Possible irq problem?

Try, adjusting respurces, a different card, and a different slot with same card.

Debian 3.0 recognizes the card.

mk2k01 wrote on 2023-05-23, 05:37:

Hello, I've problems to get USB working on my old SuperSockel7 Machine. It's a FIC PA-2013 with K6 II, a Matrox G400 and an CF-t […]
Show full quote

Hello,
I've problems to get USB working on my old SuperSockel7 Machine. It's a FIC PA-2013 with K6 II, a Matrox G400 and an CF-to-IDE adapter on Windows 2000. The CF-card is marked as swapable drive.
If I plugin an optical USB-mouse to the onboard USB ports, the red light only sometimes shines for a splitsecond. But there seems to be no power on the usb-port. On Windows 2000 I can't install the driver for onboard USB. It says the driver doesn't work.

I also have 3 USB 2.0 PCI cards. One with NEC D720100 (or NEC D720101?), two with VIA chips (6212L?). On none of this I get the mouse powered.
I've already disabled almost everything (serial, parallel, onboard USB, IRC to VGA-card) to get more free IRQs. But it doesn't help. Whenever I install the driver of one of the USB-cards and activate it in W2k, the system freezes. I can install the driver in safe mode, deactivate it and then boot into "normal mode". As soon as the USB-card get's activated, the PC freezes.

Hey everbody,
I finally have "solved" the issue. It are actually quite a lot changes, that were necessary to get one card working in a minimal way.
- I had to upgrade the BIOS to the latest version accepting also K6-2+ processors. With Windows ME USB (onboard and the VIA USB-Cards) work at this point. But I want to get the best Windows ever, so that doesn't count.
- I had to install Windows 2000 in ACPI-Mode (now powers down automatically, which is a great feature). This was possible due to bios update, prior the installation crashes.
- the VIA USB-Cards work, if you don't use any other PCI devices and have very slow RAM timings. Using a sound-card in parallel, increases risk for freezes in case of high usb-datatransfer. Copying a 200 kB file via USB 1.1 speed works most times, if you copy several Megabytes the system freezes quite often.

- The NEC USB 2.0 card works stable (although right now, just in USB1.1 mode, but anyway...), but there are quite a few side conditions:
- don't use a Matrox GPU (G400, G450). Combined with the NEC USB card instandly crashes the system during boot.
- don't use a TNT PCI GPU. It crashes the system.
- ATI Rage IIc AGP 8 MB works fine (although muddy display output). Funny: That card was marked defect, as it doesn't boot in my Pentium III.
- the RAM (100 MHz) has to be of SD-Type (no EDO allowed), or maybe you can just use 256 MB of RAM or just two of three sockets. Something of this.
- the CAS-Latency has to be set to 3
- RAM-Speeds have to be set to Normal (not Fast nor Turbo)
- The +1-Wating Tock for the RAM has to be activated. Also disable any setting that increases RAM timings.
With all this, you can copy files to USB and back AND start CPU-hungry applications (latency.exe from the CPU-Z-guys) during that and the system stays stable. It you decrease RAM timings, the file transfer will also run, but the chances of getting freezes during the start of CPU-hungry applications during the transfer increase a lot. There is no point in having a reliable OS with such a plattform.
Luckily I got an AMD K6-III, so slow RAM isn't that much of a problem. Onboard-USB is still not working. Maybe a driver problem. I just deactivated it.

On the other hand the board runs fine and stable with a K6-2 @ 550 MHz @ 2,5 V with fastest RAM timings and even SD- and EDO-RAM mixture. But maybe only because there was no PCI-Soundcard installed, that could stress the PCI bus.

To be honest, that was with distance my bloodiest pc-building-mess so far and I'm so glad, that I got that socket 370 back in that days (which actually still runs).

mk2k01 wrote on 2023-07-03, 20:53:

Hey everbody, I finally have "solved" the issue. It are actually quite a lot changes, that were necessary to get one card work […]
Show full quote

mk2k01 wrote on 2023-05-23, 05:37:

Hello, I've problems to get USB working on my old SuperSockel7 Machine. It's a FIC PA-2013 with K6 II, a Matrox G400 and an CF-t […]
Show full quote

Hello,
I've problems to get USB working on my old SuperSockel7 Machine. It's a FIC PA-2013 with K6 II, a Matrox G400 and an CF-to-IDE adapter on Windows 2000. The CF-card is marked as swapable drive.
If I plugin an optical USB-mouse to the onboard USB ports, the red light only sometimes shines for a splitsecond. But there seems to be no power on the usb-port. On Windows 2000 I can't install the driver for onboard USB. It says the driver doesn't work.

I also have 3 USB 2.0 PCI cards. One with NEC D720100 (or NEC D720101?), two with VIA chips (6212L?). On none of this I get the mouse powered.
I've already disabled almost everything (serial, parallel, onboard USB, IRC to VGA-card) to get more free IRQs. But it doesn't help. Whenever I install the driver of one of the USB-cards and activate it in W2k, the system freezes. I can install the driver in safe mode, deactivate it and then boot into "normal mode". As soon as the USB-card get's activated, the PC freezes.

Hey everbody,
I finally have "solved" the issue. It are actually quite a lot changes, that were necessary to get one card working in a minimal way.
- I had to upgrade the BIOS to the latest version accepting also K6-2+ processors. With Windows ME USB (onboard and the VIA USB-Cards) work at this point. But I want to get the best Windows ever, so that doesn't count.
- I had to install Windows 2000 in ACPI-Mode (now powers down automatically, which is a great feature). This was possible due to bios update, prior the installation crashes.
- the VIA USB-Cards work, if you don't use any other PCI devices and have very slow RAM timings. Using a sound-card in parallel, increases risk for freezes in case of high usb-datatransfer. Copying a 200 kB file via USB 1.1 speed works most times, if you copy several Megabytes the system freezes quite often.

