Can a damaged CPU still work? And how long without heatsink will a C2D last?

[Burning_Ranger]

Me: C2D E4300, Abit QuadGT, Zalman 9700
Cousin: C2D E6600, Asus Commando, Tuniq Tower


Basically, I'm trying to settle an arguement. My cousin has a C2D
E6600, he messed up the BIOS flash and sent off the motherboard for
replacement (Asus Commando).

In the meantime, I bought an E4300, mobo, RAM and Zalman 9700.

Suffice to say my system didn't boot so in the process of
troubleshooting, I tried HIS E6600 in my mobo (Abit QuadGT) with my
Zalman (it didn't work). Now the Zalman when seated on my E4300,
woudn't make full physical contact (I blame the motherboard). Same
thing happened with E6600. But we thought it should be OK for
troubleshooting. We only had the E6600 with Zalman in my CPU for no
more than 5 seconds, more like 3 or 4.

My cousin got his replacement Asus Commando back and now can't reach
as high overlcocking speeds as it did (fails on 333 FSB). Is it my
fault?

1. My cousin claims a CPU can still partially work if it blows a few
transistors. I think this is nonsense, a CPU is a precision
microscopic instrument, not a wooden cog that will work half-assed for
a while. My thinking is if my motherboard and Zalman damaged the CPU
then it shouldn't work at all. The fact that his CPU won't OC as much
as before is due to the intermittent nature of the compatiblity
between CPUs and motherboards, basically he just got unlucky with his
replacement. Who's right?

2. If I were to run a C2D E6600 CPU without a properly seated
heatsink, would it REALLY burn out in less than 5 seconds? Surely it
woudn't reach 70+ degrees celsius in less than 5 seconds?!?!

 

[CBFalconer]

.... snip ...

1. My cousin claims a CPU can still partially work if it blows a
few transistors. I think this is nonsense, a CPU is a precision
microscopic instrument, not a wooden cog that will work half-assed
for a while. My thinking is if my motherboard and Zalman damaged
the CPU then it shouldn't work at all. The fact that his CPU won't
OC as much as before is due to the intermittent nature of the
compatiblity between CPUs and motherboards, basically he just got
unlucky with his replacement. Who's right?

Basically, I agree with you. The things that blow would be near
the major heat sources, and those vary with the job. So, while not
impossible, unlikely.

2. If I were to run a C2D E6600 CPU without a properly seated
heatsink, would it REALLY burn out in less than 5 seconds? Surely it
woudn't reach 70+ degrees celsius in less than 5 seconds?!?!

Sure it could. It all depends on the source of the heat.

 

[kony]

Me: C2D E4300, Abit QuadGT, Zalman 9700
Cousin: C2D E6600, Asus Commando, Tuniq Tower


Basically, I'm trying to settle an arguement. My cousin has a C2D
E6600, he messed up the BIOS flash and sent off the motherboard for
replacement (Asus Commando).

In the meantime, I bought an E4300, mobo, RAM and Zalman 9700.

Suffice to say my system didn't boot so in the process of
troubleshooting, I tried HIS E6600 in my mobo (Abit QuadGT) with my
Zalman (it didn't work). Now the Zalman when seated on my E4300,
woudn't make full physical contact (I blame the motherboard).

What exactly does "wouldn't make full physical contact"
mean?

Same
thing happened with E6600. But we thought it should be OK for
troubleshooting. We only had the E6600 with Zalman in my CPU for no
more than 5 seconds, more like 3 or 4.

This makes no sense, if you thought it would fry it and thus
you quit trying, why did you try it in the first place? If
you didn't think it would fry it, it would take more than 3
or 4 seconds to try it, you would have to turn it on and
wait for the watchdog timer to err out and reset defaults,
or turn off system then resest CMOS and retry it which is
again more the 3-4 seconds.

My cousin got his replacement Asus Commando back and now can't reach
as high overlcocking speeds as it did (fails on 333 FSB). Is it my
fault?

It's possible, but no way to be sure.
A poorly fitting heatsink can cause the heat spreader to pop
it's bond with the CPU core and it'll not seat as well
anymore.

