My AQXEA Barton 2500+ @ 2.2GHz (NF7-S v2.0)

"Wayne Youngman" wrote
The only reason I had it enabled was it gave me quite a *boost* in Sandra
Memory tests. I'm not sure what you mean by *instabilities*. The only
way
I know something isn't working is from MemTest errors and Prime95 fails.
Do
you mean like sudden freezes and stuff?.

So what exactly will *disabling* CPU Interface give me? the chance for a
higher FSB, CPU overclock, memory timings? eh :P

_The_ major stability test for any 'general use' computer is actually using
it - Prime95 Etc. are good for testing specifics but you may not hit just
the right combination of circumstances that will crash the machine even
running a whole bunch of stability testers at the same time.

Use the thing day to day and if it keeps crashing on you then back off on
the overclock a bit or maybe try upping the Vcore a notch, that's all.

Ciao...

[UK]_Nick...
--
Nick M V Salmon Master Mariner MN(Retd.)
Mailto: My four
http://www.nmvs.btinternet.co.uk/index.htm

"Wayne Youngman" wrote
After running Prime95 for 2 days I can say the 2500+ @3200+
(2.2GHz) is completely stable.

Did you try running other stuff at the same time..? Sometimes that makes
Prime95 throw an error.


So I decided to try and get some more out of it.

stepped up to 11.5 x 200 (2.3GHz)

1.75v - FAIL Prime95 (within 1 minute)
1.775v - FAIL Prime95 (within 1 minute)
1.8v - FAIL Prime95 (within 1 minute)

I didn't try any higher Vcore than 1.8v. Hmmm I may have well hit my
chips
limit here.

Looks like it to me as though it may be - although you usually get some
minimal extra speed out of Voltage increases. 2.2Ghz does seem a little low
for (almost ;-) the latest stepping Barton 2500+to me though but YMMV as
they always say about overclocking - the only guaranteed speed from that CPU
is 1.83GHz..!


Cooling seems fine. I think I still have CPU
Interface/disconnect *enabled* in BIOS, but I switched off the spreads.

Try 11.5x200 at 1.8Vcore with both disabled as that 'CPU/FSB Interface'
especially may cause Prime95 to throw an error very quickly - those first
few seconds it's setting up a bunch of stuff in memory AFAIK. BTW: all
that 'CPU Disconnect' function does is allow the CPU to run cooler at the
idle - waste of time IMO, I prefer the CPU to stay at the same temp' myself,
ie. minimum thermal cycling.


Everything seems to be working peachy at 11 x 200 1.75v though, maybe this
is my sweetspot?

If you're happy with it, which you probably are after a 550MHz PIII (?) then
it's sweet - if you're like me and want that last possible ounce then keep
on tweakin'. ;-)


P.S: Could my 300w Antec PSU be helping Prime95 fail?

Maybe, it is a little low but it's a good quality unit & if it's a
'truepower' version then the 12V should be reading 12.0V, is it..?

Ciao...

[UK]_Nick...

"Wayne Youngman" wrote in message
...
"Nick M V Salmon" wrote
Double BTW: Why did you stop at 2.2GHz..? Most of those CPUs will do
more
than that, weird if it won't and I'd suspect something else before the
CPU.
My new Barton 2500+ (AQXEA) seems stable at 12x200MHz for 2400MHz but
not
a
midges more on the FSB - which suggests to me it'll be on the ragged
edge
at
that so I may back it off to 12.5x192 like the last one.

Could also be this particular NF7-s v2.0 and I'll be swapping the Barton
out
for a Tbred B known good at 220MHz FSB to crosscheck that. Starnge that
my
highest FSB clocking NF7 was a v1.2... Was being the operative word. :-(
But ya can't win 'em all and that's the very first mobo I lost to
overclocking...


Hi,

After running Prime95 for 2 days I can say the 2500+ @3200+ (2.2GHz) is
completely stable. So I decided to try and get some more out of it.

stepped up to 11.5 x 200 (2.3GHz)

1.75v - FAIL Prime95 (within 1 minute)
1.775v - FAIL Prime95 (within 1 minute)
1.8v - FAIL Prime95 (within 1 minute)

I didn't try any higher Vcore than 1.8v. Hmmm I may have well hit my
chips
limit here. Cooling seems fine. I think I still have CPU
Interface/disconnect *enabled* in BIOS, but I switched off the spreads.

Everything seems to be working peachy at 11 x 200 1.75v though, maybe this
is my sweetspot?

