Any updates?

I would like to share some extensive use of v0.2 adapter since my prev post.
The impressions are for Asrock Z87M-Pro4, ESS1868F + wp32. SC55mkII SF2, Arachno and Chorium SF's are okay for Doom/Duke3d and has unpleasant instruments volume balance in other titles.
I made sata to atx24 "connector" and get rid off one more PSU in the case (three of four required voltages: 3.3,12,5). -12V - direct wire from mobo ATX connector.

DOS support - excellent. Dune2/LOL/DOOM "the system" still requires SFX proper init, like fastdoom or duke3d. Without - no sfx. Also Quake dos will run with channels reversed if not initialized "proper way".
LandsOfLore/Dune2 - smooth, requires first 16M of ext mem free. Mean getxms 16384, smartdrv, freexms - okay.

Latest BIOS for Z87m-Pro4 brings 1920x1200 support for IGP, titles with support of custom resolutions runs great - qdos, q2dos, 100'ish FPS.

The issue I'm fighting now with is - win98 and/or Me (via 98quickinstall github) refused to load drivers for any graphics adapter: I have X850 PCIe, bunch of 7900 (GT/GS/Quadro), 6800GT-ish quadro 3400 - no luck. Device manager told me there's resource conflict where graphics card conflicting with 00-7F ports (!), conflicting with keyboard and something else (nvidia case) and 00-FF ports in ATI case. Spent two days trying different variants, no luck at all. Graphics adapter "from the start" even without drivers shows "?" and resource conflict . Vogons has thread for that with solution I cannot afford - disable ACPI in BIOS (no such option) and disable ACPI in windows (easy).
I have some more motherboards waiting for tests (Z87m-D3H Giga and Z97M-Plus Asus). Can't figure out what purpose for 98 setup here though, as all of titles of interest runs great with patches on w11, including Unreal Gold which by Rivatuner didn't considered as a game, so low resources rqrd.

Next steps I'll focus on - GRUB and "incorporating" sapphisa default ports init to the GRUB. For multiOS boot.

vsharun wrote on 2025-05-12, 09:09:

Next steps I'll focus on - GRUB and "incorporating" sapphisa default ports init to the GRUB. For multiOS boot.

Don't know if there's any boot loader that can touch PCI configuration registers directly.

Personally I use GRUB4DOS to perform such multi-booting, particularly for WinNT family (up to XP/2003).

- First boot to DOS (directly if possible, or use GRUB4DOS).
- Set up the ports with the utilities.
- Optionally init the sound card with UNISOUND (for PnP sound cards).
- Go back to GRUB4DOS and choose to boot to the OS of choice.

Note that PnP cards initialized with UNISOUND will work with non-PnP drivers afterwards. This is especially useful for PnP Sound Blaster 16 when using Windows 3.1's default Sound Blaster drivers (NT 3.51 also).

LSS10999 wrote on 2025-05-12, 13:40:

Don't know if there's any boot loader that can touch PCI configuration registers directly.
Personally I use GRUB4DOS to perform such multi-booting, particularly for WinNT family (up to XP/2003).

boot.S (dense space, coz should fit MBR) and diskboot.S (should be enough) both in real mode stage of GRUB, therefore no restrictions.
Also converting sapphisa.c to the GRUB module quite possible.

>vsharun
For Win98 you need R.Loew's memory patch and PCIe vram patch if your VGA has more that 256MB. I successfully ran nvidia 7600 and 7900 PCIe but I had 256MB versions that didn't need patch. I then replaced it by GTX970 - no way for win98 just generic 2D VBEMP driver ( I could put some older VGA to second PCIe if needed)

RayeR wrote on 2025-05-12, 16:24:

>vsharun
For Win98 you need R.Loew's memory patch and PCIe vram patch if your VGA has more that 256MB. I successfully ran nvidia 7600 and 7900 PCIe but I had 256MB versions that didn't need patch. I then replaced it by GTX970 - no way for win98 just generic 2D VBEMP driver ( I could put some older VGA to second PCIe if needed)

All of my PCIe's graphics for win98 are 256M.
Just put PCI S3 Virge, clean setup and voila - no issues, even AHCI SATA SSD are ok, 3.2Gb RAM available.
Once I put 6800/7900 PCIe adapters there's PCI-to-PCI bridge new devices found and after that - question mark on PCI VGA graphics card with conflicting resources range 00-7F as usual.
PS: after inserting PCIe 6800 system freezes in seconds after start. the situation exactly like in this case:
Re: NVIDIA 6800 PCIe (6-Series) works on Windows XP but not recognized on Windows 98SE ( PCIe to PCI Bridge)

