Unofficial M919 486AVIP Vesa/ISA/PCI Mainboard Page
You bought it, had high hopes, found problems
and need answers. Help may be a click away!
Here's the scoop: you got your V1.x, and your V3.x.
Setup tips: AMD 5x86-150 3X OC, Ver 1.x Cache jumpers, Ver 3.x CPU
jumpers, PCI bus speed up, and notes on video corruption. A way to fix the VIP,
corporate info, and who to thank.
Other External Sources of Information:
For generic setup tips you should browse the BIOS Survival
Guide
. If your PC is sick you may find the help you need from
Computercraft's Upgrade and PC
Help page. Two other worthy links are the Computer Hardware
Performance Page and the ComputerNerd's Upgrade Help
page.
Last, but not least, are some practical upgrade guidelines
summarized on my very own PC
Upgrade Advice Page.
The place to look to check out old newsgroup posts
DejaNews is the place to look
(hey... didn't I just say that?). Finally, if you're fed up with the
VIP maybe Dick Perron's Other-Mother Board
page is for you!
hits so far. This page last updated
on 01-09-98.
(click here to view
on-deck items)
Send mail to the Webrewer of this page: SDND@AOL.COM
M919 VIP Main Board Identification and Features
Preface: Okay, I'll admit it, this motherboard is
ancient as far as MBs go. Over two years old and
still kicking. I love it when I make a good choice! I haven't
maintained this page in over a year now and yet people still find it
valuable. You'll find some of the links a bit
old (okay they're dead-ends) so bear with me as I fix them over the
next few weeks. The basic VIP info is good as it ever was...
Are you considering, or do you own a no-name 486-class VIP
(Vesa/ISA/PCI) motherboard?
This VIP motherboard has 1 Vesa, 4 ISA and 3 PCI slots and the UMC
8881/8886 chipset with onboard EIDE and serial/parallel ports. While
there may be others floating around out there,
so far I have seen two major configurations (and two? versions each) of
this main board:
- Vers 1.2: DIP socket cache, 4-72 Pin SIMM sockets.
- Vers 1.4: Soldered DIP cache with "Write-Back" chips (same basic
configuration as the 1.2).
- Vers 1.5: They're out there but I haven't seen one yet - any
problem reports?
- Vers 3.2: 2 Square "Write-Back" chips, a COAST cache slot, 2-72
and 4-30 pin SIMM sockets.
- Vers 3.2: Same as above without the "Write-Back" chips.
- Vers 3.3: They're out there but I haven't seen one yet.
- Vers 3.4: Back to 4-72 SIMM configuration.
You can identify this board by the silk-screened M919 in the corner
near the ISA slots (on 1.2/1.4/1.5 versions) or more reliably by the
V1.2, 1.4, 1.5, 3.2, 3.3, or 3.4 in the corner near the keyboard
connector. Early versions of this MB have demonstrated serial port
flaws, intermittent mouse operation, floppy disk access, and various
video problems. With a list of "features" like this you might be scared
to touch this MB.
But Wait!, this board is an excellent buy and it was exactly what I
was looking
for: a dirt cheap 486 class bridge between my old 486DX33 and a planned
Pentium
upgrade. This MB fit my needs so well that even though I knew it had
problems
I couldn't resist. It is possibly the most full featured 486
motherboard on
the market today. On board I/O, Enhanced IDE, tons of jumper settings,
and AMI
WinBIOS that gives you exquisite control over the MB-CPU-Cache-I/O
configuration. I've been successful at running a AMD-5x86P75-133 ADZ at
a
reliable 160 MHz overclock and it screams: PCBench 9.0 Dosmark 1039.
For under $120 this upgrade shames my loaded GW2K P5-90 at work and
benches solidly between P5-90 and P5-100 levels. I've even beat a
P5-100 using MATLAB's
"bench" utility which measures a mix of integer and float point math
intensive operations.
The Early Version 1.2/1.4 Problems
Now the downside: specifically, I'm referring to the problem where the
BIOS
CMOS settings have no effect on the serial port operational modes. You
may also experience serial communication corruption or error correcting
slowdowns due to corrupted serial port operation. Further, you may see
the Floppy Disk Controller take a hike or refuse to work as it should.
I contacted the board's distributor and they were aware of these
problems and they outlined the following fix.
First, do you have anything connected to the secondary EIDE port?
This
connection causes the serial port difficulty with the version 1.x
boards.
From reports I've received even a bare cable on this port causes
problems. Try removing it temporarily and then check (using DOS MSD) if
BIOS can now alter the port modes. This works for me - COM2 can be
disabled, but, my mouse still froze when MSD COM or IRQ options were
chosen.
