Archived from groups: alt.comp.hardware.overclocking,alt.comp.periphs.mainboard.abit (
More info?)
I see your point. based on my experience with Corsair XMS3500, at 5:4 I was
able to do 244 FSB on a 2.8C CPU (IC7-G board) and with Corsair XMS4400, I
am able to do that same fsb with ram at 1:1, albeit slightly looser timings.
however, with your experience with being able to do 218 at 1:1, that sort of
indicates that the ram is not the limiter...
--
Thomas Geery
Network+ certified
ftp://geerynet.d2g.com
ftp://68.98.180.8 Abit Mirror <----- Cable modem IP
This IP is dynamic so it *could* change!...
over 130,000 FTP users served!
^^^^^^^
"Jim" <null@null.com> wrote in message news:hBRec.1603$Yf6.479@fed1read07...
> Well, if I can't run, say, FSB 235 (5:4), thereby underclocking PC3200 to
> 188MHz (vs. 200MHz spec), yet, I can run FSB 218MHz (1:1), no problem, I
> think it's safe to say memory is at least an issue. Even totally relaxed
> memory timings maxs no difference. Now, is it possible that running
asynch
> (vs. sync) is itself causing a problem?! Maybe. IOW, if perhaps I had
> PC4000, maybe I *could* run FSB 250 (1:1). I've certainly considered it.
> Problem is, at these prices, who wants to run out an get PC4000 only to
end
> up tapping out a 230 (1:1)! That's my dilemma. Perhaps if I found a deal
> on PC4000, I'd give it a whirl, but right now, it's not even a remote
> possibility. I originally got the PC3200 for a mere $176! Even that's
now
> $286, on sale! no less.
>
> Trust me, I didn't go into the half of it regarding everything I tried. I
> reseated the Northbridge, disabled overdrive features, reduced memory
> timings, water cooling, changing DIMM slots, running HSF full bore,
running
> default BIOS settings, vDIMM up to 2.8v, vCore up to 1.8v!, to name a few,
> even took to literally blowing an 80mm fan directly on the Northbridge, it
> just doesn't matter. Unless my CPU is simply a dud, something on this
mobo
> is holding me back, I'd sure like to know what.
>
> Jim
>
>
> "TomG" <tgeery-NOSPAM-@cox.net> wrote in message
> news:4SPec.6853$55.3940@lakeread02...
> > not sure I agree that your ram is not the limiting factor, even at 5:4
> >
> > --
> >
> > Thomas Geery
> > Network+ certified
> >
> >
ftp://geerynet.d2g.com
> >
ftp://68.98.180.8 Abit Mirror <----- Cable modem IP
> > This IP is dynamic so it *could* change!...
> > over 130,000 FTP users served!
> > ^^^^^^^
> >
> >
> >
> >
> > "Jim" <null@null.com> wrote in message
> news:ZHJec.1290$Yf6.377@fed1read07...
> > > I have almost the same setup, AI7 (BIOS 16), P4 2.6C (800MHz), 1GB
> > Kingston
> > > PC3200, Coolermaster Aero 4 (copper, 2800RPM, very quiet), currently
> > running
> > > 218MHz (1:1).
> > >
> > > Be forewarned, the AI7 is a fine mobo, but the OC'ing ability of this
> > > particular mobo has proven to be inconsistent, more than I can recall
w/
> > any
> > > other Abit mobo. It seems either you get a GREAT mobo (insane OCs to
> 250,
> > > 260, or more), or mediocre results (230, if you're lucky).
> Unfortunately,
> > > I'm with many others in the latter category, the system simply will
not
> > run
> > > stable above 218MHz sync. I can just barely run 230MHz FSB w/ 5:4
> without
> > > XP hanging on boot-up, so it's not the memory. And believe me, I'm
VERY
> > > experienced and worked on this mobo for TWO MONTHS, to the point of
> > > exhaustion, it just won't do it. So just beware, keep your fingers
> > crosses.
> > >
> > > As I said, I can't run the CPU FSB above 230Mhz, and can't run the
> memory
> > > past 218MHz. So that leaves me little choice, either run CPU FSB
230MHz
> > > (5:4), and thus seriously underclock the memory, or run 218MHz (1:1)
and
> > > sacrifice some CPU OC'ing to keep the memory tapped out, and sync'd.
> Not
> > a
> > > great situation, my only option being to get some PC4000 perhaps. But
> at
> > > today's prices, and given I don't think I'll being seeing anything
more
> > than
> > > CPU FSB 230MHz anyway, I decided to stay put. I was *hoping* to run
CPU
> > FSB
> > > 250 (5:4), apparently this is a pipe-dream. Even tried water cooling,
> no
> > > dice. Yet others run the mobo 250 and up, I'm stumped.
