pegasus1:

It seems totally bizarre that the QA/QC process allowed this, surely there is a QA/QC process.

QA/QC is a constant question of finding a level that is good enough. If they want to sell a "cheap" card or have a high profit margin, the optimal level needs to be found, sometimes they overstep that level where the QA/QC is no longer good enough and deliver cards with faults on it. examples like the 3090 "New World" problem that turned out to be a 3090 GPU problem instead. 3090 high memory temps in high demand workloads. Asus Vega and 5700 with coolers falling off the VRM in horizontal position. Was it GTX1080 EVGA that had massive cooler problems? GTX980 with very high VRM failure rate. GTX970 4GB memory problem There a many other examples of things like SSDs getting slower chips after launch, cases catching fire, Samsung odyssey monitors with picture quality problems, AMD USB problems....... I am guessing here, but AMD could have been let down by the manufacturing plant of the cooler, the Vapor Chamber machine program could have been adjusted by someone or maybe the filling liquid is contaminated, maybe the chamber pressure or size changes a little when mounted instead of on the table when QC is carried out. The QC on these cards are not 100% test at all conditions, that is simply too expensive and the card would be 3000€ and everyone would laugh their ass off. AMD could have received perfectly good cards from the production line and then given the go ahead for mass production. They simply found out that they overstepped the "good enough line", so the cards are no longer good enough.

Horus-Anhur:

Truth be told, most people throughout GPU history, avoided the reference models, preferring custom cards with better cooling, OC, power delivery and aesthetics. People who considered the reference model would usually be people that wanted the cheapest. Or who would want a standard PCB to install a water cooler.

Yeah for aircooling or maybe warranty or even availability. As with all things get the best you can afford.

TheDeeGee:

I prefer this era of GPU boxes ^^

I got a 9800Pro and flashed it with an XT Bios, didnt really do much apart from speed up the VRam if i remember.

They should just stop manufacturing the reference model and offer an AIB model as an exchange. I'm sure the partners would collaborate. If no offered model is fitting for the customer then replace the reference model with a tested working one, and if such is not available, then offer money back. The reference cooler is a finely designed, certainly adequate and not too big, but alas, these manufacturing issues happen, though more often on AMD's GPUs than Nvidia's, including the AIB models.

GamerNerves:

They should just stop manufacturing the reference model and offer an AIB model as an exchange. I'm sure the partners would collaporate. If no offered model is fitting for the customer then replace the reference model with a tested to be working one, and if such is not available, then offer money back. Finely designed cooler, certainly adequate and not too big, but alas, these manufacturing issues happen, though more often on AMD's GPUs than Nvidia's, including the AIB models.

I think they might do that and issue a credit note or vouvher for an AIB of the same value maybe

TLD LARS:

QA/QC is a constant question of finding a level that is good enough. If they want to sell a "cheap" card or have a high profit margin, the optimal level needs to be found, sometimes they overstep that level where the QA/QC is no longer good enough and deliver cards with faults on it. examples like the 3090 "New World" problem that turned out to be a 3090 GPU problem instead. 3090 high memory temps in high demand workloads. Asus Vega and 5700 with coolers falling off the VRM in horizontal position. Was it GTX1080 EVGA that had massive cooler problems? GTX980 with very high VRM failure rate. GTX970 4GB memory problem There a many other examples of things like SSDs getting slower chips after launch, cases catching fire, Samsung odyssey monitors with picture quality problems, AMD USB problems....... I am guessing here, but AMD could have been let down by the manufacturing plant of the cooler, the Vapor Chamber machine program could have been adjusted by someone or maybe the filling liquid is contaminated, maybe the chamber pressure or size changes a little when mounted instead of on the table when QC is carried out. The QC on these cards are not 100% test at all conditions, that is simply too expensive and the card would be 3000€ and everyone would laugh their ass off. AMD could have received perfectly good cards from the production line and then given the go ahead for mass production. They simply found out that they overstepped the "good enough line", so the cards are no longer good enough.

i completely agree imho, another likely cause could be something as simple as dialling up or changing the thermal paste (which could be deforming)

Horus-Anhur:

AMD entering 2023 with the wrong foot forward.

the standard foot of the Radeon division crew

AMD partner suspects a faulty batch of reference Radeon 7900 GPUs is affected by thermal issues https://cdn.videocardz.com/1/2023/01/AMD-RADEON-7900-IGOR-1-1200x491.png

Once again Raja has another opportunity to shine with INTEL.

Dimitrios1983:

Once again Raja has another opportunity to shine with INTEL.

XD

TheDeeGee:

I prefer this era of GPU boxes ^^

I had a Sapphire 9700, man I miss those days

Seems all negative stuff is posted by nvidia fanboys. My next card will be a Sapphire 7900xtx. I can also go and scream about faulty nvidia cards, in your forum, if you want that. Oh, I wont, as it is too easy.

patteSatan:

Seems all negative stuff is posted by nvidia fanboys. My next card will be a Sapphire 7900xtx. I can also go and scream about faulty nvidia cards, in your forum, if you want that. Oh, I wont, as it is too easy.

idiot inside

What is the actual number of complaints for MBA GPUs? I mean the sensation is usually off the wall...

I really hope this is just a limited batch QC issue. And that AMD quickly recovers from it to be a viable alternative to Nvidias milking machine.

cucaulay malkin:

idiot inside

Surely it looks like full time assignment for you. Always on, always bitching...

GREGIX:

Surely it looks like full time assignment for you. Always on, always bitching...

so everyone is just bitching,right ? the people who rma'd too. it's not amd's fault, everyone is just too negative about +50 degrees difference between die and hotspot.

Even if it's just a batch of cards, like Igor says, it's still going to be a headache for AMD and it's partners to solve.

Horus-Anhur:

Even if it's just a batch of cards, like Igor says, it's still going to be a headache for AMD and it's partners to solve.

good thing for amd there weren't too many in stock in the first place, but sucks for those who got theirs.

[youtube=rxaDZ6n2MNo]

Horus-Anhur:

[youtube=rxaDZ6n2MNo]

~ 56 out of 223+ reported with problems? 25% of the poll vs how many sold again?