I get the same thing with my 2400+. 200x11 at 1.675v perfect but 11.5x200
will not run right no matter what the voltage. The MSI is limited to 1.8V
but still....

*Impression IM-21 Black Case w/Antec Truepower 431W PS*MSI K7N2
Delta-L*Athlon XP 2400+ at 200x11--1.675v*Vantec Aeroflow*2x256 OCZ PC3200EL
2.0-2-3-7 (2.6v)*ATI 9700 Pro*2x80GB WD "SE" riding a RocketRAID100*Lite-on
48/24/48/16 CD-RW/DVD combo*Audigy*Klipsch Promedia 4.1's*NEC MS97F-BK*XP
Pro SP1*Logitech Elite/MX700*BCC

"Nick M V Salmon" wrote
Did you try running other stuff at the same time..? Sometimes that makes
Prime95 throw an error.

Try 11.5x200 at 1.8Vcore with both disabled as that 'CPU/FSB Interface'
especially may cause Prime95 to throw an error very quickly - those first
few seconds it's setting up a bunch of stuff in memory AFAIK. BTW: all
that 'CPU Disconnect' function does is allow the CPU to run cooler at the
idle - waste of time IMO, I prefer the CPU to stay at the same temp'
myself,
ie. minimum thermal cycling.

Hi Nick,

yes I ran a fair few things while Prime95 was running, including 3D demos
and stuff, all is well (it seems!). Yes I am not 100% sure if I have
reached my limit on the overclock, I would like to try for 2.3/2.4GHz so I
will follow your suggestion and *disable* those two CPU options (Interface
and Disconnect) and try 11.5 x 200 @ 1.8v. .

Wayne ][

"Nick M V Salmon" wrote
Try 11.5x200 at 1.8Vcore with both disabled as that 'CPU/FSB Interface'
especially may cause Prime95 to throw an error very quickly - those first
few seconds it's setting up a bunch of stuff in memory AFAIK. BTW: all
that 'CPU Disconnect' function does is allow the CPU to run cooler at the
idle - waste of time IMO, I prefer the CPU to stay at the same temp'
myself,
ie. minimum thermal cycling.


Hi,

well as much as I tried it just wouldn't play any higher than 2.2GHz

Attempts @ 2.3GHz

11.5x200 @ 1.8v = Fail Prime95 fast
11.5x200 @ 1.825v=Fail Prime95 fast
11.5x200 @ 1.85v=Fail Prime95 after 3mins!

I suspect that it may make it if I dare go above 1.85v (go Aero7 Lite, go!)
but after reading about a recent 2500+ that just stopped working after a few
months @1.85v I think I will call it a day. An extra 370MHz is the best
overclock I did so far :P, hehe I remember paying nearly £200.00 for a
Pentium 200Mhz.

For the record I had all the options that affect stability *disabled*
(fastwrites,cpu disconnect/interface/spreads etc), and I only had adjusted
CPU and vDimm. . .didn't touch chipset and AGP voltages (that's ok right?).

I will now begin the final stage of CPU/Memory tweaking, and try and work in
a 217-220MHz-FSB into the equation,

Thanks all,

Wayne ][

"Wayne Youngman" wrote in message
...
well as much as I tried it just wouldn't play any higher than 2.2GHz

Attempts @ 2.3GHz

11.5x200 @ 1.8v = Fail Prime95 fast
11.5x200 @ 1.825v=Fail Prime95 fast
11.5x200 @ 1.85v=Fail Prime95 after 3mins!

I suspect that it may make it if I dare go above 1.85v (go Aero7 Lite,
go!)

Don't there's no point - that is quite a big step from 2200 to 2300 but the
extra 0.1V should have had more effect than pushing the fail time for
Prime95 out from zero to 3 minutes. Be happy with 'highest end yet sold'
speed for the cost of the lowest end Barton at only 1.75Vcore aircooled with
an £11 HSF. ;-)


but after reading about a recent 2500+ that just stopped working after a
few
months @1.85v I think I will call it a day.

Heh, that was mine - Peltier cooled and still not sure if the cause was
overvolts or damp - most likely damp really - it had a spot of water on it
when I looked at it just after it died and _very_ slight corrosion to the
1st L3 bridge, so I guess water & some form of short circuit is the most
likely.


An extra 370MHz is the best overclock I did so far :P,

You'd have had a shedload more overclock MHz with a 1.5Vcore spec'd 1700+ or
1800+ - Heh, I'm getting 60% O/Cs out of all three of mine - FSB O/C is none
to shabby at an average of 50% too. I hope to get even more out of the
latest XIUHB stepping from CPU City on Tuesday. :-) £45 for a ~2.4+Ghz
CPU. :-)))


hehe I remember paying nearly £200.00 for a
Pentium 200Mhz.