On the other hand, there is no guarantee at all that a
different motherboard (even of the same make and model) will
overclock a CPU to the same level. There are lots of
reasons for this from particular specimens of chipset die to
random variations in discrete resistors and such in voltage
feedback loops making a higher or lower voltage, or a
different bios version on these boards, maybe even a slight
board redesign resulting in different revision #.

He was never guaranteed a high overclock though, so long as
the CPU is stable at default speed running at full load, it
is not something that should argued too much... life is too
short for that so how about buying him a few beers to make
peace?

1. My cousin claims a CPU can still partially work if it blows a few
transistors.

This is true, but it doesn't effect how high it can
overclock, it would outright fail at some things no matter
what the clockspeed. It can partially work but be instable,
but that would be anytime that damaged portion is *used* not
just when it's overclocked (normally, though anything is
possible it is unlikely you're that one in a billion to have
a wierd short from a core that was heat damaged, but
starting up a system from cold/off state is the # 1 way to
heat damage a Core/2 Duo, not while it's running already
(which is a more gradual heat increase so it has more time
for the integral protection circuit to sense this change in
temp and throttle or shut off).

I think this is nonsense, a CPU is a precision
microscopic instrument, not a wooden cog that will work half-assed for
a while.

Ok, an eample...

Some cars might be considered precision instruments. You
back it out of the garage and break off the side-view mirror
in the process. Rest of car works the same in every way but
as soon as you try to use the *side-view mirror* portion of
the card, you reach a logical *error* state where you don't
get back the data you needed. The card still goes as fast
as before but because something on it is broken it may be a
problem.

This isn't as likely as simply having the whole thing fail
completely. "odds" are the reason it's not o'c as well is
due to the replacement motherboard or bios/settings, but it
can't be determined for certain at this point unless you can
manage to retrieve and revive then recreate the entire
scenario again with the old board.

Incidentally, how can we assume the old board was a stable
o'c, if it died? I'm wondering if it is not a good idea to
try to o'c this board as much as the last.

My thinking is if my motherboard and Zalman damaged the CPU
then it shouldn't work at all.

As mentioned above, this is likely but there is a reasonable
chance of a popped spreader type of problem, but odds are it
is the motherboard, not that.

The fact that his CPU won't OC as much
as before is due to the intermittent nature of the compatiblity
between CPUs and motherboards, basically he just got unlucky with his
replacement. Who's right?

Either of you "could" be right but it's more likely you are.
Buy him a beer anyway. You never did tell us exactly what
the spread was in prior o'c versus current o'c?

2. If I were to run a C2D E6600 CPU without a properly seated
heatsink, would it REALLY burn out in less than 5 seconds?

Yes it could.

Surely it
woudn't reach 70+ degrees celsius in less than 5 seconds?!?!

Yes, it can if past CPUs are any indicator. Since it has a
heat spreader, that is better than nothing, but the moment
it heats so rapidly it can pop the heat spreader off and
completely kill the CPU, or just result in a poor interface
because there is no permanent bond between them anymore.

However, if both boards had the same bios version, in the
present case it would mean the CPU temp is reported higher
than it was previously. That may be hard to compare now
though, without some prior logging of CPU temp vs load state
vs room temp, etc, etc.

If he's real adventurous he could pop the spreader off
himself though it might even cause damage that didn't exist
already (it can be done but no guarantees anyone or any try
in particular is *safe*).

I'd just forget about it, be happy with whatever o'c it
yields and at the lower o'c, it runs cooler and this allows
lower system noise, fan wear, dust buildup.

 

[paulmd]

Me: C2D E4300, Abit QuadGT, Zalman 9700
Cousin: C2D E6600, Asus Commando, Tuniq Tower

Basically, I'm trying to settle an arguement. My cousin has a C2D
E6600, he messed up the BIOS flash and sent off the motherboard for
replacement (Asus Commando).

In the meantime, I bought an E4300, mobo, RAM and Zalman 9700.

Suffice to say my system didn't boot so in the process of
troubleshooting, I tried HIS E6600 in my mobo (Abit QuadGT) with my
Zalman (it didn't work). Now the Zalman when seated on my E4300,
woudn't make full physical contact (I blame the motherboard). Same
thing happened with E6600. But we thought it should be OK for
troubleshooting. We only had the E6600 with Zalman in my CPU for no
more than 5 seconds, more like 3 or 4.