I wonder why there's conflict in 00-7Fh IO range as VGA should not use such low addresses. Normally it uses something like 3B0-3DFh. Isn't it rather IRQ conflict? What PCIe slot did you used for your W98 card? The primary x16 one or second x4? I always used my PCIe cards in x16 primary slot but AFAIK e.g. Ruthan used setup with 2 PCIe VGAs (x4 for W98) and it worked for him. No idea what's going on there. Did you tried some minimal setup with all USB disabled, SATA in IDE compatible mode (no AHCI, no native)? On my P67 board I had some IRQ conflict of too many USB ports so I had to disable some of them in devmgr otherwise I got BSOD but still I have at least 2 USB2.0 ports working...
Ok, it's a bit offtopic here...

RayeR wrote on 2025-05-12, 20:20:

I wonder why there's conflict in 00-7Fh IO range as VGA should not use such low addresses. Normally it uses something like 3B0-3DFh. Isn't it rather IRQ conflict? What PCIe slot did you used for your W98 card? The primary x16 one or second x4? I always used my PCIe cards in x16 primary slot but AFAIK e.g. Ruthan used setup with 2 PCIe VGAs (x4 for W98) and it worked for him. No idea what's going on there. Did you tried some minimal setup with all USB disabled, SATA in IDE compatible mode (no AHCI, no native)? On my P67 board I had some IRQ conflict of too many USB ports so I had to disable some of them in devmgr otherwise I got BSOD but still I have at least 2 USB2.0 ports working...
Ok, it's a bit offtopic here...

Mind that PCI classic S3 Virge are ok despite working thru the bridge, with 3.2Gigs of RAM available fro w98. smooth 800x600 16bit. i5's IGP - also okay, no conflicts, but no driver.
Both PCIe slots conflicting. Ati X850 claims all first ports 0000-00FF BTW. Unsupported ATI some 3580 & 5770 - the same conflict without any drivers.
No IRQ conflicts, no memory space conflicts. I have tried himemx max=256M, the same: windows sees 256M and the same issue.
USB disabled globally - the same. IDE compatible mode - no w98 boot at all, GPF IIRC, only AHCI are bootable.
I think there's something inside Asrock's BIOS which prevents PCIe graphics in 98.
I have also Asrocks B75M-Pro4 to test, Gigabyte B75M-D3H, and (in "LDRQ works"-soldering/TPM cable production - Asus Z97m-plus. Plus two H61's. I'm interested mostly in z-series, because easy overclocking.

I have seen Omores videos for with Intel mobo and there's Asus boards everywhere:
Asus Pro H610-C D4 https://youtu.be/i_7yCmXm2wQ?si=sfO5MDX8KdfplGZw&t=112
ASUS Prime Z790-P WiFi DDR5

wait a moment........
Next, we patch the WINDOWS/SYSTEM/PCI.VXD file with ptchopt.exe.
Can't find this patch anywhere. ChatGPT said this patch is something which look like remedy for me. Part of RLoew's nvidia patch package, does nothing, mean the same issue with 00-7F conflict. Also I have tried four versions of PCI.vxd - vanilla and patched from 98SE original installation and vanilla/patched from 98lite installation - the same effect.

UPD: clean install of clean 98SE without any patches - the same
setup without /p i - clean device manager and refuses to accept nvidia drivers, adding PCI bus leads to the same issue - resource conflict.

UPD2: Gigabyte B85M-D3H - Quadro 3400 (6800GT/Ultra) - easy, 1920x1200 32bit color desktop without any w98 patches - clean 98se install with himemx /max=256M. The same w98 installation back to the Z87M-Pro4 - no way,resource conflict

Yes, it must be some odd issue of your specific MB/BIOS, normally pcie vga is no problem under w98. You can dump and compare pci bridges configuration (pci config space registers) of this MB and similar working MB if you find something...

If you ebable AHCI mode it activate IDE BIOS that provides INT13h services that DOS relies on and win98 can use them in compatible 16b mode. Also if I remember there's some Rloew AHCI driver? I rather use IDE compatible mode that also enable low level utils that access disk via legacy IO ports. Win98 sees it as native dual channel IDE. But only if I boot from HDD on sata2 channel, it doesn't work on sata3 channel (my MB has 2 sata3 + 4 sata2)

RayeR wrote on 2025-05-13, 18:33:

Yes, it must be some odd issue of your specific MB/BIOS, normally pcie vga is no problem under w98. You can dump and compare pci bridges configuration (pci config space registers) of this MB and similar working MB if you find something...

