The details which I've provided have indeed been limited in accuracy,
and one of the causes might very well be my limited understanding of
technology. But another cause for this limitation is the nature of
systems built by some OEM manufacturers, such as using customised
motherboards, which is NOT the case with all system manufacturers.
If referring to the major OEMs, yes it is the case. Where
you could have been to have not seen this for many years
will be a mystery.
If
this is wrong or right, it is not the place to discuss it - it just
precludes information gathering when troubleshooting.
Then why did you mention it? Debunking falsehoods is never
necessary if they're not put out there in the first place.
Yes, you are right about the difference between cheap and expensive
power supplies - but I have heard this argument on suspiciously
numerous occasions to justify underpowered systems, which die out after
2 years of use. True, 450W good powersuply will most likely be better
then a 500W cheap one, but powering modern computers with 250W
powersupplies (be them cheap or expensive) is underpowered by most
official information, including processor manufacturers like AMD.
That's because they overshoot. AMD does not requie 450W PSU
for any processor they make, in fact deferring to OEM
determination. What OEMs do that the casual user or
assembler does not is determine the actual need. Fact is, a
low-end modern system can run fine for years off a 250W PSU
if it is a decent quality unit. Medium tiered systems might
do it as well, largely depending on what makes it medium
rather than low-end. A faster CPU or another 512MB+ memory
is not much more power demand, while a video card,
especially a gamers card can be another matter.
Just about any mATX system is using 200-300W, usually closer
to 200W PSU. The OEM can effectively determine the real
power requirement so the remaining issue is quality of the
PSU... since an OEM may source much better quality in
200-300W than you would buy off a shelf in that wattage
range.
And
this is done on a regular basis by most system manufacturers, large and
small. Situation is even worse when the PC includes a AGP video card
with a large amount of RAM, plus two optical drives. On top of all
that, a minor number of systems I have opened had good brand power
supply - most of them were small, and cheap. All the worse.
I suspect you simply don't know decent power supplies. If
an OEM box has a major manufacturer's PSU in it, and it
bears that manufacturer's name rather than a 3rd party
relabeler who doesn't manufacture (like Antec), and if it
comes in a dull grey casing too, then to the untrained eye
it resembles a generic but is quite the opposite.
No, more ram and a 2nd optical drive are relatively light
loads, not a justification to jump from a 250W to 450W PSU.
However, I don't recall ever seeing a major OEM system with
a lot of memory (for it's era), and 2 optical drives, with
only 250W PSU. Care to name one make and model?
To come back to your question, yes, the power supply we've put in was a
cheap brand, as was the one we have taken out. I prefer 500W of cheap
power supply, over 250W of cheap power supply.
You might be causing problems then. 500W generic is
expected to last a couple years as you found problematic, as
the more common failure points such as caps, fan, diodes and
chopper BJTs are lower rated most often... how did you think
they managed to hit the low price points? Elvin magic?
To answer your question, no, we haven't touch-tested the temperature -
which, talking about accurate information, would seem rather
subjective, don't you think?
No I think it is prudent to do it failing another method to
obtain the temp. If you are setting up, servicing or
troubleshooting the system then you "need" to come to a
conclusion. While a touch-test is subjective, AFAIK the
typical human has a relatively constant body temperature and
their subjective interpretation of temp is "close enough",
it need not matter if "hot" is +-5C different from one
person to another, it's the ballpark temp that matters as
one does not try to shoot for only 5C under the threshold
for instability.
I'd also made the common considering you had no idea about
Athlon motherboards in general, and felt it beyond your
skill level to determine the specs for the board towards the
end of finding the addresses to interface with MBM5 or
similar to get a temp readout. It is most likely there is a
temp to be had but you'd already gone off on tangents about
overclocking by the OEM without merit.
Funny, same argument I've read elsewere on the Internet coming from
Medion's support - and a perfectly reasonable one, if it wouldn't be
for the evidence which suggests other reasons for locking the
frequencies.
Not funny at all, you are inept and guessed it must be their
fault. Again, the system might be flawed but you are only
guessing at the problem without evidence. I attempted to
provide a few hints to determine the CPU and it's o'c status
but had you even used one of several software tools to see
what the actual FSB frequency & multiplier was? If so, it
should have been quite clear whether the CPU was o'c or not.