For the record I had all the options that affect stability *disabled*
(fastwrites,cpu disconnect/interface/spreads etc), and I only had adjusted
CPU and vDimm. . .didn't touch chipset and AGP voltages (that's ok
right?).

Noooooooo, up the chipset to the 1.7V max a glue a heatsink on the
Southbridge - I wondered why your overclock was so low with your temps so
low too...


I will now begin the final stage of CPU/Memory tweaking, and try and work
in
a 217-220MHz-FSB into the equation,

If you're doing anything important with the machine then leave well alone -
if you want to see wht it will do for the hell of it then;

1) back off the multiplier to 10x, make sure chipset Volts are at 1.7V &
memory at least 2.7V
2) back off the memory timings to the preset 'Optimal' - that way they'll
stay 'safe' as you up the FSB without your faddling with them - you can
speed em up a bit later on using 'expert' before doing a ny length of run in
Memtest86.
3) stick your Memtest86 floppy in the FDD.
4) leave all that crap disabled, as you wrote above.
5) Set FSB to 210 & reboot - wait for Memtest86 to load and run a few
iterations, not the whole test series - if you're close to the edge then
Memtest won't even load or it will error out.
6) Set FSB 215 & reboot plus ditto. (You may need 2.8Vddr up here, some
memory does, some doesn't.)
7) set 218 & reboot plus ditto - this is spec' for your PC3500 (it is 3500,
I can't remember which 'twister' you have.)
8) Now carry on _edging_ upward but back off the multiplier another 0.5 to
9.5 first - you're not testing for raw MHz here - only FSB and memory speed
capability of the CPU, Northbridge & memory itself. ie. keep the CPU at or
under 2.2GHz which you already know is stable at 1.75Vcore.

You may find that it starts to crap out at 210ish - I think AMD speed-binned
earlier Bartons for FSB ability, not raw CPU speed capability - I've had two
2500+ Barton's now and neither likes [liked for one of 'em. :-( ] much above
200 FSB on any of my motherboards - they'll do it for a benchmark or two but
they're not rock solid stable at anything above 200MHz...

You do know where the CMOS reset link is don't you..? LOL

Once you've found an upper limit for the FSB, work out the next multiplier
_up_ with a slightly reduced FSB and start running some tests. Don't go the
whole hog in Memtest86 as yet, let it do a couple passes of the standard
tests - the really thorough test is for later when you leave it running
overnight. If it passes then take the floppy out and let it boot into
Windows - try a run of SuperPI 16M - that will error fast if you're close to
the egde on memory. The run Prime95 for a while, no need for more than an
hour - you already know the CPU's arithmetic units should be good at this
speed - its the memory subsystem you're testing now.

Now, this may be the kicker - try a 3D app' like 3Dmark2001SE - that will
load the Northbridge like crazy - no crash, no artifacts..? Go out to BIOS
and enable those things you like and see if she will still do it all. Lower
memory timings etc, BUT, one step at a time so you _know_ what has screwed
things up. You're looking for errors with Memtest86 now - where's the
edge..? Ease your way up to it and then back of a safe distance before you
try to run any longer term tests...

Enough - you can figure the rest yourself but do remember what I wrote
earlier about the best test being longer term general use...

Have fun & Ciao...

[UK]_Nick...

"Nick M V Salmon" wrote
Don't there's no point - that is quite a big step from 2200 to 2300 but
the
extra 0.1V should have had more effect than pushing the fail time for
Prime95 out from zero to 3 minutes. Be happy with 'highest end yet sold'
speed for the cost of the lowest end Barton at only 1.75Vcore aircooled
with
an £11 HSF. ;-)

You'd have had a shedload more overclock MHz with a 1.5Vcore spec'd 1700+
or
1800+ - Heh, I'm getting 60% O/Cs out of all three of mine - FSB O/C is
none
to shabby at an average of 50% too. I hope to get even more out of the
latest XIUHB stepping from CPU City on Tuesday. :-) £45 for a ~2.4+Ghz
CPU. :-)))

Noooooooo, up the chipset to the 1.7V max a glue a heatsink on the
Southbridge - I wondered why your overclock was so low with your temps so
low too...
Have fun & Ciao...


Hi Nick,

I was wondering about that chipset voltage. I didn't adjust it because I
thought that it only affected the FSB overclocks? Ok so I will have another
go with all the extra options disabled *and* the chipset voltage pushed up.

