XP2500 Barton Overclock Failure

"Simon J. Bale" wrote in message
...
My CPU idle temp is 43C and rises to 54C running the Prim95 torture test.
I
have 3x80mm case fans installed and the stock AMD HSF.

If those are external CPU diode readings that's high. CPU external 54C under
load means CPU internal diode probably in the mid 70's. Ditch that stock HSF
pronto and get something decent. I use a Vantech Aeroflow with AS5, and CPU
external (2500 @ 3200) under max load is 45C, internal diode temp is 49-50.

| multiplier is set to 11x and the CPU
| voltage is set as standard for the XP2500.

http://sysopt.earthweb.com/articles/...awb/index.html

Well, I'm getting the impression that heat may well be the problem, so I've
purchased a bigger heat sink and fan. Am I correct in thinking local hot
spots on the die could be causing the problem? The motherboard reports the
CPU temp to be around 45C when idle, but I'm assuming the thermistor isn't
located on the actual processor die so this temp could be misleading. I
can't see why the motherboard should be causing a problem because the 2500
and 3200 are fundamentally the same chip and the board should support 3200
speeds. I've also noted that the board runs the core voltage at 1.75V when
set to auto, is this usual, I thought the designed operating point was
1.65V.

"Simon Bale" wrote in message
...
Hi,

I have an XP2500 Barton core CPU and I would like to try and overclock it
to
2200Mhz. I have done some initial tests and I can't seem to overclock the
CPU at all. Increasing the FSB by only a few MHz causes windows to fail to
boot. An attempt to clock the FSB to 200MHz results in windows stating
that
it can't load several files from the hard disk during the initial stages
of
boot up. I realise that using the Via KT600 chipset means that I can not
lock the PCI bus frequency. When I bought the CPU a few months ago I
fitted
the heatsink incorrectly which caused several unexpected over-temp
shutdowns. I eventually realised and corrected the problem. The system is
very stable at the XP2500 speed but could the inability to clock to XP3200
speed be a result of damage to the CPU caused by the overheating or is it
more likely that another component is causing the problem.

NB: During the above tests the clock multiplier is set to 11x and the CPU
voltage is set as standard for the XP2500.

My set-up is as follows:

XP2500 Barton, AMD supplied HSF; Asus A7V600-X; 512MB Crucial PC3200 RAM;
Gainward FX5200 GFX.

Cheers, Simon Bale.

Simon ...
I'm unclear as to what settings you have tried in your bios.
Did you try raising your vcore to (as high as) 1.8v? AND ...
Did you try (at the same higher vcore) raising the vdimm (ddr voltage) to
2.8 or 2.9?
What have you set your ram timings to?

I don't think this is a heat issue ... not if you can run prime successfully
(54c really isn't that "dangerously" high).
and btw ... It's likely you'll never run any software that stresses your
computer anything near what prime does!

g'luck

"Simon J. Bale" wrote in message
...
Well, I'm getting the impression that heat may well be the problem, so
I've
purchased a bigger heat sink and fan. Am I correct in thinking local hot
spots on the die could be causing the problem? The motherboard reports the
CPU temp to be around 45C when idle, but I'm assuming the thermistor isn't
located on the actual processor die so this temp could be misleading. I
can't see why the motherboard should be causing a problem because the 2500
and 3200 are fundamentally the same chip and the board should support 3200
speeds. I've also noted that the board runs the core voltage at 1.75V when
set to auto, is this usual, I thought the designed operating point was
1.65V.

"Simon Bale" wrote in message
...
Hi,

I have an XP2500 Barton core CPU and I would like to try and overclock
it
to
2200Mhz. I have done some initial tests and I can't seem to overclock
the
CPU at all. Increasing the FSB by only a few MHz causes windows to fail
to
boot. An attempt to clock the FSB to 200MHz results in windows stating
that
it can't load several files from the hard disk during the initial stages
of
boot up. I realise that using the Via KT600 chipset means that I can not
lock the PCI bus frequency. When I bought the CPU a few months ago I
fitted
the heatsink incorrectly which caused several unexpected over-temp
shutdowns. I eventually realised and corrected the problem. The system
is
very stable at the XP2500 speed but could the inability to clock to
XP3200
speed be a result of damage to the CPU caused by the overheating or is
it
more likely that another component is causing the problem.

NB: During the above tests the clock multiplier is set to 11x and the
CPU
voltage is set as standard for the XP2500.

My set-up is as follows:

XP2500 Barton, AMD supplied HSF; Asus A7V600-X; 512MB Crucial PC3200
RAM;
Gainward FX5200 GFX.

Cheers, Simon Bale.

"iamnotme" wrote in message news:HSp8c.144972$Up2.69889@pd7tw1no...
Simon ...
I'm unclear as to what settings you have tried in your bios.
Did you try raising your vcore to (as high as) 1.8v? AND ...
Did you try (at the same higher vcore) raising the vdimm (ddr voltage) to
2.8 or 2.9?
What have you set your ram timings to?