For you 3.x sufferers, the secondary IDE port doesn't seem to bother
the system as much
but you may still see a smattering of these problems anyway. For the
lucky owners of the 3.4 or 1.5 VIP - I think they've finally licked
these I/O flaws, but, keep your eyes open for any problems and let me
know!
The Fix
This fix requires some bravery on your part. Locate the 4 surface
mounted chip
caps located between the first and second ISA slots. These slots are
the
VL/ISA near the edge and the one right next to it. The chip caps are
labeled
starting at C21, then C22 working towards the CPU. If you look
carefully the
C22 silk-screen is nestled along the VL slot parallel to the C22
location. Now
taking a deep breath, and a soldering iron (or sharp X-acto knife) you
want to
disconnect C22 from the resistor array by removing it or cutting the
trace.
Now, wipe the sweat off your face and relax - no more COM problems -
just try
to make warranty repair claim!
According to the manufacturer, C22 slows the serial bus down too
much and
removing it restores the proper timing. This fix WORKS!, however, I
have received one report that if you use the secondary IDE port the
problem is
reduced but not totally eliminated. I haven't confirmed this - but the
extra IDE cable length might play a role here.
Now a dose of reality: if you look closely there is little or no
room to work
in this area - the chip cap is nestled between two resistor arrays and
it is
virtually impossible to get access to either the trace or for a
desoldering
tool to remove the chip cap. With this tight fit cutting is probably
the best
approach. Using a magnifying glass and plenty of light, look for the
trace on
the VL socket side of the chip cap (the other side has a common
connection
trace to all four chip caps). You should be able to barely make out the
copper
trace just inside the white silk-screen boundary printing. You want to
sever
the trace as close to the solder pad of C22 by cutting with a very
sharp knife.
When it is through the trace you'll see small amounts of white powder
form
(glass-epoxy). Make sure to remove any metallic residue (use some
masking tape
wrapped on the end of a jeweler's screwdriver).
Note: As mentioned, there is evidence that the 3.x versions still
suffer from this set of problems. If you have a VIP version other than
1.x (like the 3.x) you will see the same parts layout but "C21" will be
numbered "C24" and "C22" has become "C25". More than one (V1.4 or 1.5)
reader has mentioned that "C21" is C5 on his board (the cap to cut: old
"C22" is now "C6") - I'll have to take a closer look when I get another
chance to find a V1.4/5. If you find you have problems and feel
adventurous, rest assured, the cut has been tried and it HAS worked.
This REVISED figure shows the problem/fix area
But I don't want to Cut It
Without the fix I personally had the V1.x serial port and mouse
problem. But even
with these flaws this MB is exceptionally stable under DOS and Windows
WFWG-
3.11. (But, I bet this causes Windows 95 plug & play problems
galore!) A simple solution for you (early) 1.x sufferers is not using
the Secondary IDE port. Unfortunately, I need it for a older hard disk
so my pre-cut non-invasive work around
uses COM4 for my modem port to eliminate bus contention between the
on-board and modem's default COM2 port. And the mouse - I'd just try to
avoid those problem applications! Again, if you have a serious case of
VIProblems, the cap clip helps.
Oh, by the way...
There have been a few instances where COM port(mouse) problems can be
traced to VIP installers who re-use the old cables in the PC's case
with the new MB. I'd recommend you change the cables with the main
board as a preventative measure - more than a few hairs have been
pulled because of header pinout mis-matches.
To wrap this issue up, I'd like to add that my 1.2 MB (with the cut)
still froze the mouse whenever I ran the COM or IRQ checks under MSD -
I'd need a full reboot to get the mouse working again. Strangely
enough, one reader found a Microsoft document that describes this
problem to a T:
CAUSE: This problem occurs when MSD un-synchronizes the mouse during
its hardware
scan of the serial ports.
RESOLUTION: Using the "MSD /I" switch usually allows
the mouse to continue functioning in
MSD until you choose either the COM Ports button or IRQ Status button.
To work around this problem, run MSD with the /I switch, or use the
keyboard in MSD. In either case, you should restart your computer after
you exit MSD to restore the serial port and IRQ states.
Gee, that's what I'd have to do anyway, some fix. To further lay the
blame on Micro(No, I don't want to go there today)soft, I recently
upgraded to a Logitech mouse.
The notable difference: with the Logitech mouse drivers I had no mouse
freeze-ups in MSD. Don't "start me up", but two things are obvious: 1)
MS doesn't have a clue about this, and 2) the Microsoft mouse software
drivers must be faulty.
