Crankin' It Up
Why do I
run lousy DOS tests? I have a theory; to date it's proven moderately useful
and certainly saves hard drives from data corruption. The theory goes that
if you crank the voltage up to 15% over spec (in this case 1.70 volts), shave
off about 10% of the max speed obtained and then run at that speed with default
voltage. When I tested the 366@550 I found that it maxed out at 616 MHz (112x5.5)
at 2.2-2.3 volts. Shave off 10% (62 MHz) and you wind up with 554 MHz, which
is close enough to 550 for me. It ran at 550 at 2.0 volts for 4 months before
being retired. The P3 650 I tested in April did not appear to jibe with my
theory though. The C1 and now the C2 series of chips do follow my theory.
The numbers quoted for Vcore are the minimum, not the maximum required for
stable operation.
DOS
Tests (boot to a floppy and run cache check 7)
|
FSB
(MHz)
|
Speed
(MHz)
|
Vcore
(Volts)
|
Chip
Temp (F)
|
Case
Temp (F)
|
|
66
|
533
|
1.50
|
78
|
75
|
|
100
|
800
|
1.50
|
84
|
80
|
|
110
|
880
|
1.50
|
86
|
82
|
|
112
|
896
|
1.55
|
87
|
82
|
|
115
|
920
|
1.60
|
89
|
82
|
Sailing past 800 without
so much as a whimper, I briefly fantasized about reaching the magic 1 GHz
mark. Then, default voltage refused to boot at anything higher than 880. Bummer.
Still, 920 Mhz is a tremendous feat from silicon rated by Intel to be stable
at 533 MHz. Notice that 1.60 volts was the minimum to pass the cache check
tests. Increasing to the Abit's maximum of 1.70 volts produced no further
gain in speed.
Windows
98SE Tests (stable with all 2D/3D applications)
|
FSB
(MHz)
|
Speed
(MHz)
|
Vcore
(Volts)
|
Chip
Temp (F)
|
Case
Temp (F)
|
|
66
|
533
|
1.50
|
95
|
89
|
|
100
|
800
|
1.50
|
95
|
90
|
|
103
|
824
|
1.55
|
99
|
91
|
|
105
|
840
|
1.60
|
100
|
92
|
|
110
|
880
|
1.70
|
102
|
93
|
Nothing to
be ashamed about here. Fully stable at 800 MHz at 1.50 volts, the 533 gains
266 MHz just from a FSB adjustment. I settled on 840 Mhz at 1.60 Volts. Note
that I used no adapter whatsoever. Adapters, by any manufacturer, introduce
some level of electrical noise just by the extra connections present. Stout
power supplies are also required at these high speeds. Don't expect that your
HP Pavilion box, etc., sporting a wimpy 150 Watt supply could even begin to
handle the close to 16 Amp peak current draw from the +5V line for any length
of time.