My cousin got his replacement Asus Commando back and now can't reach
as high overlcocking speeds as it did (fails on 333 FSB). Is it my
fault?

1) Motherboard is different (yeah it's still an asus commando, but
that doen't guarentee that the components are all made by the same
people, and have the same tolerances as before)

2) Overclocking is a dumb idea.

1. My cousin claims a CPU can still partially work if it blows a few
transistors. I think this is nonsense, a CPU is a precision
microscopic instrument, not a wooden cog that will work half-assed for
a while. My thinking is if my motherboard and Zalman damaged the CPU
then it shouldn't work at all. The fact that his CPU won't OC as much
as before is due to the intermittent nature of the compatiblity
between CPUs and motherboards, basically he just got unlucky with his
replacement. Who's right?

Yes, a CPU can partially fail. They have the stability of a drunken
horse when they do, but they can half fail. I've seen it.

2. If I were to run a C2D E6600 CPU without a properly seated
heatsink, would it REALLY burn out in less than 5 seconds? Surely it
woudn't reach 70+ degrees celsius in less than 5 seconds?!?!

I can't speak to the core 2s, but i can personally attest that a
900mhz duron without a heatsink can burn itself to a crisp in about 3
seconds.

Supposedly later cpus have better thermal protection, and will shut
themselves down to prevent thermal death. That doesn't mean no damage
will result.


Bottom line: cousin is an idiot for overclocking, and screwing up the
bios flash. You're an idiot for trying to run it without a proper
heatsink.

Solution: run the cpu non-overclocked, and be grateful it still works.
Accept mutual blame and chock it up to lessons learned.

 

[Stefano]

Me: C2D E4300, Abit QuadGT, Zalman 9700
Cousin: C2D E6600, Asus Commando, Tuniq Tower


Basically, I'm trying to settle an arguement. My cousin has a C2D
E6600, he messed up the BIOS flash and sent off the motherboard for
replacement (Asus Commando).

Few years ago I helped a friend managing a place for FPS
enthusiasts... practically they had all the CPUs overclocked... and
you can imagine the kind of work of these machines.

Well, they ruined most of them and the place closed because there were
no funds for replace them.

NEVER OVERCLOCK!!! NEVER.

 

[kony]

Few years ago I helped a friend managing a place for FPS
enthusiasts... practically they had all the CPUs overclocked... and
you can imagine the kind of work of these machines.

Well, they ruined most of them and the place closed because there were
no funds for replace them.

NEVER OVERCLOCK!!! NEVER.

I've overclocked every single system I used for gaming in
the past 10 years. Besides one eventually having vented
capacitors, the overclocking was not a problem.

The key in overclocking is being reasonable, not trying for
the very last MHz possible but retaining a reasonable margin
which requires finding that highest MHz possible then
backing off of that some. It is best not to put excessive
voltage through the CPU, again a bit of moderation as most
semi-modern CPUs (past roughly the K6 and Cyrix (or more
modern Via processors) have a fairly log curve of voltage
increase vs frequency increase past a certain point, so
finding the "sweet spot" for a given specimen of CPU is
helpful such that you use least power possible to attain ~
90% of the overclock possible.

It also requires checking parts temps (not just CPU) to
ensure all stay cool enough, and in some cases upgrading the
cooling of the motherboard VRM subsystem.

As with any (non-o'c systems also) system, ensuring stablity
is key. The higher it is o'c, the more frequent it might
need retested for stability. A system that has aged a few
years may not o'c as high as it did when brand new and the
margin you left will counter that, but unless the system is
only reserved for trivial uses (gaming, etc), you don't want
to find out it drifted into an instable zone while using it
(for intended purpose rather than the stability test).

Like anything else, you need to know the finer details of
something to have a good result. There is no reasonable
claim to "never overclock"! The more reasonable claim is
don't overclock if you don't know enough of the finer
details to attain a result you find acceptible. Similarly
you shouldn't scuba-dive or do auto repair if you don't know
what you're doing. It's all about details...