What's New with the 3.2?
The latest version of this MB (Ver 3.x) is now on the street and I had
a chance to look at one recently. The VIP has grown four 30 pin and
lost two pin 72 SIMM sockets (you can also find 4-72 pin only
variations) and now has a re-arranged MB layout. According to Eurone
LA, this version has "fixed the COM port problems once and for all!"
However, they obviously didn't tell the rest of the world, because I
still get E-mail reports of COM problems with this MB. If you have a
3.3 version it is a incremental revision that had a different board
layout and this version was later superseded by the 3.2 MB and a simple
BIOS update (4-15-96).
The most interesting development: the M919's manufacturer has
changed the cache architecture of this MB. It now has one of those odd
brown sockets for a strip of RAM which the very sparse user pamphlet
indicates would hold either a write-back (now) or a pipeline cache
announced by the
cryptic phrase: "Future will available". With this cache slot there may
or may not be two surface mounted square memory modules with the now
infamous words "Write-Back" on them. If you have the square chips enjoy
the way they look
because that's what they do best. Close inspection shows the traces
around these
chips go absolutely nowhere.
I've asked Eurone about this latest version's cache configuration
and their response mentioned the socket was a proprietary design and
has no commonalty with the Pentium COAST modules. Make sure you have
the right one because one reader reports if you happen to swap one of
these modules with a Pentium COAST you can kiss your power supply
good-bye.
The scoop on the VIP pipeline module: you can say "so long" to this
scheme! The 5x86 is a low-end main board and there was minimal expected
demand for the incremental performance of the
pipeline cache. In other words: if you're trying to find a pipeline
COAST module - give up.
My Experience With the 3.2
At one time I got my grimey meathooks on a 3.2 MB with the 2/16/95 BIOS
date. I must say it fired up a lot quicker than my older 1.2 (but I
know what I'm doing now!). This baby came with the regular 256KB cache
module and I ran some tests with it installed and missing. Cachechk
reports that this module is faster than the 1.2's DIP configuration.
Initially, I had no COM or mouse problems and it ran just as fast as
the 1.2. But, later I started seeing occasional floppy disk controller
errors and a "missing mouse" error on bootup. In fact, I have even seen
the
LPT1 port disappear once. I've cut the cap on this bad-boy V3.2 and so
far haven't seen those
problems come back!
While I like the new layout the power sockets and all of the ports
are kind of cramped in the corner - I guess that keeps the cables in
one place. The pins for all of the external connections are now
arranged in a single row along the opposite corner and that's nice but
I had a bit of a problem getting my 3 wire turbo switch plug to fit
into a 2-wire pin header space.
But the SIMMs setup, that's another story! The two 72 pin SIMMS work
great for the "on the move to Pentium crowd", but I was really
surprised when I dropped a notoriously picky set (you must use 4
identical SIMMs) of 30 pin SIMMs from my old DX-33 system . The thing
fired up, saw the extra 4 Meg and ran without a hitch. Although I had
to adjust the timing to make the older 70 nS memory work with my 60 nS
72 pin SIMMs, this combination worked far better than those "other"
mother boards that give one or the other but not both memory
configurations at the same time!
Important Memory, um, er, oh... I forget!
A couple of words of caution: if you use the VIP with 72 pin SIMMs and
want to push both the clock and CMOS to the fastest possible settings,
then make sure you have identical 72 pin SIMMs. You may have difficulty
if you mix and match these 72 pin wonders. Besides, if you ever plan on
re-using your memory on a Pentium they have to be matched anyway. BTW,
identical SIMMs are those purchased at the same time from the same
place (same brand and hopefully lot). If you have the 30/72 pin combo
board, and you're already using 30 pin SIMMs, then to use both 72 pin
sockets you MUST have the same SIZE 72 pin SIMMs.
Finally, if your MB doesn't boot after messing with memory, then try
one or more of these tips:
- Remove all V/I/P cards, leave only the video card.
- Reduce to a single bank (1-72 or 4-30 pin) SIMM setup.
- Disable auto configuration of CPU/Memory BIOS settings.
- Back-off all DRAM/Cache timing to the lengthen wait/refresh
states.
- Disable L2 cache.
- If it works, then add new "types" of SIMMs one at a time.
- For dual 72 pin SIMMS installs try bank pair 0-2 or 1-3.
- Sneak up enabling BIOS features and pushing your CPU/Memory
settings.