If you ebable AHCI mode it activate IDE BIOS that provides INT13h services that DOS relies on and win98 can use them in compatible 16b mode. Also if I remember there's some Rloew AHCI driver? I rather use IDE compatible mode that also enable low level utils that access disk via legacy IO ports. Win98 sees it as native dual channel IDE. But only if I boot from HDD on sata2 channel, it doesn't work on sata3 channel (my MB has 2 sata3 + 4 sata2)

AHCI Rloew path seems incorporated in 98quickinstall and has 6 channels. And according to Disk Speed v2.0 performance of Intel X25-M 80gb with 92% wear (yes, 8% left, SMART "red") exceeds sata1 level with 160-180MiB/s, not bad for sata2 device 17 years old.
I almost forgot how painful win98 days was. New hardware, some crashes = new Windows (re)install. I did some tens of windows reinstalls just for past three days.
Gigabyte B85M-D3H rev 1.0 of mine with F14 bios has "enough" overclocking headroom, so 4670k @4.5GHz 1 core 1.32V stable and only 20%-ish slower than 12700k @4.9 in SuperPi (104s vs 80s 8M). I have no TPM cable just now, friend of mine doing another one for Z87M-D3H, but according to him this B85M-D3H was flawless with 0.2 dISAppointment.
Some 82800 3dmark99 max, 165k CPU marks on 7900GT stock Leadtek WinFast.

Yes, crippling Win98 installation was easy so I learned to make a backup after every successful step. I have multiboot on the same disk so I used other windows to boot on and just copy Win98 directory away to make backup (no need to backup whole partition), it's just a few seconds on SSD 😀 Nice to hear that AHCI driver works.

RayeR wrote on 2025-05-13, 18:33:

Yes, it must be some odd issue of your specific MB/BIOS, normally pcie vga is no problem under w98. You can dump and compare pci bridges configuration (pci config space registers) of this MB and similar working MB if you find something...

If you ebable AHCI mode it activate IDE BIOS that provides INT13h services that DOS relies on and win98 can use them in compatible 16b mode. Also if I remember there's some Rloew AHCI driver? I rather use IDE compatible mode that also enable low level utils that access disk via legacy IO ports. Win98 sees it as native dual channel IDE. But only if I boot from HDD on sata2 channel, it doesn't work on sata3 channel (my MB has 2 sata3 + 4 sata2)

I'm not sure about the reliability of Rloew's AHCI driver, as I'm having issues with it on my ASRock X99M Killer/3.1 with dISAppointment v0.2.

Simply put, the AHCI driver would cause I/O errors (which begins with an abrupt stop of disk activities) from time to time on my system, that Windows would show a blue screen every time it happens warning about data loss, though I don't really notice anything apparent as the failed I/O may have been successfully repeated afterwards. Inspecting the partition from Linux via fsck doesn't show any apparent error on the file system.

The I/O errors caused by the driver would register on my SSDs' SMART log which would additionally cause Windows to think my disk has developed bad sectors that it keeps insisting on doing a surface scan, that I had no choice but to disable such checks altogether via msconfig, since the partition was rather big that it's going to take hours to finish even for a SSD, and I worry whether ScanDisk can properly handle disks larger than 137GB. It's just the I/O error counter was incremented, not the amount of pending/reallocated sectors.

As such I had to run Windows 98SE without using any disk driver. The disk performance is as fast as using the aforementioned AHCI driver, just I lose the ability to use animated cursors.

LSS10999 wrote on 2025-05-14, 02:37:

As such I had to run Windows 98SE without using any disk driver. The disk performance is as fast as using the aforementioned AHCI driver, just I lose the ability to use animated cursors.

Does using non-AHCI drivers disable animated cursors?! That's bizarre

EDIT : yep - https://www.betaarchive.com/wiki/index.php?ti … _Archive/123334

LSS10999 wrote on 2025-05-14, 02:37:

Simply put, the AHCI driver would cause I/O errors (which begins with an abrupt stop of disk activities) from time to time on my system, that Windows would show a blue screen every time it happens warning about data loss, though I don't really notice anything apparent as the failed I/O may have been successfully repeated afterwards. Inspecting the partition from Linux via fsck doesn't show any apparent error on the file system.

