Here is the Text and PCX files for the DX2-66V16BGC and the DX4-100NV8T Also attached is the picture of the CHKCPU test for the DX2-66V16BGC.

And one last thing. I had an Intel 486DX4-100 in another machine. I figured I would test it. I am seeing weird things. Maybe the heatsink and fan isn't powerful enough. Text is not being displayed properly, either with pixel crawl or the entirely wrong letter displaying, as seen below. Also, when running the Quake Benchmark I get, for lack of a better term pixel sparkling while the game runs. I'm not sure how to capture this effect. Is this a cache issue? The chip isn't described as having write through or write back, so could that be causing this issue? Any suggestions would be very much appreciated.

nathanieltolbert wrote on 2021-11-01, 23:49:

Here is the Text and PCX files for the DX2-66V16BGC and the DX4-100NV8T Also attached is the picture of the CHKCPU test for the DX2-66V16BGC.

Thank you.

It seems your DX2-66 runs with clock tripled, so according to its data sheet it is basically jumpered wrong.
Try to find where it has the 2x/3x jumper. Compare it with the possibilities in your mainboard manual.

However, since I also think that it is binned, why not continuing at this frequency?
I guess it will work, at least as far as thermal limits are concerned.
But when you try to run it with higher FSB you may want to run it with 2x multiplicator.
As timing gets tight with higher clock frequencies please remember, 40 MHz FSB is tricky but 50 MHz is a challenge. Some brave or crazy guys like Feipoa try to run special Cyrix processors with 60 MHz FSB or even 66 MHz. But this is science. Beyond these frequency we start with liquid nitrogen.

One thing you need to know is, that the processor changes CPUID output when you change the multi.
As far as I know all AMD Enhanced 486 processors are concerned. I refer at least to all AMD 486 with 16 KB cache. For the AMD 486 DX4 SV8B I have to check. The AMD 486 DX4 NV8T should not support the CPUID instruction.
Perhaps this is the reason why your DX2 is detected as Enhanced DX4.

Conclusion: It is nothing to worry about.

nathanieltolbert wrote on 2021-11-01, 23:56:

And one last thing. I had an Intel 486DX4-100 in another machine. I figured I would test it. I am seeing weird things. Maybe the heatsink and fan isn't powerful enough. Text is not being displayed properly, either with pixel crawl or the entirely wrong letter displaying, as seen below. Also, when running the Quake Benchmark I get, for lack of a better term pixel sparkling while the game runs. I'm not sure how to capture this effect. Is this a cache issue? The chip isn't described as having write through or write back, so could that be causing this issue? Any suggestions would be very much appreciated.

I don't think it has anything to do with the processor.
Your description lets me think about a graphics card which is just not good enough.
We'd like to know more about model and RAMDAC.

Disruptor wrote on 2021-11-02, 05:20:

I don't think it has anything to do with the processor.
Your description lets me think about a graphics card which is just not good enough.
We'd like to know more about model and RAMDAC.

I am using an ARK1000PVL with 1MB of RAM. It's currently set to 0 Wait states. The strange thing is all of the weirdness with the Intel DX4-100 go away with the AMD DX4-100. The ARK1000PVL is the 2theMax version. I have another 1MB of 50ns DRAM coming, but I didn't check where the NOS was coming from, and it's coming from Israel, so it may take some time to get here. Is there an easy way to check the RAMDAC on the VLB card? I can always test with the built in CL-GD5428 and see if it persists.

Disruptor wrote on 2021-11-02, 05:16:

Thank you. […]
Show full quote

Thank you.

It seems your DX2-66 runs with clock tripled, so according to its data sheet it is basically jumpered wrong.
Try to find where it has the 2x/3x jumper. Compare it with the possibilities in your mainboard manual.

However, since I also think that it is binned, why not continuing at this frequency?
I guess it will work, at least as far as thermal limits are concerned.
But when you try to run it with higher FSB you may want to run it with 2x multiplicator.
As timing gets tight with higher clock frequencies please remember, 40 MHz FSB is tricky but 50 MHz is a challenge. Some brave or crazy guys like Feipoa try to run special Cyrix processors with 60 MHz FSB or even 66 MHz. But this is science. Beyond these frequency we start with liquid nitrogen.