EDO Memory and the VIP
Yeah sure, you thought all you had to do was change the CMOS settings
from "Normal" to
"EDO" and you'd be on your way running one of the newer DRAM fads.
Think again!
Whether the VIP takes EDO RAM depends upon the version of the
UM8881F fitted. There is a three letter code after the year/week code.
If the FIRST letter of
this code is 'E' then the board will take EDO. Other VIPs may have a
'B' code letter and it has been reported that during experiments with
EDO and this chip version it doesn't work and the cache chips became
extremely hot. Thanks to Ian for this tip.
VIP Memory Gluttony
Just because memory is cheap, a lot of you VIP users are packing way
over the sedate and economical 16 MB of RAM. It has come to my
attention that when you pig-out over 32MB your L2 cache mysteriously
(according to CACHECHK) refuses to work with memory above this level.
This phenomena has been reported in the USENET groups and the
resolution appears to be a 512K cache memory upgrade. I don't have the
bucks to experiment here so let me know what you find out. BTW, if you
got the COAST version of the VIP don't hold your breath - I kind of
doubt if 512K VIP COAST modules will ever be produced.
Forget Write Back or Write Through - Here's MY Cache Policy
Because the VIP, as well as other types of main boards, have been know
to contain decorative cache chipsets you should take several
precautions prior to handing over your cache cash (mail orderers will
have to take their chances). You should have a copy of Cachechk and
other checkout programs handy to assess the performance of your MB
upgrade anyway! Here is my cache of cache rules to live by:
- If your MB has soldered cache chips have your dealer run
Cachechk and really inspect those results - if he won't do this don't
buy it - a high percentage of soldered cache is non-functional.
- If it is a VIP 3.2 and it doesn't have a COAST strip ask where
to get one and how much it'll cost. When you get one do the Cachechk
routine to make sure it works. (as a baseline 8-15 nS chips cost around
US $40). For grins, ask when the pipeline cache is going to be
available.
- Make sure you have a money back full refund policy established
before you buy ANY main board that comes with cache - if the cache
turns out to be fake return the MB.
My Opinion - You Didn't Ask but You Get it Anyway!
Don't get me wrong, I think the VIP one of the best MBs as far as
features, demonstrated performance, and cost. They have even made it a
bit better for the upgraders by adding the 30 pin SIMM socket option.
Unfortunately, the cache module design excludes the possibility of
plugging
in your own selection of cache chips in the more contemporary DIP
sockets.
My read of the pipeline cache attempt was right on: too much for too
little way too late. IMO this was
a technical sales ploy to hype the oh-so-valued "pipeline cache" that
you all have heard about and really do need! Market pressure negated
this 486 development as demand shifted towards the Pentium MB designs.
Enjoy it while you can, these main boards and the CPU chips that
make them go are NOT destined to get cheaper and cheaper - they are
going to disappear entirely... Remember, you heard it here first!
BIOS and Mainboard Setup
The VIP board uses a AMI BIOS (WinBIOS as opposed to HighFlex) 10-10-94
core
date and (depending on delivery date) reports 11/08/95, 2/16/96,
03/06/96,
04/15/96, or 05/06/96 as the release date when the machine boots (The
latest flash BIOS is dated 05/31/96). This BIOS is chocked full o'
features so good luck. Three big tips:
- Although you can CMOS disable the secondary IDE port, there are
no user
selectable drive Type options in this BIOS release. The BIOS must
autodetect
secondary drives so be warned if your balks at autodetection. My crusty
Conner
drive posed no problems and the BIOS popped the Type settings easily.
AMI's
documentation on the 10-10-94 Core says that the next version will
provide
manual settings. Look for BIOS1010.ZIP on AMI's FTP server via AMI's Web Page. (Select
"downloadable documents" and then "other manuals".)
- If your machine exhibits random functionality each time you boot,
you may want to check if BIOS Chipset Autoconfiguration is enabled.
This feature is just like a roulette wheel: each time you boot you get
to spin the wheel and see what lady luck
will give you. My advice: use manual settings and keep conservative
until you get the thing running.
- Don't use EMM's HIGHSCAN or your AMI BIOS machine will not boot -
I don't know why but (see below), contrary to Nike's advice, Just
(Don't) Do It!
To examine CMOS settings (including hidden ones!) in detail a useful
utility is
AMIsetup an external setup program for 386/486/Pentium PCs with AMI's
HighFlex
BIOS or WinBIOS. Look on the Simtel Mirrors for the file AMIS2990.ZIP.
