hacked PCEM to have edit boxes to control my filter's settings for further interactive tweaking and forgot how annoying the Windows resource handling stuff was.

(also turns out my cpu filter isn't getting a lot of precision and there's heavy streaking........ still have a long way to go 😒 it's like it's going in two directions at the same time despite my functions only writing a pixel to the right, i might not fully understand voodoo emulation anyway since I assumed P8A8 is RGBA4444)

I was out shopping at my local mall when I spied an old beige case sitting in one of the hardware shops. 5 bucks later and its mine. Been spending the evening cleaning it up and thinking about what hardware would be nice in it.

Today Im tinkering with my new AIR P6NDI dual Pentium Pro motherboard.

So far I have tested it as it came without changing anything in the BIOS setup, next I will try to see how much I will be able to tweak the performance.

It came with 320MB EDO memory and two 200 MHz 256 KB CPUs, I will remove some memory and one CPU to see if I can get the speed up to 233 MHz as a first step. I will also see if any of the memory modules can handle running at 50ns (I think they are 70ns modules). If these modules wont play nice I know I have two 128MB 60ns modules in a box somewhere, in worst case I could borrow a 50ns module from my PR440FX system if it is compatible. Its bit unclear what voltage the 168pin memory run at with this board, the manual only states that 5V and 3.3V modules are keyed differently so I will have to see if 3.3V modules fit the sockets or if all memory run at 5V on this board. Edit I found a jumper for switching between 3.3V and 5V memory, both fits in the dimm sockets. /Edit

When I have tweaked the system as much as possible I will try a 512KB CPU to see if that improves the performance. I have a feeling the 512KB CPUs wont overclock much so perhaps I will use them in my PR440FX system running at 208 MHz and use the best of my 4 256KB CPUs with this board as I intend to use it as a pure DOS system.

The top realistic overclock with this motherboard seems to be 4*60 = 240 MHz but 3.5*66 MHz should be slightly faster. Its a shame this board dosnt support 70 MHz FSB like the PR440FX as I think 3.5*70 = 245 MHz should be possible.

I think I will start a thread sharing what I find out concerning this motherboard as there is hardly any information at all available online, it might be useful for other people with this board.

Just finished my overkill glide rig. Well, it's allmost finished - still looking for a Voodoo 5 5500.

- Athlon XP 1900+ 1,6GHz (will replace with a 2400+ soon)
- Gigabyte GA-7VRXP KT333 + Universal AGP
- Creative Sound Blaster 128 On-Board^
- 1GB DDR400 Kingston running at 400MHz (yes, this KT333 allows DDR overclocking)
- 3DFX Voodoo 3 3500 TV EU AGP
- Antek NSK4000 Black
- Enermax NAXN 450W PSU with strong 5V rail
- 80GB WD Caviar Black
- Win98SE + WinXP SP3 dual boot

Scores 2452 pts in 3DMark01 and 7862 in 3DMark99.

That's a lovely glide machine! Played some Unreal yet?

philscomputerlab wrote:

That's a lovely glide machine! Played some Unreal yet?

Thanks 😀 I'm installing UT Gold and Unreal on it as we speak 😀

I removed the 128MB + 64MB "168p 60ns Buffered ECC EDO" and left the 4*32MB normal? 72 pin EDO modules in their sockets.
How these different kinds of memory could work together is a mystery but the 32MB 72pin modules seem to run fine with the 50ns setting so I will use those.

Speedsys memory throughput gained 10MB/s but I did not see any improvement in the Quake timedemo or PCPBench.

RWARGH the onboard audio isn't working! Don't know how I missed that... Such a shame - it's a creative chip too... (Well... crative branded ESS chip...)

Front panel header jumpers are in their correct locations, and the correct drivers have been installed from Gigabyte's support page... Guess I'll plug in my Audigy 2 ZS and hope for 98 drivers... 🙁

More likely to be a Creative badged Ensoniq, like ES1370, ES1371 or ES1373... They were from the Ensoniq AudioPCI and Sound Blaster PCI / 128... They're horrible anyway to be honest, so you're not missing anything. Every system I even found with one in ran with stability problems, there's a Gateway one with an amp onboard which is just as bad as the rest but when it has been running for a while smells like burning plastic and causes the screen to black out. Never looked into it as I just threw a generic PCI card at the machine, can't remember the chipset, might have been a Crystal one.

Yeah, I disabled it from bios and stuck my Audigy 2 ZS in there. Downloading the driver CD now.

I did think something was wrong when I first installed it in win98 - using the installer caused BSODs. Manually installing it seemed to do the trick but there was no audio. Maybe the motherboard has an extra header I'm missing? Well, nevermind. The ZS was recognized by XP and is running fine. Now for the tricky part - getting it to work in 98...

EDIT: HOLY F*** the Audigy 2 sound f**king GREAT! I thought most newer Creative boards sucked! I mean my X-Fi Xtreme Music sounds allmost identical to the Realtek ALC668 in my G751JY...

kanecvr wrote:

Yeah, I disabled it from bios and stuck my Audigy 2 ZS in there. Downloading the driver CD now.

