of msi furmark, and it emerged unstable and constantly crashing minutes after boot.

[Maplez]

2 hours ago, Haswellx86 said:

That's 10 years of hard work. Someday, something has to die. Rip .

 

1.3s and 1.4s are some high voltages to run at. The 290X also consumes a ton of power and run hot. They have low lifespan because of this.

Thanks to the 3 fan design it ran a much cooler life then most R9s.

2 hours ago, Haswellx86 said:

Drama aside, I think the GPU is fine, and what is causing problems is probably some power supply mechanism on the PCB. You should take it to a repair center, where they can take a look of each component using their tools, and identify which component is failing here.

I'm just curious about what failed on it, since parts are cheap I could do some soldering but if I pay anyone for a diagnostic of that type it'll just be cheaper to buy a used R9 290X 3 fan for 90 CAD

[-rascal-]

2 hours ago, Maplez said:

Thanks to the 3 fan design it ran a much cooler life then most R9s.

I'm just curious about what failed on it, since parts are cheap I could do some soldering but if I pay anyone for a diagnostic of that type it'll just be cheaper to buy a used R9 290X 3 fan for 90 CAD

 

PSU tripping sounds like a short.

 

If a fuse/resistor/capacitor blows, it would result in an open circuit, it wouldn't cause the PSU to trip.

An inductor may fail as a short, if it is overstressed.

A failed Controller / Transistor (e.g. MOSFET, VRM) can cause a short, depending on how it fails. 

[Maplez]

1 hour ago, -rascal- said:

 

PSU tripping sounds like a short.

 

If a fuse/resistor/capacitor blows, it would result in an open circuit, it wouldn't cause the PSU to trip.

An inductor may fail as a short, if it is overstressed.

A failed Controller / Transistor (e.g. MOSFET, VRM) can cause a short, depending on how it fails. 

I plugged it back in and I saw good ol smoke coming from the power delivery side of the board. Funny enough I can't find any signs of damage on any of the power delivery components

[xAcid9]

IIRC, my ol' 290x start tripping with voltage over 1.275v, I remember 1.25v-1.3v was like the "safe" limit for lots of people that time. Only the "chosen" one will go above that safely on air. 

 

Also by tripping I mean visual tripping, not PSU tripping. 

[Maplez]

49 minutes ago, xAcid9 said:

IIRC, my ol' 290x start tripping with voltage over 1.275v, I remember 1.25v-1.3v was like the "safe" limit for lots of people that time. Only the "chosen" one will go above that safely on air. 

 

Also by tripping I mean visual tripping, not PSU tripping. 

Ah, my gpu never artifacted unless raised the clocks to high, it was chill at 1.3V.

I remember now that at stock the vrms were getting 95C+ and I never bothered to check their temps after increasing the voltage to 1.3c. The core never got hot but those poor vrms

[xAcid9]

5 minutes ago, Maplez said:

Ah, my gpu never artifacted unless raised the clocks to high, it was chill at 1.3V.

I remember now that at stock the vrms were getting 95C+ and I never bothered to check their temps after increasing the voltage to 1.3c. The core never got hot but those poor vrms

What's the card clock speed @1.3v btw? 

[Maplez]

1 hour ago, xAcid9 said:

What's the card clock speed @1.3v btw? 

1125mhz or 1150mhz, according to msi afterburner the voltage wasn't going beyond 1.28v for some reason

[-rascal-]

20 hours ago, Maplez said:

I plugged it back in and I saw good ol smoke coming from the power delivery side of the board. Funny enough I can't find any signs of damage on any of the power delivery components

Welp, once you release the magic smoke, you can't put the smoke back in.

Something on that ol' 290X is Cookin'