The I/O errors caused by the driver would register on my SSDs' SMART log which would additionally cause Windows to think my disk has developed bad sectors that it keeps insisting on doing a surface scan, that I had no choice but to disable such checks altogether via msconfig, since the partition was rather big that it's going to take hours to finish even for a SSD, and I worry whether ScanDisk can properly handle disks larger than 137GB. It's just the I/O error counter was incremented, not the amount of pending/reallocated sectors.

As such I had to run Windows 98SE without using any disk driver. The disk performance is as fast as using the aforementioned AHCI driver, just I lose the ability to use animated cursors.

This is why for W98 installation I stick with "20-40Gb partitions and drives <137Gb" rule.
Thanks to BIOS, flash drives are visible in W98 as a real mode DOS access devices, mean easy data transfer. No animated cursors though 😀

myne wrote on 2025-05-12, 07:50:

Any updates?

I've been working hard on the project and making significant strides. I just got done helping my wife take a broken TLA714, fixing it, upgrading it, and setting it up for PCI and ISA analysis. Right now I am in the middle of developing both a PCI and ISA DLL addon for TLA 700 series Logic Analyzers and PCI + ISA interposer boards that have logic analyzer probe connectors (Mictor 38 pins) for easy probe connections. I did build a single Mictor38 Pin LA probe connector to breadboard spacing pcbs (on the github in my sig and on pcbway shared projects), but having to rewire them for specific signals is quite annoying and time consuming. Yes, yes, I could just have a few of these Mictor2Breadboards made and connect a few to each bus, but the interposer boards are not that complicated and I might as well just build them. This probably should have been done before I built the prototype IT8888 card but you live and learn.

On the prototype IT8888 based card front, It is largely built but I was having trouble with the configuration of the chip. I needed a faster and better way to probe the PCI and ISA busses. Thus Interposer boards with actual protocol analyzers are required so that I can confirm the configuration and operation of the IT8888. Without a solid way to see both buses via an analyzer, I'm basically dead in the water as it's guess, configure, and check what the IT8888 is doing. Which as you can imagine takes significant time, also it's tedious. I am now at the point where I need to be able to verify the entire bus state at once and not just a few signals. So making PCI and ISA interposer boards and the TLA700 protocol analyzer addons are quite necessary imo. Plus with the interposer and analyzer addons I can confirm the operation of the hardware before throwing it in a test machine. I'd rather not blow up perfectly good motherboards if I can help it, even though I do have an Adex PCIx32. I mean there's probably well over 500 wires that needed to be routed, soldered, ensuring they are where they should be, mistakes are bound to happen.

You might think that none of this is necessary and I should just keep smashing my head against the hardware wall, but I disagree. It's still not even proven yet this will work. If it doesn't, none of this analyzer and interposer work will be a loss since it will be helpful in the future for other attempts like trying with the PIIX4 and if it comes to it, the FPGA route. If it does work, than these interposers and analyzers will be highly useful for testing PCBs after manufacturing and ensuring cards soldered are functioning correctly. (If and when it ends up working, I will like to keep a stock cards fully assembled for anyone who wants them.) The protocol analyzers and interposers will also be useful for people who decide to build their own ISA or PCI cards in the future, which is a good thing to support.

At this point I'm basically fully committed to building the card and trying future stuff if needed. My wife has been extremely encouraging and none of this would have been possible without her footing the bill for everything so far. She bought the TLA714, probes, TLA plugin modules, components, connectors, chips, pcbs, ISA DMA cards, pretty much everything. So really, a huge thanks to her!

That's where I am at. I will keep working as fast and hard as I can.

Wow.
Here's me thinking I was probably helping a fair bit but you're way ahead 🤣

Hehe, it's extraordinary to see someone's wife/GF is so enthusiastic and supportive about electronics/HW, she must be from 'outer space' 😀
BTW Raspberry Pico can be also turned into cheap logic analyser, there was implemented some support in Sigrok SW unofficial fork...
I think that for basic debugging you may start with some kinda ISA POST card - many years ago I built this ugly one but it has one big advantage - it has fully configurable address decoder with DIP switch so it can be used to test any IO address if it's properly passed through bridges cascade 😀

myne wrote on 2025-05-14, 12:42:

Wow.
Here's me thinking I was probably helping a fair bit but you're way ahead 🤣

No, you've helped me quite a bit. I would never have thought to turn the IT8888 by 45 degrees, that is huge for routing and placement. Plus the files and stuff for kicad, you've helped a lot. Everyone here has been instrumental.

