1. That test did not validate every case of GPU failure. A machine that wouldn't power on couldn't be checked and the customer would sometimes be screwed (or supported; it all depended on the Apple Store they went to). You'll have to go reading old threads on forums (like I did for months while my MacBook Pro 3,1 started showing signs of GPU failure- mine didn't start failing until I started using it heavily with gaming and 3D software, suggesting thermal extremes were the issue).stratology wrote:Again, the Nvidia issue from 7 years ago had nothing to do with thermal problems. The issue was faulty Nvidia GPUs. Apple took responsibility and offered a replacement program for the motherboards.Jace-BeOS wrote: The GPU problem I mentioned and you blew off as nvidia's problem (rather than Apple's) shows us the thermal problems in tightly packed computers and showcases their lack of component-level repairability.
If you had an affected Mac repaired, you may remember that Apple Authorised Service Providers used testing software on a USB stick, to verify that the issue is actually with the GPU (as opposed to, say, an issue with the connection between the MLB and display).
Common sense: if it had been a thermal design problem, not a GPU chip problem, the issue would have re-occurred after the motherboards of affected Macs were replaced. Replacing the logic board does not change the thermal design in any way.
2. The solution was to give users another board with the same GPU on it, which would fail. There was no correction of the problem.
3. The replacement offer lasted as long as it took to use up all the unused motherboards. This is why mine was never "repaired" and it sits uselessly in its box, and no one seems to want to buy it for parts.
4. The ones that failed were the ones used most for heavy tasks (gaming and 3D killed mine), while other users who did nothing more than Internet, word processing, etc, never saw problems. Non-heavy she didn't kill thr GPU, suggesting thermal design issues.
5. The MacBook Pro 3,1 machine from 7 years ago was not the only example of this happening. AMD had a similar failure. It primarily happens on laptops and other densely packed all-in-one systems, suggesting thermal issues are the real trigger.
Is it a compact computer design issue or a chip design issue? It keeps happening, so either both AMD and nvidia are incapable of making non-defective GPU chips (unlikely, since this level of failure isn't reported in desktop GPU boards) or it's the thermal constraints of compact computer designs and the GPUs should never be okayed for use in these machines.
Even if we blame nvidia and AMD for releasing chips that claim thermal tolerances compatible with laptop designs that end up killing the chips, the manufacturer of the laptops (such as Apple) should acknowledge what is happening and accommodate for this situation by not continuing to push the thermal design harder every revision of the machine. Instead, that's exactly what they do. They continue to unnecessarily shrink the enclosure and decrease all kinds of material tolerances.
There are lawyers and accountants who tell Apple executives whether a potential design flaw is or is not on the unprofitable side of risk assessment. It seems the failures are far apart enough that Apple can still make a killing on machines that risk burning out prematurely (especially when they get a few million dollars in settlements from GPU companies to handle the proportionately small percentage of MacBook owners who will find out about and take advantage of any replacement offer).
In fact, Apple's choice to turn to more disposable machines (non-serviceable, soldered RAM/storage etc) just proves this is their preference, since it helps them sell more machines every three years. It's not economical for consumers or for the consumption of the rare materials being mined to build this technology. That's on Apple, not the GPU manufacturer.
