They won't -- part of the reason I'll likely buy another AMD.
I do feel a little bad laying it on about AMD since the temperature aside my chip seems to do the job and the motherboard and chip were reasonably priced and I didn't have to trawl through lots of specification lists to know whether or not it had virtual machine acceleration enabled as I would have to with an Intel chip.
I also get the impression there's nothing particularly unreliable about AMD either.
It really does come down to me not being able to trust the chip I have nor verify whether or not it's faulty. If I knew something like the temperature sensor were faulty or I've managed to constantly mess up cooling I may well suggest AMD to others but as it stands I can't.

Your answer confuses me.

1. AMD boards will not take Intel chips, yet you would buy another AMD chip.
2. And "I may well suggest AMD to others but as it stands I can't"

1. I could do with a CPU upgrade so rather than spend twice the money on a new motherboard, CPU and RAM I may buy another AMD CPU later this year in the hope that my current CPU is faulty. I wish I had gone down the Intel route when building this PC but AMD was a simpler and cheaper choice so for me the gamble of another AMD chip to keep it going another couple of years until I feel it worth buying a new motherboard, CPU and 32GB of RAM is worth it.

2. For value for money and that they work I could almost recommend AMD to anyone who realises that they're not going to be as fast as an equivalent Intel part but when I don't trust the CPU I am running myself I can't. The bigger picture of the confusion over temperatures still stands but for some that's not an issue either. In other words I wouldn't want to completely dismiss AMD on my experience as I know others who don't have these issues. However, I can't in any conscience recommend them as things stand.

Apparently you 2 guys are younger, or at least less aware or experienced in the history of Intel and AMD. Like most large corporations a high degree of effort (man-hours and money) goes into tripping up the competition, not just making "a better product". Before the year 2000, Intel abandoned Socket 7, which was the last socket that allowed compatibility with any other company's CPUs, and introduced the proprietary Slot 1. AMD was ingenious in a 2 pronged response, the Super 7 socket and Slot 2 which had the same number of channels and connectors (making it less burdensome on mobo manufacturers) but grouped electrically differently to avoid patent infringement.

Around this time AMD hired numerous engineers from the DEC Alpha project (awesome CPUs years ahead of Intel in every way) when DEC went under largely because x86 architecture was so firmly entrenched. It was shortly after this that AMD invented the 64bit architecture that even Intel was forced to use since Intel's IA 64 had many problems and was far less efficient and trustworthy. Shortly after this, AMD focused on the lower and midrange price market figuring that the bulk of sales are in simple workstations and OEM SOHO Desktop systems, realizing that Intel had held dominance of the top end so long that it will be a very long time (if ever) before they will likely be beaten at what is certainly "their own game".

Apparently neither of you clicked the link I provided since now both of you have concluded that AMD temp reporting "is useless" when it is not!.... excepting at idle, obviously of almost no concern.

Neither of you seem to grasp how "useless" (to use your term for "not accurate") BOTH Intel and AMD are, nor the value of having "inaccurate" temperature sensing. Just look at Intel's own design and function here -

Intel Temperature Sensing

Since you are using AMD in Linux here is important documentation on using the "force' parameter on the newer AMD CPUs in Linux

Kernel Issues and Fixes for AMD Temp Sensing

If you think AMD is the equivalent of the Pinto, a dangerous failure, guess again. Here is this years Stock Index and Sales Volume

Current NASDAQ AMD Charts

I hope this resolves both issues and mistakes for you, assuming you might actually click and look at hard data.

The article does say how to use the force=1 parameter.

This didn't work for me.

Maybe it will for you. You'll need sudo for those too.

That parameter seems to only be for Socket AM2+ and not later? So, while it may (or may not) resolve the original; question it does nothing to prove whether or not my CPU is faulty. The sensor may work well to tell the chip when to throttle back and when it can "Turbo" but since the fan speed on my CPU goes down while temperatures still appear over AMDs (this will kill your CPU) temperature I'm doubtful of that too. That is unless the temperature is completely useless to the user of the CPU -- which is completely possible but, then, AMD should document that the CPU will shut down in the event of thermal problems and not give a "this will kill your CPU" temperature of 55C when, in my example, the CPU can hit 75C reported.
In other words, reporting inaccurate temperatures, is bad no matter how useful they actually are to the chip itself.
I'm aware of the history of AMD and Intel (also being aware that, while AMD seems to make some movement Intel quickly becomes the better chip overall).

A post I received.

Thanks Fixit, that seems to be my issue and without an actual set of numbers from AMD it renders the temperature gauge useless to me (and others, I'm sure) hence my not knowing whether my CPU is defective in some way or not.

What makes you think your cpu is defective ?

It either runs too hot or reports it runs too hot. The fan will slow down while it's under full load and the temperature seems to be going up. But, since none of the temperatures I am reading are supposed to be the real one it might not be at all.
As I mentioned I'll likely be buying an FX-8370 to replace it fairly soon and I'll see who that behaves in the (hopefully) couple of years or so it holds me over for until I'll look to building a new system.

I can see why you are frustrated.

FX-8370 is a fast chip. :-)

Good luck.

Thanks, hopefully it will build some trust in AMD for me.

The thermal issue is far more important than mere temperature reporting, even though those numbers can be helpful in revealing the effects of improvements. Example - I recently bought a used Thinkpad T61p and there are several ways in which I know that the Core 2 Duo is operating within spec but that is by no means good enough for me. I want this Notebook to actually be a Laptop, something I can comfortably rest on my lap or chest when lying down. Beyond that a lifetime in electronics taught me that "Heat Is The Enemy" and experience proves it's validity in longer life and more stable operation. I will be absolutely shocked if I am unable to drop temps by at least 20 C. The ultimate arbiter though will be whether I feel I can avoid sterilization LOL.

Bottom Line - It is well worth hours of effort and many dollars (rarely exceeds $50-$60 USD) to get your system as cool as possible and is something I always do from the beginning by default. Who cares whether the resulting avg temp is 35C or 39C? As long as it does not feel hot to humans it is good design in electronics. Some stuff must run hot but Computers are not one of them. Otherwise, Cray would never have resorted to cryo-cooling and modern server farms would not spend the vast sums they do on water cooling. Maybe stop worrying and work to cool it instead.

I've already fitted an aftermarket cooler and it's in a Fractal R4 case with two fans at the front and one at the top (the CPU cooler vents upwards). If the CPU needs more than that it's junk. A friend has a(n admittedly more modern) Intel unit in a very similar case and cooler setup and no issues and I've not heard of anyone in this climate needing any more.
If AMD gave a user-readable temperature and a limit value or a lot more information it might be OK. Doesn't help also that the "this will damage your CPU" temperature is about 55C. Whilst we all know that that ought not to be reached in ideal conditions it's hardly a large margin of error if values aren't accurate.