One thing you need to know is, that the processor changes CPUID output when you change the multi.
As far as I know all AMD Enhanced 486 processors are concerned. I refer at least to all AMD 486 with 16 KB cache. For the AMD 486 DX4 SV8B I have to check. The AMD 486 DX4 NV8T should not support the CPUID instruction.
Perhaps this is the reason why your DX2 is detected as Enhanced DX4.

Conclusion: It is nothing to worry about.

That's the problem. I can't seem to find the 2x/3x jumper. It's different than what the jumper manuals have listed. For example all of the jumper manuals indicate a JP35. I cannot find that anywhere on the board. I may just be missing it, or it could be that it is labeled something else. Is there a trace or something I can trace to back identify that jumper? If I could get the other board I have working, maybe someone who is more well versed and intelligent than I could trace out all of the jumpers and we can create a jumper manual for this board specifically. For the most part, it seems that the jumpers are the same, as the ones linked to above in the thread, but there are just a few differences. In one they say to set JP40 and JP4, but on this board those both pertain to the Cache chips. And I can't find JP35. And the JP20 block seems to be right up next to the JP19 block with no space, and the only way I came to that conclusion is because there is the white block around one of the pins in the middle of the large block that would indicate a first pin on a jumper block. And I'm still not certain that I am write in that thought process. And I'm super nervous that I'm going to blow up this board, because I fear I did that to the other board, which never worked from the very get start. I can see traces under where the battery was that are damaged, and a via that looks completely corroded, but I don't know if they are actually bad or not, and my multimeter doesn't make a sound to test for continuity. I never got a better one because all sorts of other things have come up.

-Edit- Corrected spelling errors

nathanieltolbert wrote on 2021-11-02, 15:50:

... The strange thing is all of the weirdness with the Intel DX4-100 go away with the AMD DX4-100. ...

May you specify the model of your Intel DX4 too please? It's the S-Spec number and the &E or &EW thing.

Certainly the Intel 486DX4-100 has the following information on the front of the chip.
intel
IntelDX4
iCOMP index=435

A80486DX4-100
C5160801
&E 3Volt SK050
Intel '89'93

The Back of the chip has this printed in black ink

A5024944CB
A4
HJ 15

Is there anything else I can provide to help?

Also I think I found the RAMDAC for the video card using UVCONFIG I don't know if it's 100% correct, but when I run UVCONFIG this is the information presented.

Video Card: ARK Logic SuperVGA (ARK1000PV) with 1Mb, VBE 1.2
RAM DAC: AT&T ATT 20c491 24 bit DAC
. Configuring VBE 2.0 32 bit protected mode extensions
. Configuring linear framebuffer located at 64 Mb

Does that help?

Disruptor wrote on 2021-11-02, 05:20:

nathanieltolbert wrote on 2021-11-01, 23:56:

... The chip isn't described as having write through or write back, so could that be causing this issue? ...

...
We'd like to know more about model and RAMDAC.

nathanieltolbert wrote on 2021-11-03, 04:45:

Certainly the Intel 486DX4-100 has the following information on the front of the chip. IntelDX4 A80486DX4-100 &E 3Volt SK050 […]
Show full quote

Certainly the Intel 486DX4-100 has the following information on the front of the chip.
IntelDX4
A80486DX4-100
&E 3Volt SK050

Is there anything else I can provide to help?

That's description enough.
According to cpu-world.com it has 16 KB unified L1 cache.
&E is Write Through
&EW would be Write Back
A write through chip should have less problems.

I'm sorry, I never have had an Intel DX4 and I'm neither a RAMDAC expert.

Disruptor wrote on 2021-11-03, 09:12:

That's description enough. According to cpu-world.com it has 16 KB unified L1 cache. &E is Write Through &EW would be Write Back […]
Show full quote

That's description enough.
According to cpu-world.com it has 16 KB unified L1 cache.
&E is Write Through
&EW would be Write Back
A write through chip should have less problems.

I'm sorry, I never have had an Intel DX4 and I'm neither a RAMDAC expert.

