Five dead AMD socket A Athlon XP CPUs. WTF is going on?

I've tried googling to see whether there's any reason why five second
hand AMD cpus, that I recently acquired last week from a trader in a
local flea market, should all exhibit the same symptoms of 'deadness'
but, to absolutely no avail.

The reason I'm posing such a question here, is that whilst I might
expect the odd 2nd hand AMD cpu to be DoA, it seems extremely improbable
that I should see five such examples on the trot.

The first three were in a batch of 4 'pulled' socket A full size MoBos
with chips (along with a bag of 3 Durons - which, annoyingly, tested
ok). When I tested the boards with the cpus as fitted, only one worked
(the one with the unlocked XP2600 (AXMA2600FKT4C) which was reported as
a 500MHz mobile cpu).

It was only when I started testing the cpus in my trusty Jetway V600DAP
MoBo, did I think to retest the three 'faulty' MoBos, discovering, by
making good use of those 'pesky Durons', that only one was actually
faulty.

When I spoke to the trader a few days later to remark on the
improbability of getting three dead CPUs (I wasn't complaining, for the
money, I'd gotten hold of 3 decent socket A boards), he handed me
another two Athlon XP chips for free. Subsequent testing revealed
exactly the same symptoms of 'deadness' as the first three, taking my
count of dead athlons to an improbable high (even for AMD cpus, 5 on the
trot is exceptional).

Close visual inspection failed to reveal the slightest signs of
physical abuse (broken or contaminated pins or alteration attempts on
the L Bridges) nor even the slightest hint they'd been overheated. In
short, they looked in 'prime condition'.

Now I, unlike most posters, am priviledged to own an analogue wattmeter
which reveals that the test setup draws pretty well the expected power
for the clock speed setting, showing less power draw on the 100MHz FSB
setting than when on the 133 MHz one. This implies that the CPU under
test is responding to the clock signal just like a fully functioning
chip would, except that it looks like it is being held in a permanent
reset state since there is absolutely no hint of the test setup even
starting a POST (not even when I try sans graphics adapter and ram -
when I'd expect to at least hear the long mournful beeps from the
attached speaker to indicate missing ram).

The rather curious thing is that all five chips show exactly this self
same symptom, leading me to wonder whether there is any one particular
mode of failure that could account for this. I know that ESD damage due
to incompetent handling could account for failures that lack any
evidence of physical damage as in these five examples, but I wouldn't
have expected to see such a systematic response to ESD induced damage.

As it happens, I've got more than enough XP athlons to be going on with
as well as the, usually more rare commodity, _working_ socket A boards
(full size, rather than the micro ATX I was hoping to get), so I'm only
really interested in replies that can definitively suggest a common fail
mechanism or else offer an interesting theory on how I've managed to
corner the market in five non-working examples of otherwise perfectly
blemish free XP chips.

--
Regards, John.

Please remove the "ohggcyht" before replying.
The address has been munged to reject Spam-bots.

.

Relevant Pages

  • Re: 32-bit vs. 64-bit x86 Speed
    ... these chips run legacy 32 bit code faster than 32 bit chips do, ... personal computers then, were, in hardware: ... More powerful graphics cards. ... chip with 32 CPUs which can be shut down to conserve energy, ...
    (comp.compilers)
  • Re: Mystery Alpha CPUS, what do they fit in?
    ... >> that the mobo's were never made and all of these Alpha CPUs are ... Maybe they simply made more chips than mobo's. ... the chip area and wafer size. ... simply had better yields that expected and wound up with more cpu's ...
    (comp.os.vms)
  • Re: PCI Soundcard for XP
    ... Their chips would sometimes just 'go away' for ten milliseconds or so, ... You get weird locking conditions and race issues that ... The notion that modern CPUs go into "unknown states" is categorically ... that we can't write 100% provably reliable software (even operating ...
    (rec.audio.pro)
  • Re: VIA C7 server
    ... Christian Weisgerber wrote: ... C7 chips? ... C7 CPUs? ... Good luck. ...
    (comp.unix.bsd.openbsd.misc)
  • Re: Querying CPU frequency
    ... work on all "latest" Intel and AMD CPUs? ... With "latest" I'd say the CPUs that were thrown on the market in the last ... The TSC drifts for low power ... it's neither AMD nor Intel ... ...
    (comp.lang.asm.x86)