VOGONS


First post, by Boctor

User metadata
Rank Newbie
Rank
Newbie

Recently, I have been trying to install Windows NT4, 2000, or even XP (purely for the sake of testing) on a build with an ECS P6FX1-A motherboard, revision 1.2. For a few years, I have only used this computer and board with Windows 98SE, and everything has worked perfectly, even for playing games like Quake 2. I found this board in a computer an office was getting rid of, running Windows 95. It was still completely working, and spent practically its entire life on the same desk. Being curious about the reported performance improvements from a pure 32-bit operating system on a Pentium Pro, I decided to try my luck. Other forum posts about this motherboard occasionally mentioned running Windows 2000 on it, or at least NT at all. I already have a Super Socket 7 box for 16-bit DOS things, so why not?

Unfortunately, every NT-based installer I've tried seems to freeze during the initial loading, before even reaching the formatting or file copying stages. 2000/XP will freeze when trying to start Windows, toward the end. NT4.0 freezes while just loading things on floppy 2 of 3, usually the FAT, keyboard, or video driver. I stopped lurking and finally registered, in hopes I could at least get a second opinion. Best case scenario, it's something obvious that someone with relevant experience can isolate. Worst case, at least I know this motherboard isn't worth more of my time. There are no visible problems with the motherboard, physically. It appears to be in excellent shape, not even dirty and no corrosion anywhere. I can provide photos of the board or the internals, but nothing obvious seems to be going on in there.

Full system specifications

Motherboard: ECS P6FX1-A Rev1.2
BIOS: 1.5
Processor: Pentium Pro 200Mhz (256k L2)
Memory: 64MB (16 x 4) EDO SIMMs
Floppy: One very normal Sony 3.5" microfloppy drive
Optical: HP dvd640, also tried a noisier Lite On DVD drive with no differences
HDD: Tried both CF card and 20GB Seagate IDE HDD
Power Supply: Corsair CX500
Cooler: More-than-adequate
VGA: SiS 6326 4MB
Audio: PCI "SoundBlaster"
Other expansion cards: USB; Netgear RJ45 10/100 Ethernet
Peripherals: IBM Model M and IntelliMouse Explorer, both PS/2

Things I have tried so far
  • Reading the manual
  • Installing a new Dallas RTC module with battery holder (DS1687-5 compatible one from glitchworks. The datasheet showed it as pin-compatible the DS12B887, which was installed before, so please tell me if I've fudged anything here!)
  • Flashing all the BIOS versions available (1.4, 1.4d, 1.5) with no change in results, aside from 1.4 booting much slower and generally being worse at doing so
  • Swapping out the floppy drive, HDD, and DVD-ROM drive
  • Using different installation media, in case of astronomically bad discs
  • Resetting CMOS settings to BIOS defaults and to setup defaults
  • Clearing CMOS data entirely, via jumpers, software, and Uniflash 1.40
  • Removing all expansion cards except for the SiS 6326 video card
  • Swapping the SiS video card for an older Trident VGA card
  • Setting the jumpers to run the CPU at 166Mhz instead of 200Mhz
  • Disabling the L2 cache
  • Swapping the memory, installing only one bank of SIMMs, changing memory from 60ns to 70ns, almost everything RAM-related
  • Loading the Intel PIIX Bus Master driver from an F6 floppy, while loading the NT4.0 installer
  • Re-seating all possible connectors
  • Booting an installer with the HDD, optical, or floppy drive completely disconnected (depending on which media it's run from to begin with)
  • Precopying the Windows 2000 installation files onto the hard disk from in Windows 98SE, and attempting to boot the installer from there

Other observations:

  • Attempting to boot a different machine's hard disk with Windows 2000 already installed also freezes, with the loading bar full. Guessing during ntoskrnl initialization, it panics or something.
  • When the NT4.0 installation freezes, the top half of the screen is blue and the bottom half is black. There is no text, no seams, nothing but these two solid colors. The floppy I/O LED stays lit.
  • TinyCore Linux i386 works perfectly fine. The graphical desktop, common shell commands, and drivers for the NIC all seem to work without issue.
  • I am grateful for any help or second opinions. I will provide pictures of the hardware/screen, elaborate on specs of parts as needed, etc., as needed.

Attachments

Reply 1 of 3, by vetz

User metadata
Rank l33t
Rank
l33t

I had similar weird issues in NT 4.0 on my slot1 dual build and Win95 working beautifully. Turned out it was two issues:

1. 3 sets of memory I had gave errors in Memtest86 on this specific system. So when I tried to eliminate the RAM as a variable when I swapped SIMMs I put in new faulty RAM. Have you tested the machine with Memtest86?
2. I have hardware installed in all slots and the case had poor supports, this allowed the motherboard to flex a tiny degree which would cause strange behaviors if the system was in vertical position. I noticed this when I had the case in horizontal that everything worked fine. I made some extra supports underneath the motherboard and everything was in the clear again. This might not be the situation for your machine as you dont have every slot full.

3D Accelerated Games List (Proprietary APIs - No 3DFX/Direct3D)
3D Acceleration Comparison Episodes

Reply 2 of 3, by Boctor

User metadata
Rank Newbie
Rank
Newbie

Thank you for the suggestions, I ran memtest86 from a floppy for several hours and made sure it performed multiple full passes without reporting any errors. Everything is mounted in the case to the best of my ability. All the ATX mount points are used, and none of those weird plastic PC/AT isolation columns are involved. At this point, I was feeling nervous that the fault would be something impossible for me to find, with only my basic knowledge of low-level electronics. The same basic knowledge that turned out to be the cause of this mess, in the first place. I could have sworn I'd tried to boot NT before and after swapping the RTC chips, but I must be the one with the bad memory.

memtest86-running.jpg
Filename
memtest86-running.jpg
File size
326.87 KiB
Views
226 views
File license
CC-BY-4.0

The fault was my replacement RTC chip! What I'd thought was compatible with the DS12B887 (it was compatible to a limited degree) was actually the culprit. According to a useful write-up I ran into during desperate searching, Windows NT talks directly to the RTC, unlike Windows 9x and DOS where timekeeping is done with the BIOS as an intermediary. This was where it all finally clicked. The RTC access had to be where it was freezing. I ran back to the desk, reinserted the old Dallas chip with the flat battery, and finally made it past the FAT/video/keyboard drivers, into the NT4.0 installer. The black/blue screen that I'd been stuck on for days is actually the one that flashes right before the actual setup program displays anything.

nt40-success.jpg
Filename
nt40-success.jpg
File size
373.13 KiB
Views
226 views
File license
CC-BY-4.0

Hopefully, no one repeats my error and assumes pin-compatible equals compatible. Bottom line: A DS1687 and DS12887 are not fully interchangeable.

Reply 3 of 3, by evasive

User metadata
Rank Oldbie
Rank
Oldbie

Some of the differences are mentioned in here if not directly, you can see they have a different feature set:
http://pdf.datasheetcatalog.com/datasheet_pdf … 5_to_DS1687.pdf