Its hard to say but from my experience if it has that much damage to one corner it usually means its got very little chance of actually working, you can only see the external damage and not the micro cracks that are internal to the silicon die.

I wouldn't pay for it myself, but if it was for free .. then sure Id fire it up and check to see if its still alive and then hit it with a Prime95 run to see if its stable or if the damage has caused instabilities.

I've seen some pretty awful looking chips work fine back in the day.

Cross your fingers, carefully install the hsf with the intention of it being the last time ever, and press the button.

It's hard to say - I've seen worse - it really depends if the chip on the corner reaches far enough in to expose the die.

But... in my experience, working things are worth considerable more than non-working things and the effort to do a
basic test is minimal (doesn't need much to see if POST will happen) - when sellers say "untested", it often really
means "failed tests".

In as much as you can tell if any CPU works by looking at it, I say more likely dead.

Now I've seen some banged up Athlons and Durons that look real bad but still work, but these were usually in the hands of hardware enthusiasts. Now the difference between those and regular users, is that if they hear or feel the crunch, they double check what they are doing, while the other 95% don't. So it's sometimes not about that chip, it's about what happens immediately after. Chips on the left side of die often indicate that the heatsink was going on backwards, the step in it being the wrong side to rebate for the hinge mechanism area on the socket. Some heatsinks can still fasten in this position. If the system was then turned on, with the heatsink making virtually no contact with the rest of the die, it blew. There is no safe amount of time to have an athlon powered with no sink contact.

So in order for a left side chipped CPU to survive, it has to have been in the 5% where the installer has gone "Wait a sec, I'm messing up" rather than the carry on regardless bunch. It still could have had cracks radiate subsurface and kill it though.

Some of them just chipped even when pros did it.

Once upon a time I had some official amd certificate saying I knew how to do it. Still chipped one - and I wasn't alone.

No way to tell. It isn't too bad but I wouldn't pay more than 5€ for it.

Well, this works:

Did not expect it to (and got it for free), but somehow it is still alive.

That said IMO as already mentioned - the safest assumption is: untested=tested, bad. This is how it is most of the time. So paying any amount of money for this is questionable.

These are mine, all working fine. 100% lottery with this stuff.

That last one ...oof

I've seen much worse too, so I do think it has an OK chance of working.

But when I see seller is selling for parts only due to being "untested", I almost always assume the item(s) was actually tested and found not to be working, especially for newer hardware that's easy to test - e.g. PCI-E video cards, modern RAM, and etc. When it comes to older parts, though, then it's possible that the seller might be telling the truth if he/she does not have an old system to test the hardware. So with old hardware, it's possible to have them truly be untested rather than tested and failed, but only listed as untested. Another tell-tale sign is to look at the seller's other items - if he/she has the same type of CPUs (socket A) listed as tested and working, then the one sold as "untested" is more than likely a lie.

That said, let the price be your guide. If it's just a few $ and considerably less than what you'd spend on a working CPU of the same kind, then flip a coin and maybe try it out.

Again, I have seen s462 CPUs with much worse damage than this and they worked fine.
For those of you that are worried: keep in mind that the actual CPU core is at the bottom of the silicon die, and most of the silicon material on top is just inert silicon (i.e. not part of the circuitry in the core.) So that's why surface damage on these often doesn't result in core damage. Of course, cracked corners with really deep cracks all the way down to the bottom can kill the core. So just watch out for those.

Now, I'm not going to show my poor old Duron 750 with all of its abuse marks from make-shift CPU cooler experiments... but the core looks more like an oval. For those who would like to picture it in their head, take Mandrew's 4th pictures above the way the CPU is manged in that upper left corner, and apply it to all 4 corners of the die. 🤣
Yeah, I'm not proud of that abuse. But long time ago (~2008-2009) when I found that CPU and motherboard it came in, retro computing with such "new" components wasn't really a thing and the thing was practically worthless to me. I didn't have any good socket 462 coolers at the time, and being annoyed by the whiny tiny stock cooler the CPU came with, I decided to rig my own with an Xbox 360 CPU heatsink and a 80 mm fan. The good news is that I did make a relatively quiet contraption. The bad news is that it's hard to make a proper clamping mechanism for socket 462 that uses the socket ears, so I ended up with a lot of heatsink "slips" and "tilts", leading to dis-forming the core the way it is now. But hey, that CPU still works to tell a story.

Thanks for the replies! Really interesting.
It's sold in a bundle with 2 other CPUs (Athlon XPs - which don't interest me), for 10 € total.
Still thinking about it 😀