RayeR wrote on 2025-05-14, 14:31:

Hehe, it's extraordinary to see someone's wife/GF is so enthusiastic and supportive about electronics/HW, she must be from 'outer space' 😀
BTW Raspberry Pico can be also turned into cheap logic analyser, there was implemented some support in Sigrok SW unofficial fork...
I think that for basic debugging you may start with some kinda ISA POST card - many years ago I built this ugly one but it has one big advantage - it has fully configurable address decoder with DIP switch so it can be used to test any IO address if it's properly passed through bridges cascade 😀

Yeah, she's out of this world for sure. I'm very lucky. That's funny you mention GusmanB's project, I actually put in a pcb way order for 10 of them 3 days ago. They were quoted at $125 for all assembled and shipped. Looks like it will take over 25 days for assembly + 1 week shipping. It does say it's already started assembly so maybe it will be faster. I have doubts they have enough buffer space for full bus signal capture. The 100 MHz would be enough for ISA, but PCI transactions might need specialized triggering at such slow samples. It might be just enough for configuration analysis. I know it's capable of 200MHz sampling, but I'm not sure about the 400 MHz claims, i understand it's likely a combination of sample interleaving and pico2 overvolt/overclocking. The limited buffer size presents the biggest challenge imo. For PCI at 400MHz sampling: 32k samples at full sampling rate is filled in just 80μs (32k/400MHz). A typical PCI transaction might take 3-5 clock cycles at 33MHz (90-150ns). This means it could capture approximately 500-800 PCI cycles in a single buffer. For ISA at 100MHz sampling: 32k samples provides 320μs of capture time. A typical ISA cycle takes 125ns or more. This allows capturing around 2,500+ ISA cycles. These do seem like on the surface capable of doing such analysis, however there is no protocol analysis and would have to be added into their tool, or manually interpreted. In any case they will be useful i'm sure. While we were setting up the TLA714, I was using the DSLogic U3Pro32. The main problem with USB analyzers is that none of them have actual ISA or PCI protocol support, which isn't surprising or expected. I'm not sure how to add that to them, I was always told they can only trigger on simple patterns of signals, but this may be untrue, Idk. My TLA ISA and PCI protocol DLLs can actually understand the PCI/ISA protocols and shows transactions (like "memory read" or "I/O write") With Pico2 or U3Pro32, you interpret everything manually; basically the TLA shows what's happening in plain English. The TLA can find exactly the event you're looking for, while the others require capturing everything and searching manually. Furthermore, the TLA can monitor multiple buses simultaneously with perfect timing correlation. This is ideal as it can find protocol violations and timing issues that would be nearly impossible to spot manually.

While it might seem simple enough to verify the configuration of the chip is correct, the only real fool proof way I can see how to do that is by fully validating the incoming ISA and outgoing PCI. Without better IT8888 documentation, there's quite a bit of guess work on our part. Basically with a full protocol analyzer you can: Measure Transaction Latency, Analyze Retry Behavior, Track Buffer Utilization, Driver Compatibility Problems, Device-Specific Quirks, System Resource Conflicts, DMA Transfers, I/O vs. Memory Mapping, Power Management Interactions, Correlate Cause and Effect, Identify Protocol Violations, Provide Timing Context, this is just some of it, there's a bunch of stuff not mentioned. This can reduce debugging time from days/weeks to mere hours. I think the work I put in here will pay dividends in the future no matter the road taken. The whole multibus simultaneous monitoring with perfect timing correlation is particularly important. When troubleshooting, seeing both the ISA and PCI sides with precise timing alignment is invaluable for understanding cause and effect relationships. I'm not sure it's even possible to do this on hobby grade hardware.

That's an interesting ISA post card and a very cool feature. Have you posted about that somewhere? I'm sure the larger community would be interested in schematics if you're willing to share.

this is what I use when I want to analyze the ISA bus - Analyzing the ISA bus cheaply

It migh be usable for some PCI stuff in a pinch

dartfrog wrote on 2025-05-15, 05:35:

No, you've helped me quite a bit. I would never have thought to turn the IT8888 by 45 degrees, that is huge for routing and placement. Plus the files and stuff for kicad, you've helped a lot. Everyone here has been instrumental.

😀