What you had claimed as evidence was wholley insufficient
and even false towards your conclusion. In short, it
revealed you don't know what you're doing, are incompetent
to set it up... which is fine, that's why people buy OEM
systems (among other reasons) but when there is a problem
you are not fit to solve, you need to refrain from guessing
and do some research.
I perfectly agree with you that the temp information in these
circumstances is not infallible. But considering that a software
released by the manufacturer of the motherboard has been used, ...
No, NOT considering that because one cannot presume an
earlier software has a valid readout on a newer motherboard-
that it came from same manufacturer is quite irrelevant. If
it had been the other way around, if a newer software than
board, the situation changes as they are typically backwards
compatible at least for a period.
I would
say there are very good chances that the reading was correct.
Another jump. It might be correct, but you have no
reasonable expectation that it is and what you have
suggested is not correct, it matters not that it came from
same manufacturer but rather the date of release.
I would
doubt that when customising the motherboard, they would have had any
reason to move registers around, thus making the reading incorrect.
You guess again. Fact is, YES they do in fact use different
registers on different motherboards, particularly when the
technology changes over a period of time.
And
anyway, correlated with system instability and cpu overclocked, I would
say that it seems more accurate than "touch testing"
"IF" the report is correct, then yes it will naturally be
more accurate. Problem is your random assumptions and
further, that one can be a confirmation of the other. You
are not logically attacking the issue.
IF the system is overheating, that in itself may be the
problem but you use it as some support for your conclusion
that it's overclocked which is not supported by any evidence
and not even implied by the observed details. You guessed
it based on a prior guess that an "auto" or default POST on
a board would always result in the final FSB speed a CPU
uses, when in fact that is not true, rather boards do POST
at a lower FSB default fairly routinely when a new CPU is
installed.
We've purchased "Arctic Cooling Copper Silent", the one with 3 manual
speeds. My research and experience indicates they are quality, well
proven coolers. If you have different personal oppions, that's
something else. And it was fitted correctly.
Apparently it wasn't good, or wasn't installed correctly.
If both of those were true, you would not have had the CPU
overheating.
Do you even understand yet that if the CPU had been
overclocked by raising the FSB, it would NOT produce more
heat than if the CPU had been the "official" part, a genuine
3400?
Your incorrect guessing game has again caused further error.
You are incompetent and even worse, so full of
misconceptiosn that it will take longer for you to RElearn
things properly than someone who wasn't so prone to guessing
towards false conclusions. This may seem offensive but it
matters not- the end is necessarily that you realize where
you're going wrong else it persists, which is no good for
you in the long run.
Case ventilation is good, and all dust has been removed during a
previous session.
Perhaps, but after what's been read thus far, it would seem
best for you to doublt-check that, perhaps pointing a fan at
the open case to see if it improves temp much. Even though
your heatsink is not very good, in a well cooled case it
should be keeping the CPU cooler than it is. It would not
matter whether CPU was overclocked by FSB or not towards
heat generation, only the CPU family, MHz speed and vcore
will determine that providing the system load remains equal
which it should relatively speaking in same system.
Unless the processor is overclocked.
Nope, I already explained why.
However, if you want to consider overclocked CPUs, I have a
few I keep around for gaming systems, which have raised
vcore (CPU voltage), raised FSB, run significantly faster
than a 3400, and significantly cooler than 69-74C without
extreme measures- quiet, low-RPM fan on the heatsink, quiet
chassis cooling.
Any way you look at it, your account of the details does not
support your conclusion.
My omission again. An Athlon XP 2200 has been tried as well. Still no
success. Besides, if I remember correctly a Sempron 2800 (socket A) is
built on the same core as the Athlon XP, and so far I've had no trouble
running them in all boards which support Athlon XP.
At this point, it can be ignored. Primary focus is on why
the original system configuration is instable- which is what
should have been the goal all along but you got side-tracked
on this whold false-presumption-of-overclocking tangent.
You are absolutely right again in the general assumption, unfortunately
that was not the case. The old processor has been run in the new
motherboard for several hours, including a Windows XP home edition
installation. During the process, I have entered the BIOS several
times, changed different options, and exited, saving changes. Processor
remained recognised as XP2500.
Did you do the simple thing I had mentioned, looking at the
label on the CPU? You have not changed the appropriate bios
settings if you "always" see it as an Xp2500, because
changing the FSB, as required on many motherboards, will in
fact make the difference.
Let me remind you that just few paragraphs above, you actually agreed
with me: "> So? Of course they are, you paid for "X" speed so why
would
No, it's just that both ways you were wrong.
