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

Some findings reported by mindfactory on GPU failure rates.

Joined
Mar 10, 2010
Messages
11,880 (2.12/day)
Location
Manchester uk
System Name RyzenGtEvo/ Asus strix scar II
Processor Amd R5 5900X/ Intel 8750H
Motherboard Crosshair hero8 impact/Asus
Cooling 360EK extreme rad+ 360$EK slim all push, cpu ek suprim Gpu full cover all EK
Memory Gskill Trident Z 3900cas18 32Gb in four sticks./16Gb/16GB
Video Card(s) Asus tuf RX7900XT /Rtx 2060
Storage Silicon power 2TB nvme/8Tb external/1Tb samsung Evo nvme 2Tb sata ssd/1Tb nvme
Display(s) Samsung UAE28"850R 4k freesync.dell shiter
Case Lianli 011 dynamic/strix scar2
Audio Device(s) Xfi creative 7.1 on board ,Yamaha dts av setup, corsair void pro headset
Power Supply corsair 1200Hxi/Asus stock
Mouse Roccat Kova/ Logitech G wireless
Keyboard Roccat Aimo 120
VR HMD Oculus rift
Software Win 10 Pro
Benchmark Scores laptop Timespy 6506


As expected really , AMD do indeed have a marginally higher return rate, skewed that way by power colour.

The highest failure rate for a card is the 2080Ti again no surprise.

None of the failure rates is beyond the scope of normal ie 6% , that's not great but not terrible either.

It's only indicative no more, though over 120000 cards were sold so it's far from a guess.

No one like facts?
 
Interesting read but yeah Powercolor do need to get their act together, I recently sold my Powercolor 5700XT it was fine, and replaced with Sapphire simply because I needed a reference ca5rdd for a waterblock, kinda thankful I did lol. In All seriouness sure the cards are fine but that is quite a bit higher at 6%
 
Did anyone expect anything different? Really? Navi on top, with 5700 variants leading the way... and the lone Turing outlier... a 2080 Ti.
Not really but the devil's In the details, and power colours lineup, without power colour, they're about equal all told.

Doesn't account for anything but hardware issues(mostly , driver's would be not counted though some returns could end up as such) but does quantify the issues both companies have had recently.
 
Not really but the devil's In the details, and power colours lineup, without power colour, they're about equal all told.

Doesn't account for anything but hardware issues(mostly , driver's would be not counted though some returns could end up as such) but does quantify the issues both companies have had recently.
I did, once, buy cheap, XFT for some WCG work. They didn’t last long running 100% for weeks. I guess we all have to learn. I’m little surprised the MSI/AMD models fare that bad. The Nvidia models that I punnish runs for years 24/7/365. Maybe just luck
 
RTX 2070 Super @ 1% failure rate, phew. But do these failure rates take into account the cards usage. Say gaming over content creation?
Or the fact some are mishandled or even tinkered with.
 
I did, once, buy cheap, XFT for some WCG work. They didn’t last long running 100% for weeks. I guess we all have to learn. I’m little surprised the MSI/AMD models fare that bad. The Nvidia models that I punnish runs for years 24/7/365. Maybe just luck

I have only bought sapphire Radeon reference card's recently, no issues, well my main rig is down ATM, has been a few days, I'm away ATM and it sprung a leak before I left so it's off short term but will be back crunching and folding when I get back home.
RTX 2070 Super @ 1% failure rate, phew. But do these failure rates take into account the cards usage. Say gaming over content creation?
Or the fact some are mishandled or even tinkered with.
No, they only take into account returned for whatever reason card's.

No one makes consumer cards that are designed to be hammered 24/7 ,due many years folding I learned this, you can use them but it's not made for it and compromises need to be made, better cooling than usual, lower clocks and volts perhaps on stock cooling or ,and putting the noisey mostly stock stuff in a room your not in for both noise and heat issues.

Most I have ran folding is four cards, 36 once mining ,either way no heating was required and I live in England.

@EarthDog I wasn't suggesting Nvidia are worse.
Just that neither significantly lead in failure rates.
 
Something I've never tried. Some 3d work and photo - video editing is all my cards are used for.
Neve had one across a few brands fail on me to date.
I have had cards fail, but mostly after a few years folding and never a waterblocked card, in saying that in a way most failed eventually, I only really punish water cooled cards for year's, and eventually they needed a re Tim or new thermal pads, all of them, I don't downclock or undervolt and frequently oc the snot out of them to game or bench but stock for folding.

And brand doesn't matter here if you Use the shit out of them you Will have to service it to keep it alive, even watercooled.
 
Source spreadsheet is here for those who don't want to dip their toes in the liquid sewage that is WccfTech:


Weirdly, the "Returns" column is now a decimal - which makes me question the validity of the whole thing...

1596569421404.png
 
Isn't AMD ~50% more? Sounds like a lot... but... it isn't when you're going from ~2% to ~3%...
Sounds like a successful launch and life if it's lower than 5% to me, both camp's are within reasonable failure rates in reality with even the worse AIB having the same reasonable failure rates besides a few outliers, it depends on your inner narrative I suppose, sure in some fields(cars, plane's) .05% is too many but when it comes to consumer gaming tackle, it's fair enough.

@Assimilator there was two such table's you think Nvidia's is shy?.
 
Sounds like a successful launch and life if it's lower than 5% to me, both camp's are within reasonable failure rates in reality with even the worse AIB having the same reasonable failure rates besides a few outliers, it depends on your inner narrative I suppose, sure in some fields(cars, plane's) .05% is too many but when it comes to consumer gaming tackle, it's fair enough.

@Assimilator there was two such table's you think Nvidia's is shy?.

Nah they're both weird, I just took a screenshot of the first one in the doc to prove for posterity I wasn't crazy (since I'm assuming the sheet will be fixed eventually).

Do we know that this is actually from Mindfactory, though?
 
Nah they're both weird, I just took a screenshot of the first one in the doc to prove for posterity I wasn't crazy (since I'm assuming the sheet will be fixed eventually).

Do we know that this is actually from Mindfactory, though?
Other site's also ran with it and there's little point in faking it.
 
Back
Top