I did think something was wrong when I first installed it in win98 - using the installer caused BSODs. Manually installing it seemed to do the trick but there was no audio. Maybe the motherboard has an extra header I'm missing? Well, nevermind. The ZS was recognized by XP and is running fine. Now for the tricky part - getting it to work in 98...

EDIT: HOLY F*** the Audigy 2 sound f**king GREAT!

Have you done any double blind testing against a 8 bit Soundblaster 😉

Im now testing jumper settings with the board

The manual only gives you the jumper settings for PPro 150, 180 and 200 so I only know the settings for 60 and 66 MHz fsb and 2.5 and 3.0 multipliers.

The FSB jumper block has 4 jumpers and the multplier block also has 4 jumpers... trial and error ftw..

I found the 4x multi and the system seems stable at 240 MHz, 266 Mhz posts but produces some divide error before booting DOS.

I have also found undocumented FSB settings of 50 and 55 MHz

The Pentium PRO's cache is very speed sensitive... overclocking the CPU might be dangerous (you could end up damaging the cache)

kanecvr wrote:

The Pentium PRO's cache is very speed sensitive... overclocking the CPU might be dangerous (you could end up damaging the cache)

I have never heard of anyone killing a Pentium Pro CPU if not letting it run very hot? My cooling is rather good 😀

I have now mapped out the FSB jumpers on the AIR P6NDI dual Socket 8 motherboard

JS1 to JS4

1-2, 1-2, 1-2, 1-2 = 60 MHz (This is the corrcet setting for 60 Mhz FSB if you go by the manual))

2-3, 1-2, 1-2, 1-2 = 60 MHz

1-2, 2-3, 1-2, 1-2 = No post

1-2, 1-2, 2-3, 1-2 = 55 MHz

1-2, 1-2, 1-2, 2-3 = 50 MHz

2-3, 2-3, 1-2, 1-2 = 60 MHz

1-2, 2-3, 2-3, 1-2 = No post

1-2, 1-2, 2-3, 2-3 = 66 MHz

2-3, 1-2, 1-2, 2-3 = 50 MHz

1-2, 2-3, 1-2, 2-3 = No post

2-3, 1-2, 2-3, 1-2 = 55 Mhz

1-2, 2-3, 2-3, 2-3 = No post

2-3, 1-2, 2-3, 2-3 = 66 MHz

2-3, 2-3, 1-2, 2-3 = 50 MHz

2-3, 2-3, 2-3, 1-2 = 55 MHz

2-3, 2-3, 2-3, 2-3 = 66 MHz (This is the correct setting for 66 Mhz FSB if you go by the manual)

Its clear that JS3 and JS4 is enough to set all 4 FSB options. The combination of JS1 1-2 and JS2 2-3 always fails to post regardless of the other FSB jumper settings.
Im wondering what JS1 and JS2 are used for...

I did also map the multiplier jumpers.

JSS1 to JSS4

Multiplier

JSS1 to JSS4

1-2, 1-2, 1-2, 1-2 = 2.0x

2-3, 1-2, 1-2, 1-2 = 2.5x

1-2, 2-3, 1-2, 1-2 = 3.0x

1-2, 1-2, 2-3, 1-2 = 4.0x

1-2, 1-2, 1-2, 2-3 = 2.0x

2-3, 2-3, 1-2, 1-2 = 3.5x

1-2, 2-3, 2-3, 1-2 = 2.0x

1-2, 1-2, 2-3, 2-3 = 4.0x

2-3, 1-2, 1-2, 2-3 = 2.5x

1-2, 2-3, 1-2, 2-3 = 3.0x

2-3, 1-2, 2-3, 1-2 = 2.0x

1-2, 2-3, 2-3, 2-3 = 2.0x

2-3, 1-2, 2-3, 2-3 = 2.0x

2-3, 2-3, 1-2, 2-3 = 3.5x

2-3, 2-3, 2-3, 1-2 = 2.0x

2-3, 2-3, 2-3, 2-3 = 2.0x

Well the multiplier jumpers are even more confusing than the FSB jumpers.
Its obvious that the CPU runs at 2x when it dosnt support the multiplier selected.
The function of JSS4 is a bit of a mystery.

There is no difference in speedsys mem score with the same multiplier and fsb using different jumper settings, for example the L2 cache stays activated.

Quake demo1 at 233 Mhz with a Matrox Mystique 4MB and a SB AWE 64: 46 FPS. Its not a very impressive score but I guess Quake is one of few games running faster on a vanilla Pentium at the same clock speed. The 512 KB CPU adds 4 FPS to the score and gets 50.1 FPS at 233 Mhz.

Its a bit annoying that a 512KB cache CPU performs better even in DOS as I would like to put those on my PR440FX board running NT4. I wonder if a P-Pro with 1MB cache can handle 233 MHz...

With the arrival of the proper I/O controller card from Australia, I managed to get my 486 rebuild to a bootable state today. 😀 Still waiting on a few more parts to finish it off (Case, sound card and the correct video card), but it's to a point now where I can actually begin doing stuff with it!