Besides, don't processor manufacturer warn everybody that the warranty
is void through any attempt at overclocking? I'm guessing they have a
good reason for that.
You have no reason to believe it's overclocked. You just
guessed and that guess is most likely wrong.
If you would have read my post carefully, you would have seen I was
called in because of stabiliy issues (with system in stock settings, as
bought - which can't be changed anyway)
Yes, since they can't be changed, that makes it even less
likely that CPU was overclocked. You had a basic problem-
CPU "appeared" to be running too hot. You then failed to
take adequate steps to determine without a reasonable doubt
that the temp was too high, then guessed that the temp was
too high because it was overclocked... when even if it were
overclocked, it would NOT raise the temp higher than it
would have been if it had been the official 3400 part, which
it most likely IS.
... - and not becuase we were
trying to overclock it. Viceversa, we would have tried to underclock it
if possible, to make it more stable, and that is why we have checked
the BIOS.
None of this is necessary. There is only one simple
resoltuion here:
Determine why it's overheating. NOT "do you think it's
overclocked", rather, why is the cooling system not keeping
it below what you "think" the temperate is.
If you lower the CPU speed below it's spec'd value, it will
be producing less heat but that does not address the
problem- which was why it was overheating. It was not
possible to do that from only having higher FSB, it would
have produced the same heat running at same MHz either way.
Nothing to comment here, obviously just personal stuff. Easy to write
about others, not so easy to work for real in real life. I don't
pretend to be the absolute expert, just looking for answers and hoping
others might find some in the information I've posted.
Nobody said you had to be an absolute expert, but it is
necessary to only go on solid facts and not jump to
conclusions particularly when a "client" is depending on the
result. If you can't fix a problem (which you have
demonstrated) you need to defer to someone else whose
expertise CAN solve that problem. By learning what was
necessary to solve the problem, your knowledge then grows.
Guessing and pointing blame only builds false illusions.
As I have
demonstrated above, I have taken the suitable technical approach in
troubleshooting the problem, and I am still open to suggestions.
As I have mentioned repeatedly, no you have not taken a
suitable, nor technical approach. You have guessed,
essentially built a house of cards that would only stand if
you further guesssed the CPU must be overclocked- when even
a very simple check could determine this.
If in fact it were overclocked, even then there was an
obvious solution, to demand the correct part from the OEM.
Any way this is read it's still not suitable technically.
Simply
being looked and talked down to doesn't seem like much of a productive
suggestion.
I have stated what needs to be done. I was even being
generous to point out where you went wrong so you could
avoid further errors. Go ahead and be offended- but either
learn from it or you are all the more negligent for
persistiing in false diagnosis and libelous claims against
OEM(s).
It seems that you have missed the point. If I wanted a law suit, I
would have kept things as they were for "evidence". My purpose was to
help, if possible, other owners of similar systems who might be going
through similar difficulties. And yes, I would love if companies who do
such systems would change their ways in the future.
That's just it- you have done nothing to help. What
would've helped is to pinpoint the problem, why the heat
isn't being removed effectively. There is no need to
consider whether the CPU were overclocked (which it does not
appear to be), only why the heat generated by a CPU running
as a 3400, is building up onto the point it seems to be
overheating. Doing anything before resolving that is only
negligent and a waste of time.
Already answered that. Maybe you would care for a bit more research
about overclocking? Same "resultant heat"? Come again?
Same family of CPU, same vcore, same frequency, results in
same heat generation. If you take a Xp2500 and raise the
FSB or multiplier till it appears to be a 3400, it produces
some heat as the genuine 3400 regardless of overclocking.
The simple fact is that your cooling system wasn't
effective... if it was overheating at all which we can't
even confirm since you didn't use appropriate measures to
confirm it, only a single old software which was not
supported nor implied by anyone to do adequate for the task.
All is left is just to question if you don't have some personal hidden
interests in the whole matter which makes you so biased and dismissive
of the information I've presented. Yes, there is a chance that I am
wrong, but I looked at facts, not dismissed the whole thing from the
top. Maybe you prefer systems with built-in obsolescence, though?
You have no reasonable justification for presuming
overclocked CPUs. You have an easy way to check the CPU for
it's spec'd speed. You have no reason to believe there's
any problem except that it's overheating.
You have wasted enough of our time, I hope the OEM sues you
if you persist instead of taking appropriate measures to
check your errors.