I don't think this is a heat issue ... not if you can run prime successfully
(54c really isn't that "dangerously" high).
and btw ... It's likely you'll never run any software that stresses your
computer anything near what prime does!

g'luck

"Simon J. Bale" wrote in message
...
Well, I'm getting the impression that heat may well be the problem, so
I've
purchased a bigger heat sink and fan. Am I correct in thinking local hot
spots on the die could be causing the problem? The motherboard reports the
CPU temp to be around 45C when idle, but I'm assuming the thermistor isn't
located on the actual processor die so this temp could be misleading. I
can't see why the motherboard should be causing a problem because the 2500
and 3200 are fundamentally the same chip and the board should support 3200
speeds. I've also noted that the board runs the core voltage at 1.75V when
set to auto, is this usual, I thought the designed operating point was
1.65V.

"Simon Bale" wrote in message
...
Hi,

I have an XP2500 Barton core CPU and I would like to try and overclock
it
to
2200Mhz. I have done some initial tests and I can't seem to overclock
the
CPU at all. Increasing the FSB by only a few MHz causes windows to fail
to
boot. An attempt to clock the FSB to 200MHz results in windows stating
that
it can't load several files from the hard disk during the initial stages
of
boot up. I realise that using the Via KT600 chipset means that I can not
lock the PCI bus frequency. When I bought the CPU a few months ago I
fitted
the heatsink incorrectly which caused several unexpected over-temp
shutdowns. I eventually realised and corrected the problem. The system
is
very stable at the XP2500 speed but could the inability to clock to
XP3200
speed be a result of damage to the CPU caused by the overheating or is
it
more likely that another component is causing the problem.

NB: During the above tests the clock multiplier is set to 11x and the
CPU
voltage is set as standard for the XP2500.

My set-up is as follows:

XP2500 Barton, AMD supplied HSF; Asus A7V600-X; 512MB Crucial PC3200
RAM;
Gainward FX5200 GFX.

Cheers, Simon Bale.






Dont use booting windows as test of your hardware!!

If youre going to play get the tools to monitor it properly.

first off use memtest86 to see if you memory is solid at normal speeds
: crucial is good kit but your should test your mem in situ.

This test boots straight off a floppy, ie it does not load your
operating system so it is completely objective. If your BIOS can boot
you can run memtest86.

Next setup your temperature monitors (see below) BEFORE you start
over-clocking.

If temps are safe try some stress tools - I suggest cpuburn suite of
tests. Note the max temps you get.

Now start to push up the fsb bit by bit and use memtest86 bootdisk as
a first-line test when pushing fsb and other o/c parameters.

If it works OK, boot to DOS cmmand prompt (if you have some temp
monitors for DOS) and run some of the cpuburn tests. I find burnBX is
the first to show up problems.


Suggested tools:

memtest86 : very thorough memory check. Will pick up over agressive
clocking.

mbm5 to keep an eye on temps.

cpuburn for stress tools : tests more specific than just runnig prime.

If you can get 10min stable on burnBX and burnP6 you can be pretty
confident that Windows not be any less stable than it is designed to
be.

cpuburn tools are available for Linux as well so you could boot to a
SUSE or Gentoo liveCD and do you testing from a stable environment if
you prefer.

I've been running BurnP6 on my Gentoo Linux while I've been visiting
this news group.

My soak-tested stable settings : fsb=176 x 12.5 (2.2GHz) vcore 1.65
vio 3.65 command=fast. Barton XP2000+ KX7-333 PC2700
cpuburn 54C room=14C.

HTH.

I can tell you from personal experience that incorrectly seating a heatsink
CAN destroy your CPU. I'm looking at an XP 2000+ here beside me that I
screwed up by putting the heatsink on backwards. I hate to admit it,
because it's kind of a bonehead move, but such things happen. In my case,
after I corrected the mistake, the CPU showed up as an "Athlon XP 1700+
mobile" CPU. Then, over the next few months the performace deteriorated and
finally the computer ceased to boot. Replaced the CPU and all is fine.
Ouch.

jakesnake







"Simon J. Bale" wrote in message
...
I've tried the steps suggested previously and the processor will boot into
windows but crashes almost immediately after a login. Does this indicate
that I have an XP3200 which has been speed binned or is it likely that I
caused damage when I had an incorrectly fitted heatsink.

Are there any other steps that I could try to overclock the device?

Best Regards, Simon Bale.

"Wes Newell" wrote in message
newsan.2004.03.20.22.47.19.423484@TAKEOUTverizon .net...
On Sat, 20 Mar 2004 14:28:28 +0000, Simon Bale wrote:

XP2500 Barton, AMD supplied HSF; Asus A7V600-X; 512MB Crucial PC3200
RAM;
Gainward FX5200 GFX.

Here's what I'd do.
Make sure you hve plenty of power in your PSU.
Replace the stock HSF with something decent ($8 TR2-M1 should be fine)
Raise vcore to 1.85v
Raise ram voltage by .1 or .2
Slow down ram timings, try CL3.
Set FSB to 200 and give it a try.
Make sure it's using a 6 divider at 200MHz.
If that works, adjust accordingly.

--
Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
http://mysite.verizon.net/res0exft/cpu.htm