The latest revision of AMIsetup (2.99) is not a beta version, but
addresses WinBIOS features prior to the next major release. AMIsetup
was written by Robert Muchsel (muchsel@acm.org).
For an example of what this program can do, check this listing of my BIOS settings as
produced by AMIsetup. The same information can be viewed on my easy to
read formatted BIOS
page.
Latest BIOS Information
The 3.x Version of the MB's AMI WinBIOS is still the 10/10/94
core and the latest release should show a startup date of 5/31/96.
If your thinking of buying this MB make sure you a see the date
04/15/96 (or later) or else you may have a potential problems with
Sound Blaster PNP cards and you will need the Eurone PNP patch to fix
it. You can try to locate this file at the VIP MB sites or do a FTP of
this file from "ftp://users.aol.com/sdnd/pnpfix.zip".
For all of you Flash BIOS fans out there you should be aware that
AMI sells the BIOS core and a development kit and it is up to the MB
(or chipset) maker to whip up their own custom BIOS to match the design
that they've put together. That's why, except for extremely
popular MB/chipsets, you can't get BIOS upgrades just any ol' where.
Eurone told me I'd have to replace the chip to upgrade. Even though my
Ver 1.2 MB had jumpers for configuring the flash BIOS modes - I suspect
the only flash I'd experience is the electrical arc when I tried to do
it.
Lots of people have asked for BIOS upgrade information - sorry to say
the PNP patch is the only upgrade that EURONE generally offers.
M919 Flash BIOS files
Recently I got a tip from John Z. about a company called PC Chips who also distribute the
M919 VIP MB. The really interesting thing is that they have a web page
with some FLASH BIOS files that
you can download. They're
not the latest versions I've seen and you have to make sure you know
what BIOS EEPROM chip you have.
Another distributor, Fugutech, is also selling the VIP now. They
must have taken a clue from PC Chips because they too have site with
BIOS files for downloading. Take a look at Fugutech's VIP
page.
If you try a BIOS burn, let me know and I'll update the page! This
seems way too risky to me: clobber your BIOS without a backup? I
suspect you'd want to copy an image of the BIOS that works and have a
reliable way of restoring the chip before you take the plunge.
Consider that a bum re-load will render your VIP useless if it locks-up
and/or you have to reset it. You NEED a back-up method to generate a
good BIOS chip, but then, that's the safe and sane approach and you may
not be as cautious as me!
AMI BIOS and EMM HIGHSCAN Details
If you use the EMM386.EXE HIGHSCAN option when you setup your DOS based
system you probably want to include the parameter "X=F000-F7FF" or
avoid Memmaker's "aggressive high memory" scan option. Although
targeted at early AMI BIOSes, this recommendation from Microsoft seems
to still ring true because my Ver 1.2 always hangs when I select the
Memmaker option that activates HIGHSCAN. Here's what the boys and girls
in Redmond have to say:
If you are using the HIGHSCAN switch and it is incompatible with
your
machine, you may encounter one or more of the following symptoms:
- The machine may hang when you restart it.
- The machine may hang when you attempt to access your CMOS setup
program.
- You may receive "out of memory" errors because your CMOS settings
were overwritten.
- Microsoft Anti-Virus or Microsoft Anti-Virus for Windows may
incorrectly identify a virus in memory.
- You may have problems with device drivers or
terminate-and-stay-resident (TSR) programs that are loaded into upper
memory.
Some Interesting Software Finds
On one of my many mining expeditions on the Web, I found this wonderful
source of drivers that
you will want to add to your hotlist. Of particular value to VIP users
are UMC's PCI IDE drivers (file name
UMC8886B.EXE for DOS, Win 3.1,95,NT, O/S 2, or Unix). You can browse
the site yourself or FTP the file from
"ftp://users.aol.com/sdnd/UMC8886B.ZIP". One note of caution, I've
gotten many reports from Windows 3.x users who can't get the UMC
drivers to work with their system - any clues?
If you need to debug lingering VIP serial I/O problems you may want
to browse Computer
Telecommunications System's page or FTP the Serial Port Utilities
(SPU) file CTSSPU34.ZIP from "ftp://ftp.comminfo.com/users/www00402".
The CTS SPU is a collection of COM Port setup and control utilities
for anyone with a serial port problems. From the looks of it this
program should be OEM standard issue software for VIP sufferers.
Video Problems
Sad to say but the VIP has some reported video problems. There is a
(remote) possibility that problems with PCI video "pixelation" and
"color pallet" errors in some DOS based games could be caused by
overclocking. Running your standard "33 MHz" PCI card at an out-of-spec
40+ MHz may be at fault. Seems like a luck of the draw situation if you
have this particular problem - I use the VL slot for video cause I knew
it'd hit 40 MHz without a squawk. Other's have reported no problems
even up to 50 MHz.