Im studying Intels documentation for the Pentium Pro 200 CPU.

I thinking of testing a VID mod sometime in the future. The normal VID for a Pentium Pro 200 is 3.1V to 3.3V but the maximum possible is 3.5V. If you short all VID pins by connecting them to eachother the board should give the maximum VID of 3.5V if I read the paper right. The maximum voltage the CPUs can take according to Intels spec is VID +1.4V = 4.5V - 4.7V! 😁 To get that much voltage some serious mods are needed, not something I will try.

All of my Pentium Pro CPUs seem to run fine at 233MHz but non of them can run at 266MHz, perhaps one of them will with 3.5V. I loop the built in demo function in Quake to test stability, even with the CPU fan set to the minimum speed non of the CPUs seem to get hot even when looping the demo for an hour or more.

My mind and resources are totally focused on my XT computer build. I have an XT-IDE board to assemble and to test on my Juko XT motherboard.
The XT-IDE board will be flashed with XT-IDE BIOS for NECV20, for better speed, and will have an additional 8Kb BIOS ROM extension for 1.44MB floppy support. This will allow me to boot from 1.44MB floppies on the fly.

Will install MS-DOS 5, and will most likely use a compact flash card for ease of use.

carlostex wrote:

My mind and resources are totally focused on my XT computer build. I have an XT-IDE board to assemble and to test on my Juko XT motherboard.
The XT-IDE board will be flashed with XT-IDE BIOS for NECV20, for better speed, and will have an additional 8Kb BIOS ROM extension for 1.44MB floppy support. This will allow me to boot from 1.44MB floppies on the fly.

Will install MS-DOS 5, and will most likely use a compact flash card for ease of use.

May I please refer you to this thread of mine and notify me if you have any problems that relate to the item !? Pleaaaaaaaaaase.. 😀
I've never been able to get mine working on my XT, but it works great on other machines.

Anyone got the XT-CF card working on an XT 8086?

Got to work with a Micronics M4PI, it is quite possibly the worst 486 motherboard I have ever encountered. 63 Topbenches with a 100MHz CPU? I didn't have the patience to test anything else as it was too slow (Time was limited) and with most CPUs it either won't post or does not run in a stable manner, so I kept returning the AMD DX4 to the socket.

Also, for a board with so few jumpers, they sure made the layout suck as much as possible. At least I have something for the "potential videos" pile I guess.

HighTreason wrote:

Got to work with a Micronics M4PI, it is quite possibly the worst 486 motherboard I have ever encountered. 63 Topbenches with a 100MHz CPU? I didn't have the patience to test anything else as it was too slow (Time was limited) and with most CPUs it either won't post or does not run in a stable manner, so I kept returning the AMD DX4 to the socket.

Also, for a board with so few jumpers, they sure made the layout suck as much as possible. At least I have something for the "potential videos" pile I guess.

It could be cache, depending on what's on it. One of my 486 boards would not run with the DX4-100 nor the AMD 133 chip with 20ns cache in it, I had to upgrade it to 15ns cache before it would work. I was having similar issues with it before with slower cache, not posting, instability, ETC (Even when I disabled cache in bios, which had me all confused). Then with the proper cache, everything just worked flawlessly and works great now. And it's not that the cache chips I have are bad, the 20ns ones work great in another motherboard with a DX2-66, just not with the 100/133 chips.

Came with 15ns cache (256K), replaced with known working 15ns cache early on as my mind lead me to the same thought - though half of it is soldered to the board so I can't change that easily with it being SMD. No difference. Nor is it the RAM. It likes reporting 486SX as the CPU type, but it does see a 487 installed alongside it. It also doesn't support the POD. Interestingly the Cx486DX4 reports 52MHz in Chkcpu, which is impossible on a 33MHz BUS (Board has no 40MHz or 50MHz options) as far as I am aware.

I think it just sucks, a good candidate to replace the Batman as my Joker machine... I didn't realize until now what I had done there, and I hate Batman.

I find it rather novel and it does have the Intel 420 chipset which makes it have some historical value I guess. It also features a lovely CMD0640 IDE chip, great for corrupting the hard drive.

I'll keep tampering with different components anyway, you never know, I might find some combination that works well.

Edit: First motherboard I have ever encountered that doesn't POST with an X5 installed. Tried a different CxDX4... It's faster for some reason.

Edit 2: Eugh, horrible. My 486SX tramples it... I think I'll leave it alone, every problem I have can be found in Usenet archives and usually conclude with someone calling Micronics in the 90's only to be told "That doesn't work this way/isn't compatible with this board." I think it will get a video first though because I am stunned at how terrible this thing is, I thought the faulty MSI was bad, I thought the FIC was slow, I thought the Elpina/PCChips had problems and I thought the BioStar was probably the worst 486 board ever produced... I was wrong. This takes the cake. Sucks, every other Micronics I've messed with has been great even when a little stiff on compatibility.