That is okay. Thank you for the information. On the RAMDAC I'm no expert either. However, reading up on my ARK1000PV card, it appears that the RAMDAC is labeled as another ARK chip rather than the reported AT&T chip. I will try and find a datasheet on the AT&T listed chip and see what I can find. However, I probably won't understand what I read. I really wish that I had been given access to this information when it was new back in the mid 90's when I was a teenager. Maybe I would understand it better now.

Okay I looked up the Datasheet on that specific RAMDAC. It's listed as being capable of 100/80/66/55MHz operation speeds. I am guessing the speed is set by the BIOS or maybe a crystal on the card? But knowing my luck the DAC is probably running at 55MHz rather than 100MHz. It lists support of 24- (18-), 16-, 15-bit true color along with 8-bit pseudocolor apps. It also lists support for a MPU Interface? I'm guessing that's different than what I think of when I see MPU. Is there a benchmark or something that I can run that will give me the speed on the RAMDAC, does anyone know?

Please also check the datasheed of your RAMDAC whether the effective speed has to be divided when switching from 8 bit modes to 16 bit or 24 bit.
The speed is basically set by video mode & refresh rate. Some common VESA modes are stored in the video BIOS. But wjth custom resolutions or other refresh rates you have to calculate. However, you have to add overscan "pixels".

Hmm. The only thing I am finding regarding effective speed is a Maximum Resolution by Speed Grade and Color Modes table. It indicates that the chip runs at 50Mhz for 800x600x72Hz for 8-bit pseudo-color, 65MHzat 1024x768x60Hz, 80MHz at 1024x768x75Hz and 100MHz for 1024x1024x60Hz.

Is that referencing the dividing of effective speed you are referring to? I changed the card from 0 wait states to 1 wait states and the issue with the pixel crawl and the text corruption seems to have ended. I don't know if that's because of changing the wait states or not though, but I will take it.

Interesting. In one old game I am getting that weird pixel crawling and displacement. It's only one game tested so far, and it's in SVGA mode I think? The game in question is Fable from SciTech. I will see if I can get a screen capture of it at some point. I don't really know how to describe it other than when there is say a face animation on screen for a character talking pixels are displaced across the screen that should be part of the animation? I used to get the same thing on my old Packard Bell that had a HT216/32 built in graphics card.

Hi Nathan,
Did you ever find a jumper manual for this motherboard (H21D, ver. 2.1)?
I got the same NOS system and trying to get a 486DX 33MHz and perhaps later DX2 66MHz going on it.

Thanks!

Hello MagicVoodooable. Sadly no. I have not found a 100% accurate jumper manual for this motherboard. All of the manuals that I did find on UltimateRetro are pretty good if you just want to set the bus to 33MHz and if you are using a DX/DX2 chip. UltimateRetro does not link those manuals any more to that board because of their inaccuracy I think, but if you look up the manuals and check the jumpers, the available jumpers should work for a DX/DX2 chip. Interestingly the board itself is pretty fast and using the integrated CL-GD5428 or whatnot is stable and quick. Overall it's a really nice board I think. The integrated IDE isn't the fastest, but it's fast enough for old DOS stuff.

Thanks Nathan! Unfortunately I haven't had any luck getting this board to boot. I cleaned it from the RTC battery damage and repaired the traces that got disconnected.

I tried various jumper settings (I came across this thread here -- which was super helpful):
How to identify this 486 motherboard techmedia UMC um8498F

I wonder if my BIOS chip has lost its memory and I need to reprogram it. Did you have such issues, or any need to reflash the BIOS? If so, where did you get a BIOS image?

Ultimate Retro has a copy of the BIOS image, actually pulled from my board. So if you look up the Markvision Flex board there should be an attachment of the BIOS file there. Here's a link to the UltimateRetro page regarding the bios - https://www.ultimateretro.net/en/motherboards/8470#downloads

I finally got my XGecu T48 programmer and tried to re-flash the BIOS on my board. However, I noticed when I dump the content of my BIOS chip, the file is 128KB. The BIOS file from the UltimateRetro Site for this board is only 64KB and the programmer doesn't let flash it.

Do you have any ideas what I can do?