Aside from that clocking problem, I have been running across regular
"chips" and "video" news postings and receiving E-mail messages that
have a common thread: a user's video is hosed (corrupted) especially
when playing games like DOOM. I submit that if you're playing DOOM it
may not be just your video card that's corrupted! ;-)
A common thread appears to be the use of digitized sounds with the
DOS extenders (like Phar Lap or Rational Systems' DOS/4GW) used by a
majority of DOS games. Perhaps digitized sounds use DMA modes that
overlap with graphic card DMA. One suggested solution is to alter the
DMA channel setup on your sound card. Odd as it sounds - this fix has
worked for some! While this may be clutching at straws, it doesn't look
promising that a simple fix exists - any help?...
Your Input on Video Problems
Because this problem is SO random, I am performing a running survey to
gather your VIP video experiences. If you are having problems E-mail me
with the subject VIP-Video PR. Include a note about
your OS, video card type, video bus, sound card, CPU type, speed, and
clock, and then add a short line about your general symptoms (what
program and if it uses a DOS extender).
If you had problems and got them fixed, then E-mail me with the
subject VIP-Video PR Fixed. Include the above info
plus what you
did to solve your problem.
As a result of this on-going survey, I have compiled a list of VIP compatible video cards
based on your reports. (Want a clue? - the problem seems to start with
a V.)
SAD NEWS: It doesn't look like this problem can be
solved on existing VIPs (it appears to be a flaw of the UMC chipset).
Check this link out for details and a last ditch checklist of things to
try to solve your VIP-Video problem.
A new VIP-Video problem: as if the pallet changes
weren't enough, I have been getting reports of a new breed of V3.4B/F
video problems. In this situation
owners of PCI video cards have been reporting that
their VIP refuses to boot and gives them the ol' eight beep start-up
failure prior to hanging with a black screen.
I don't have the particular card IDs identified or collated right now -
but if you have experienced this please let me know. BTW the solution
is to replace the PCI card with either a new PCI or a throwback ISA
(yuk) card. My recommendation for you low-buck VIP
users is the generic Trident 9680 PCI with 2 MB. Average DOS and
reasonable Windows
performance for under $40 (US) - not too shabby.
Problems with SCSI Bus Mastering and AMD-133 CPUs
A whole slice of the world is experiencing problems with SCSI bus
mastering and the enhanced AMD 5x86 series of chips. The solution:
disable write-back caching and use the write-through option only.
You'll take a hit on performance but a functional system runs circles
around a non-working one!
If you read the AMD 5x86 CPU manual you will find that a
busmastering device takes control of the system memory from the CPU.
When it does this the busmaster is responsible for maintaining both the
memory and cache state as it does it's job. Unfortunately, the
busmaster design must flush and refresh the cache when it takes and
returns memory control to the CPU. This handoff process works well in
write through (the cache and main memory are exact copies) but falls
apart in write back (which tracks cache changes that have to be written
back to main memory when the cache is no longer valid).
The AMD write back scheme and common busmaster designs are simply
incompatible.
My Sound is Screwed Up when I overclock!
I overclock my 586 to 160 MHz from time to time. Because my SB16 - ASP
sound
card doesn't like this speed under certain games I use a clever
workaround:
You know that worthless keyboard lock on your PC? I just connect that
switch to the "A"
jumper on the CPU clock selector JP7 block. When locked I boot at 133
MHz and
everything works normally. (This is because at startup BIOS squirrels
away
some bus timing parameters and sets up your PC based on the CPU clock
at boot
time.) When I feel the need for speed, I unlock 'er and have a 160MHz
bit
blaster for those hard to reach spots. Haven't had a hint of a problem
switching the CPU speed smack dab in the middle of a running
application - the
better "Turbo Button". Actually, when you consider the information
below, it appears that the ISA is running too slow when booted
at 160 MHz!
PCI (hard drive) Speed Boost
For the fastest PCI bus, boot the VIP at 133 MHz and then switch the
jumper to overclock to 160 MHz. I do this routinely to avoid problems
with my Sound Blaster on the ISA bus. But then I discovered the
computer ran faster in this configuration.
The way I figure it the VIP's AMI WinBios must set up shop based on the
boot speed. It seems the PCI speed is 20 MHz after a 160 boot and 33
MHz when booted in the 133 mode.