> > >
> > > Since the P4 CPU is far more likely to be overclockable, I like to
> isolate
> > > the memory by overclocking the CPU FSB and memory in sync (1:1). IOW,
> try
> > > running 210, 220, 230, etc. In most cases, you'll tap out memory
before
> > the
> > > CPU (for me, that was 220MHz). I determined this using Memtest-86,
NOT
> > > Windows. Windows is too demanding, takes to long to reboot, etc. And
> > just
> > > because Memtest-86 runs clean, doesn't mean Windows will. Many
attempts
> > to
> > > OC in Memtest-86 ran fine, then hung Windows within seconds.
> > >
> > > I usually keep the memory at SPD until I tap it out, then start
losening
> > the
> > > timings, to see how much more I can gain. At that point, that's my
> upper
> > > memory limit. I then return to the SPD settings, then start
tightening
> > the
> > > timings, again, until it becomes unstable under Memtest-86, that's my
> > lower
> > > limit. When completed, I have a low/high range for the memory. Of
> > course,
> > > during this process, I'm adjusting vDimm to see if it helps, within
> > reason,
> > > as necessary.
> > >
> > > Now I turn to the CPU. In this case, I suggest runnung 5:4 so that
> memory
> > > now doesn't hinder the results (i.e., it's always underclocked, at
least
> > > until CPU FSB 250MHz). As with memory, I'm increasing voltage within
> > > reason, as necessary when instability occurs.
> > >
> > > Again, all this is under Memtest-86. When completed, I have a pretty
> good
> > > idea what the the CPU and memory can do, individually. I now try
> running
> > > Windows, which invariably can't handle the maximum OC under Memtest.
I
> > > start working backwards, reducing CPU and memory OC's until Windows
> > > stability returns. Once Windows loads and run (or appears to run)
> stable,
> > I
> > > run Prime95. Ultimately, I never accept OC'ing results until Prime95
> runs
> > > 24 hours, no errors.
> > >
> > > I strongly suggest keeping a diary, it really helps. You're adjusting
a
> > lot
> > > of variables, and it can get awfully difficult to keep track. The
last
> > > thing you want to do is spend two hours mucking w/ the system, then
lose
> > > track of where you are. List all the variable (FSB, memory, voltages,
> > > etc.), I even track memtest wall time to see if perhaps the OC is
> higher,
> > > but results are deminishing (sometimes happens on asynch CPU/DRAM
> ratio!).
> > >
> > > HTH
> > >
> > > Jim
> > >
> > >
> > > "Wayne Youngman" <waynes.spamtrap@tiscali.co.uk> wrote in message
> > > news:407b2a4d$1_2@mk-nntp-2.news.uk.tiscali.com...
> > > > Hi,
> > > >
> > > > I'm just going through my last *stability* tests, and would like to
> > begin
> > > > testing the overclocking aspects of my new rig.
> > > >
> > > > I wanted to get the machine *Rock-Solid* using stock settings, now I
> got
> > a
> > > > solid base to work from on my overclocking!
> > > >
> > > > Intel® Pentium® 4 - 2.6GHz
> > > > Intel® Retail HSF
> > > > ABIT AI7 (BIOS #16)
> > > > 512MB CORSAIR PC3200 (on Loan)
> > > > ANTEC TRUE550
> > > >
> > > > I also have a THERMALRIGHT SP-94 + 92mm YS-TECH on standby, but I
> wanna
> > > see
> > > > what I can do with the retail unit first.
> > > >
> > > > What would you say is the standard procedure for testing your CPU
and
> > > mobo's
> > > > FSB?. I know I should take the memory out the equation by setting
> loose
> > > > timings, and also running the 5:4 divider, but what about the CPU?.
> > > >
> > > > This is my first attempt at overclocking a Intel® P4, so I wanna
take
> it
> > > all
> > > > in. Do you guys say just push up the FSB like 5MHz at a time, and
> then
> > > try
> > > > to boot up and run some Prime95, then if you fail that, nudge up the
> > vCore
> > > a
> > > > bit?.
> > > >
> > > > I read some reports of people running great overclocks using stock
> vCore
> > > > (1.525v). I don't expect that myself, but I have been reading that
up
> > to
> > > > 1.65v is fairly safe, as long as proper cooling is in place.
> > > >
> > > > So is that what I should do, relax the memory and start pushing 5MHz
> at
> > a
> > > > time, then if I hit problems just bump up the vCore?
> > > > --
> > > > Wayne ][
> > > > <Intel® Pentium® 4 - Online!
>
> > > >
> > > >
> > >
> > >
> >
> >
>
>