"Hi Mike,
As a frequent user of this site of late I am glad to be able to make a
small contribution. I have successfully boosted the core voltage of the
Yikes!' G4 400 Mhz cpu to run at 500 Mhz.
As posted on the G4 Forum page, I was not able to get my Yikes! 400 to
run at even 450 Mhz for any length of time by simply changing the
jumpers. Although with the adjustment to the heatsink and fan attachment
I was able to drop the reported cpu temp range from 28-32C to 24-28C,
the machine would freeze with any rise in ambient room temperature
beyond say 24C. I decided that in my case it would be necessary to boost
the cpu core voltage.
I studied the two articles here on boosting the G3 voltage [linked on the CPU Upgrades page-Mike], paying
careful attention to the location of the controling resistors in the one
article with pictures of the G3 ZIF module. I compared the Yikes!' G4
module to the G3 one in the article as well as two out of three other G3
ZIFs I had on hand.
As correctly extrapolated in the more recent of the two articles, none
of the modules were identical in layout. However, after comparing all
the modules and referencing the settings of various candidate resistors
against the PCM/Voltage Settings table from the articles, and the
Motorola specification pages for G3/G4 core voltages, I concluded that
the correct grouping of the five resistors would be near the NO PIN
corner of the ZIF board as evidenced in the four samples.
Also as reported in the original article, some boards indeed had 0 ohm
resistors, some with 1k ohm. But in all cases the resistor represented 0
in the VID code.
On my particular G4 module, the controlling resistors were 43, 25, 26,
27, 28, in the location described above. No 43 was populated which
translated to 2.10v according to the table (11110) and confirmed by
Motorola's specification of 1.8/2.15v. Note that 0 = resistor and reads
backward.
Judging by the previous results as obtained with the G3s I at first
decided to not exceed 2.5v where 2.6v was the maximum recommended core
voltage range even for the higher volt G3 (PPC 750). Additionally, given
that I had only the one 1k ohm resistor that came with the module, the
only permutation available to me was the VID coding of 11011 which would
result in the core voltage of 2.4v. Quite simply, I moved the surface
mount resistor from position 43 to position 26. And presto!
The MacBench 4.0 scores with the cpu at 449 Mhz and the cache running
1.5:1 at 299 Mhz were: cpu 1480, fpu 1244, disk 1236. Note that the
disk was an upgraded Maxtor 27 GB 7200 RPM. The scores at 399/266 were:
cpu 1319, fpu 1111, disk 1208. As reported by the G3 article, my
temperature reading dropped with the rise in cpu core voltage. According
to that article, the actual temperature mightbe more than 20C higher,
while the report was only 20C. However, my fan blown heatsink felt
barely warm to the touch.
Well, knowing me, after much philosophizing that it wouldn't be a good
idea to even try to run at 500 Mhz because of the wellknown errata bug,
I went ahead. The machine died with extensions loading. Never one to
take no for an answer, I decided to increase the cpu voltage to within
limit. First I was going to wait to try to find some surface mount
resistors in town. But having tried once before without success I gave
in and liberated one from the more current 400 Mhz G3 I just got last
week where it had been overclocked to 466 Mhz with complete success in
the Beige G3 DT. This would not matter to the G3 because then I also
thought to take it beyond 500 Mhz as well. I chose 2.6v to gain the
maximum headroom possible for error-free operation.
The VID coding for this would be 111001 = 2.6v Vdd. On my particular
module, it meant populating the 25, 26 resistor locations in the series,
43, 25, 26, 27, 28. As now fully confirmed by the excellent result, this
was the definitive setting. The machine booted quickly and I obtained
the following MacBench 4.0 scores: cpu 1586, fpu 1375, disk 1252. (Note
again that the disk is an upgraded unit.) Needless to say, I was on
cloud nine, which was why you were subjected to a barrage of emails from
moi at 3 AM.
Since then I have subjected the machine to constantly running Graphing
Calculator Full Demo, Netscape, MacBench, ..., switching in and out of
programs, quitting and opening, visiting lots of sites in browsing,
sending and receiving mails, etc. I HAVE NOT HAD ONE CRASH! When I
started it up this morning the temp read 16C upon bootup and quickly
climbed by one or two degrees increments to 20C. The reported
temperature at rest is always 20C, climbing to 24C when running Graphing
Calculator Full Demo in the foreground, going back to 20C even as GC
Full Demo runs visibly in the background. But as noted earlier, there's
probably a corrective factor of 20C so the actual operating temps should
be in the 40s.
As you've brought up the issue of cache corruption and the errata bug,
I've successfully mounted a 34.6 MB MacOS 8.6 Update.smi. I have hereby
enclosed various screen shots documenting these facts. I must report a
small incident, however. The file was transferred over an Ethernet 100
BT network. It was slow using the G4 as the server and dragging the file
from the 8500 end to the G4. The transfer also resulted in a corruption
when I tried to mount the SMI. But when I used the 8500 as the server
and drag the file from the G4 end, the transfer went very fast and
resulted in no corruption at mounting. I would not be quick to blame the
500 Mhz speed setting though as this behavior has been observed before
with various machines, notably the Sawtoothless 400 open box which was
in my possession for a couple o' days. Frankly, I was confident that it
would pass this test and never thought to do it until you brought it up.
Afterall, I'd been running GC Full Demo constantly and subjecting the
machine to all kinds of actions and simultaneous running of several
tasks piled on and I could not provoke a crash! Wow!
Mikey, you're reading from one very happy reader. I wish other
AccelerateYourMac readers a similar success in their quest for greater
speed under their choice operating system, without paying the heavy toll
to the multi-zillionaire Mr. Jobs and the greedy Apple Computer
Corporation. $451 more for a 50 Mhz increase to 450 Mhz at the Apple
Store? Phooey! The Yikes! 400 is the biggest bang for the buck ever.
Payless and get more. It's the new paradigm. The PC world is firmly in
this new era. Apple Computer will be dragged along kicking and
screaming. The 800 Mhz Athlon soon to come out in a month or so, and
even more so the 1 Ghz version coming out early next year will see to
it. And I am certain that AccelerateYourMac will be there to help even
the playing field.
Thank you for a great site that has truly been an inspiration to us all.
Tarn
"