BTW: I did test the FSB overclocks on this NF7-S just last week (ok I'm
slow!), I was 100% stable @ 217x 8.5 with my TwiSTER PC3500 @ 8,3,3,2.0
(SPD). I did have it booted into windows @ 225MHz-FSB (450DDR) but Sandra
threw a wobbly when I tried to test the memory, so then I went back to
MemTest and of course it threw errors at me. Does everyone have that
chipset voltage maxed? (1.7v).

Yes I *do* know where the CMOS reset is :P funnily enough I must have used
it 20 times when building this NF7-S and pushing the FSB. Then I read about
holding the insert key while booting to reset CMOS. . .doh!!!

Thanks Nick, I will give that last push a go now, hmmm I wonder how bumping
up chipset from 1.6v to1.7v can assist the CPU overclock?

Wayne ][

Wayne Youngman wrote:


"Jerry McBride" wrote
Running at 10 x 200 is a breeze and so far never a temp reading higher
than
36c. This is what it should be like... :')


Hi,

36°c load temps @11x200, that's pretty impressive! what Vcore are you
running that overclock at? what is your HSF, case temps?

Wayne ][

Hi Wayne...

Sorry for the late response...

I've been running my xp2500+ at 11x200 at 1.75 vcore for... what... a week
now? Cpu temp floats around 36c and a case temp of 27c. It's my normally
chilly office. :')


************************************************** ****************************
Registered Linux User Number 185956
http://groups.google.com/groups?hl=e...ff&group=linux
Join me in chat at #linux-users on irc.freenode.net
This email account no longers accepts attachments or messages containing
html.
6:12pm up 73 days, 4:10, 4 users, load average: 0.07, 0.08, 0.08

"Wayne Youngman" wrote
I was wondering about that chipset voltage. I didn't adjust it because I
thought that it only affected the FSB overclocks?

More stable chipset = better signals on the Northbridge = CPU bus =
higher CPU overclock. No tguaranteed but it won't hurt. Only thing is, I
got SATA and IDE errors when I pushed the chipset to 1.7V without a heatsink
on the Southbridge - I've written that before in threds replying to you but
I thought I'd better re-iterate knowing you don't have a heatsink on it yet.


Ok so I will have another
go with all the extra options disabled *and* the chipset voltage pushed
up.

BTW: I did test the FSB overclocks on this NF7-S just last week (ok I'm
slow!), I was 100% stable @ 217x 8.5 with my TwiSTER PC3500 @ 8,3,3,2.0
(SPD).

You want to have the CPU somewhere near it's most stable highest overclock
when you check for highest possible FSB / memory clocks...


I did have it booted into windows @ 225MHz-FSB (450DDR) but Sandra
threw a wobbly when I tried to test the memory, so then I went back to
MemTest and of course it threw errors at me. Does everyone have that
chipset voltage maxed? (1.7v).

Those errors will likely have been caused by low chipset Voltage.

You do realise that anyone running really 'silly high' FSBs (for NForce2)
longer-term will have Volt modded the board to get even higher chipset
Voltage..?

If you can run Memtest86 'All' test clean at 217MHz with the chipset at only
1.6V then that's pretty good - you may be able to run something like 9.5x233
for 2200MHz with chipset at 1.7V & the Vddr notched up a little - dunno what
timings that Twister will do though - I'd tend to stick with the 'Optimal'
setting initially whilst pushing the FSB up - then it will auto-backoff the
timings per the SPD for the speed you're at.

NB: Get a darned heatsink for the Southbridge before you screw up your
Windows install if you're going to run the higher chipset Voltage
long-term..!


Yes I *do* know where the CMOS reset is :P funnily enough I must have used
it 20 times when building this NF7-S and pushing the FSB. Then I read
about
holding the insert key while booting to reset CMOS. . .doh!!!

Doesn't always work if you've gone a little too far in one jump but it
usually does. Epox go one better still with their "watching dog" in BIOS -
if the board doesn't boot straight off it auto-resets to default CPU without
you having to do a thing.


Thanks Nick, I will give that last push a go now, hmmm I wonder how
bumping
up chipset from 1.6v to1.7v can assist the CPU overclock?

Signal stability...

Ciao...

[UK]_Nick...

running my 2500+ on NF7-S 2.0 at 215x11 at 1.9v, Prime95 ran for 38
hours with a max temp of about 47C with a Delta fan (maxed, have fan
on a radioshack pot) and MC462 heatsink.