Archived from groups: alt.comp.hardware.overclocking.amd,comp.sys.ibm.pc.hardware.chips (
More info?)
Bitstring <9561A8484537931E75@127.0.0.1>, from the wonderful person
Franklin <franklin_lo@mail.com> said
<snip>
>>> (Q.1) Can I simply raise the cpu voltage to get the Duron to
>>> go faster?
>>
>> Yes, right up until it burns out. However if you are raising
>> the FSB and not the multiplier (which requires the chip be
>> unlocked) then you are probably stressing the memory, PCI
>> peripherals, disc channels, etc. more than the CPU (so they'll
>> probably quit working first .. not destroyed since you are not
>> over-volting them any, although raising the FSB has been known
>> to trash disks on rare occasions, since corrupt disk writes are
>> possible).
>
>From what you say I think it seems that simply raising the voltage
>on my mobo results in speeding up the FSB. Is this
>understanding correct?
No, you need to increase the FSB or increase the multiplier. Since you
said you had it running at 730 Mhz I assumed you had increased the FSB.
The only point in raising the voltage is to enable the CPU to still work
at faster speeds. The manufacturer (AMD) tests these things to work at
the rated speed (700Mhz) at the rated temperature (95c core temp) at the
rated voltage (I forget what your was, but probably 1.75v +/- 10%).
Typically they run worst at the highest temp and lower voltage .. so
it'll do 700 Mhz at 1.6v and 95c, or it'll fail AMDs testing, but at
1.9v and -20c it'll probably do 900 Mhz or something similar .. AMD just
test it for meeting spec under worst conditions.
Sadly raising VCore raises power dissipation and thus temperature, so
you need some really fancy cooling (Peltiers etc.) to achieve the ideal
of 'higher voltage and minimum temperature both at once'.
>You seem to be quite right about the memory. In a spate of
>enthusiasm I made changes to the memory settings and also to the cpu
>voltage. (I didn't consider changing memory settings to be
>overclocking but I get I was too narrow in thinking of overclocking
>applying only to the cpu.)
>
>I increased the cpu voltage even more (to about 12% higher) and I
>very quickly got system instability. Blue screen on XP. So this is
>probably the memory struggling then, is it?
No telling what is falling over .. I supposed you were increasing the
FSB (which increases the speed memory has to go at). if you were just
winding up the voltages, I'd stop right now and go and do a lot more
research. There are copious forums and newsgroups dedicated to
overclocking (but it'll be how to get an XP3800 to go at XP4800 speeds,
not how to get a Duron 700 to go 20% faster, unless you go dig into the
back issues).
>
>> I'm not sure of that particular motherboard, but most (of that
>> age) derive the PCI clock (33 Mhz) by dividing down the FSB
>> clock (66, 100, 133, 166 or 200 Mhz). Modern boards or
>> 'overclocking friendly' boards lock the 33Mhz, independent of
>> what you tweak the FSB to.
>
>>> (Q.4) Is there an approximate calculation or estimate of what
>>> extra power my PSU will need to deliver as it is close to its
>>> max? (I really don't want to have my disk drives going into
>>> error because they are not getting enough power.)
>>
>> Not a big issue, the chances of you hitting the PSU limit due
>> to winding the CPU VCore up by 10% or 15% is pretty small.
>
>Seems that my memory settings were too aggressive then. Maybe I
>should work with the cpu first and see how far I can push it
>sensibly and then adjust memory timings.
>
>Is this the best way to do it in my case?
The best way to do it is normally to unlock the CPU at increase the
multiplier - that doesn't stress anything else at all. Where that isn't
possible, you increase the FSB, being aware that this may cause memory,
graphics cards, disks, and all sorts of other things to fail.
You increase the voltage if you have to to get the system stable - if
things get too hot, or if the system isn't stable at the higher voltage
either, you then reduce it to the minimum voltage which is stable at the
speed the thing will actually go at. Your Duron700 isn't going to go any
faster than 900 Mhz no matter what voltage you try to apply .. and
you'll be rather lucky if it will actually go that fast.
>
>> Personally I'm not sure why you'd bother, unless you want to
>> play .. you can buy a 3x faster processor for about $25, and
>> you can probably find something that'll work with your
>> motherboard 2x faster than what you have for peanuts
>> second-hand. Heck, I've =thrown away= Duron 800s that would run
>> at 1Ghz at 1.75v.
>
>You're right. But for now this is my main system. I find that it
>is probably working mostly just to keep the OS happy. The extra few
>percent power is therefore quite noticeable.
Nope, if you've got the OS tuned properly (even with WinXP) it doesn't
use very much CPU at all. Go do some more web research on what you can
turn off int he OS in oredr to get better system performance.
I personally would not dick around with my main machine .. overclocking,
especially if you are a novice at it, is quite likely to trash
something. Hope your backups are all current, and there's nothing on the
machine that you couldn't either re-create or live without. 8>.
Overclocking only makes sense when you either can't buy a faster CPU
(overclocking that XP3800+ to an XP4800+ for instance, since you can't
buy an XP4800+) or if the faster CPU is outrageously expensive.
Overclocking an obsolete Duron 700 to replace an equally obsolete Duron
900 (you'll never do much better than that) is a waste of lifespan.
--
GSV Three Minds in a Can
Outgoing Msgs are Turing Tested,and indistinguishable from human typing.
Facebook
Twitter
Instagram