HOT switching to 160 MHz must keep the dividers disabled and you end
up with a actual 40 MHz PCI bus speed. Connect a switch (the unused
keyboard lock works for me) to the "A" jumper on the CPU clock selector
JP7 block. Switch
on=133 off=160. My hard drive really perks up when I do this: take a
look at the dismal 160/160 boot/operation rates.
NOTE: if you have the 3.x version of the
motherboard you will notice that the CPU clock jumpers are labeled
differently: JP3 instead of Version 1.x's JP7.
Hard Disk Performance KB/Sec
Boot |
Run |
SI-7 |
CORE |
Comp |
PCB9 |
160 |
133 |
1765 |
2925 |
2589 |
781 |
160 |
160 |
1954 |
3577 |
2972 |
927 |
133 |
133 |
2509 |
5083 |
3987 |
838 |
133 |
160 |
2594 |
6235 |
4500 |
931 |
Notes:
- SI-7 Norton SI v7 Hard Drive Test
- CORE CORE test v2.6
- Comp Comptest v2.59
- PCB9 Ziff Davis PC Bench 9.0 (w/Smartdrv)
- Minimal config.sys and autoexec.bat.
- No Smartdrv for SI/CORE/Comp.
PCI Speed Boost: The rest of the story OR Reality Bites
For PC Bench 9.0, however, "Diskscore" shows only a very slight
improvement
when running with a Smartdrv cache (2 MB). I don't like to run PCB
without
Smartdrive 'cause it takes soooo looong. Further, compared to the other
benchmarks PCB9 causes very little stirring of the disk - lots of cache
hits. The other benchmark tests really create a racket - more direct
disk access I suspect. So if you use a disk cache your PCI speed up
might be less dramatic. All in all, this is the ONLY way to boot a VIP!
Another Option For Those With the "Can't Run 160 MHz" Blues
If your machine won't run at 160 or you don't like the switch over
option, you can speed up the PCI bus by running at 50 MHz with the 3x
multiplier for a 150 MHz CPU speed - it will perk things up on the I/O
side (PCI at 25MHz, VLB at 50 MHz) but performance suffers slightly at
CPU (150MHz) intensive tasks.
How to setup a AMD5x86 for 150 MHz operation
Okay, you may ask, "How do I set up my VIP (Ver 1.x) for 3x operation?"
If you're running the
AMD 5x86 P75-133 CPU the answer is simple: just run the beast as an Enhanced
AMD DX4 (remember Intel's ploy 3 x DX = DX4). You can do this by
removing JP33. Sad to say this is where my ancient VL video card fell
off the bus. I can run my bus at 25, 33, or 40, but not at 50 MHz.
Maybe your luck will hold. I think I'll go cry...
BTW, if you have the AMD 5x86 and V3.x MB just pull the plug on JP6
to go from 4X to 3X operation.
NOTE: the AMD5x86-P75 CPU runs with one of two MB clock multiplier
jumpers: 2x (which enables AMD's on-chip 4x multiplier) or 3x (which
gives a straight 3x). If you have a Cyrix CPU and you set the MB for 2x
or 3x your CPU will run at 2x or 3x - AFAIK, there is no Cyrix
equivalent to AMD's 4x step-up.
Hey, How About a 180+ MHz AMD 5x86?
Okay, there is now more than one brave soul out there that insists that
you can generate a 60 MHz bus speed by pulling all but the "B" jumper
from the CPU speed jumpers (V1.x JP7 or V3.x JP3). Setting the CPU to
run in 3x mode results in a 30 MHz PCI, 60 MHz VL, and a 180
MHz CPU setup. Boot at 133 and then switch to 180 and you
might end up with a 60 MHz PCI, 60 MHz VL, and 180 MHz CPU or you just
may reduce your PC to a smoking hulk of silicon. Talk about potential
electromigration and asking for trouble. I have tried with a newer PCI
video card, but couldn't find the right mix of CMOS settings and secret
incantations to get it to boot. I heard
that if you draw down the BIOS to a clean state (use the transit
shorting jumper) then boot it works better. I'm not up to this myself
so if you try it, let me know if and how it works for you!
Version 1.x Cache Jumper Settings
First make sure your cache is real, use CACHECHK 4.0 and study the
numbers you
get with and without cache enabled (a software switch in the options).
You
should see the memory times step off at the 16K internal L1 limit and
again at
the 256K L2 limit (your cache size here). If they're constant across
the
screen you got faked out or have disabled you cache in CMOS. Another
method
uses the CMOS cache settings and a reliable performance test program.
Just boot
to BIOS, disable the L2 cache and see if your benchmark program detects
the
difference. this second method isn't as elegant as CACHECHK but it
comes free
with your VIP mainboard.
M919 V1.x 486 Class VIP MB Cache Memory Setup
Single Bank Cache Jumpers
1 Bank |
Config |
JP8 |
JP8 |
JP8 |
JP8 |
JP9A |
JP9B |
JP9C |
JP9D |
512K |
128K x 8 |
1-2 |
3-4 |
5-6 |
7-8 |
Off |
On |
On |
On |
256K |
64K x 8 |
1-2 |
3-4 |
5-6 |
- |
Off |
Off |
On |
On |
128K |
32K x 8 |
1-2 |
3-4 |
- |
- |
Off |
Off |
Off |
On |
|
|
|
|
|
|
|
|
|
|
Dual Bank Cache Jumpers
2 Banks |
Configuration |
JP8 |
JP8 |
JP8 |
JP8 |
JP9A |
JP9B |
JP9C |
JP9D |
1024K |
128K x 8 |
2-3 |
4-5 |
6-7 |
8-9 |
On |
On |
On |
On |
512K |
64K x 8 |
2-3 |
4-5 |
6-7 |
- |
Off |
On |
On |
On |
256K |
32K x 8 |
2-3 |
4-5 |
- |
- |
Off |
Off |
On |
On |
128K |
16K x 8 |
2-3 |
- |
- |
- |
Off |
Off |
Off |
On |
M919's Alter Egos
Psst, you want to know a secret?... promise not to tell?
The Eurone VIP 1.x is the Amptron DX-9500 and the Eurone VIP 3.x is
the Amptron DX-9700.
How did this happen? Well informed sources tell me that the same
manufacturer supplies both Amptron and EURONE LA with this motherboard
design. Eurone LA is the one of the largest US distributors of UMC
products and, according to my contact, sells significantly more of
these MBs than Amptron. Lately, however, there seems
to be multiple "primary" outlets for the VIP (the PC Chips group and
Fugutech). As it turns out PC Chips (a group of 5 companies) appears to
be the originator of the M919 VIP. BTW, if you want to try to contact
the chipset maker using an Internet search on "UMC" you'll need to use
"United Methodist Church" as an excluded keyword phrase.
You can contact Fugutech,
PC Chips, or try EURONE's
relatively new a web site: Eurone's
Home Page. But, because it is new and developing content, and we
want MORE information, you might consider checking out Amptron's Product Page. If
you think you'd like to download the MB manuals from their site, good
luck, they look surprisingly similar to the Eurone brochures. I
certainly hope these MB guys aren't on the cutting edge of a trend - at
least those books they used to write looked like might (and sometimes
actually did) have some information in them!
Disclaimer: Amptron probably holds all sorts of copyrights so I hereby
acknowledge and admit that the VIP CPU Jumper Diagram
is from their site. (Note that I removed this diagram to a separate
page to load faster and
give you the option of viewing it.)
On my last few trips to Amptron's site, I noticed that they are
slowly removing references to the 9700 from the product listing. They
still have the manuals for downloading, but, I wonder what is
going on with the 9700 disappearing act?
How Did This Page Originate?
Hey, I can't claim to know everything and I certainly didn't discover
all of this myself. I had studied 486 MB upgrades for many months
before purchasing the VIP. When I found it (warts and all) I liked what
I saw and am very happy with my choice.
Fortunately, a very active Usenet and E-mail community allowed me to
find and collect the information I've reported in this page. From users
to manufacturers thanks for all your help.
I'm an engineer at a local aerospace company and have been working
on PCs since the old Commodore PET 2001 in 1978. I really like to putz
with PCs and look at it as the
1990s version of the shade tree mechanic - tweaking and fiddling to
wring the last erg
of performance out of underdog equipment. The way I see it anybody can
plop down a few
grand for a fast system - but working on a shoestring well away from
the foreskin of technology winds my clock! Good luck to all of you
VIPioneers out there - it is a very fast board when setup just right.
I'll answer any questions I can and like to hear of your experiences.
This page is a collection of the user reported problem set for this MB
so send me a line about any specific troubles you may have had. And if
you want, just e-mail me if you've been helped by this info!
NOTE: I can't be responsible for the accuracy of this information,
nor can I promise
that you'll be able to solve all of your problems. Further, you must
consider and understand you do any modifications at your own risk.
Sorry, but lawyers would probably eat me alive if someone toasted their
PC or themselves trying the stuff I've mentioned here!
Steve Dabrowski
Michigan, USA