- The NEC USB 2.0 card works stable (although right now, just in USB1.1 mode, but anyway...), but there are quite a few side conditions:
- don't use a Matrox GPU (G400, G450). Combined with the NEC USB card instandly crashes the system during boot.
- don't use a TNT PCI GPU. It crashes the system.
- ATI Rage IIc AGP 8 MB works fine (although muddy display output). Funny: That card was marked defect, as it doesn't boot in my Pentium III.
- the RAM (100 MHz) has to be of SD-Type (no EDO allowed), or maybe you can just use 256 MB of RAM or just two of three sockets. Something of this.
- the CAS-Latency has to be set to 3
- RAM-Speeds have to be set to Normal (not Fast nor Turbo)
- The +1-Wating Tock for the RAM has to be activated. Also disable any setting that increases RAM timings.
With all this, you can copy files to USB and back AND start CPU-hungry applications (latency.exe from the CPU-Z-guys) during that and the system stays stable. It you decrease RAM timings, the file transfer will also run, but the chances of getting freezes during the start of CPU-hungry applications during the transfer increase a lot. There is no point in having a reliable OS with such a plattform.
Luckily I got an AMD K6-III, so slow RAM isn't that much of a problem. Onboard-USB is still not working. Maybe a driver problem. I just deactivated it.

On the other hand the board runs fine and stable with a K6-2 @ 550 MHz @ 2,5 V with fastest RAM timings and even SD- and EDO-RAM mixture. But maybe only because there was no PCI-Soundcard installed, that could stress the PCI bus.

To be honest, that was with distance my bloodiest pc-building-mess so far and I'm so glad, that I got that socket 370 back in that days (which actually still runs).

Getting these USB 2.0 cards working on SS7 motherboards be quite finicky. I'm glad you got it working on your setup.

I have come to the conclusion that the most convenient choice for a versatile retro experience is the S7 or SS7 in terms of how much you can slow it down. Using a Pentium MMX + setmul for example. You can cover from 1981 to the late 90's. Native USB, PS2 mouse, SS7 has AGP...

I use a USB 2.0 Koutech io-pu520 PCI card.

It has an NEC chipset.

The driver works under many operating systems and it has drivers for lots of different parts.

mk2k01 wrote on 2023-07-03, 20:53:

Hey everbody,
I finally have "solved" the issue. It are actually quite a lot changes, that were necessary to get one card working in a minimal way.
...

Some major update. It seems the USB issue is related to a problem with the MVP3 northbridge and them SD-RAM modules (1x128 MB single sided + 1x128 MB double sided). I testet the modules (without PCI-cards) in MemTest86 some time ago. They were fine. But The files copied from the USB-drive to the hard disk with these two modules were defective and I got strange phenomenons that 7zip can't be executed. Reboot the machine and it works and so on.
As I got plenty of 256 MB double sided modules I switched to them and the problem is(/seems to be) gone. Strangely I can't mix 128 and 256 MB modules and the board runs only stable with 512 MB RAM. But with this amount the times are razor sharp, the USB-Port is reading and writing data and the machine is / (seems to be) stable. The Matrox G450 and the NEC USB-card still can't do together and I only get USB 1.1 speed, but maybe it's a good as it gets with this kind of system.

tauro wrote on 2023-07-03, 21:09:

mk2k01 wrote on 2023-07-03, 20:53:

Hey everbody, I finally have "solved" the issue. It are actually quite a lot changes, that were necessary to get one card work […]
Show full quote

mk2k01 wrote on 2023-05-23, 05:37:

Hello, I've problems to get USB working on my old SuperSockel7 Machine. It's a FIC PA-2013 with K6 II, a Matrox G400 and an CF-t […]
Show full quote

Hello,
I've problems to get USB working on my old SuperSockel7 Machine. It's a FIC PA-2013 with K6 II, a Matrox G400 and an CF-to-IDE adapter on Windows 2000. The CF-card is marked as swapable drive.
If I plugin an optical USB-mouse to the onboard USB ports, the red light only sometimes shines for a splitsecond. But there seems to be no power on the usb-port. On Windows 2000 I can't install the driver for onboard USB. It says the driver doesn't work.

I also have 3 USB 2.0 PCI cards. One with NEC D720100 (or NEC D720101?), two with VIA chips (6212L?). On none of this I get the mouse powered.
I've already disabled almost everything (serial, parallel, onboard USB, IRC to VGA-card) to get more free IRQs. But it doesn't help. Whenever I install the driver of one of the USB-cards and activate it in W2k, the system freezes. I can install the driver in safe mode, deactivate it and then boot into "normal mode". As soon as the USB-card get's activated, the PC freezes.

Hey everbody,
I finally have "solved" the issue. It are actually quite a lot changes, that were necessary to get one card working in a minimal way.
- I had to upgrade the BIOS to the latest version accepting also K6-2+ processors. With Windows ME USB (onboard and the VIA USB-Cards) work at this point. But I want to get the best Windows ever, so that doesn't count.
- I had to install Windows 2000 in ACPI-Mode (now powers down automatically, which is a great feature). This was possible due to bios update, prior the installation crashes.
- the VIA USB-Cards work, if you don't use any other PCI devices and have very slow RAM timings. Using a sound-card in parallel, increases risk for freezes in case of high usb-datatransfer. Copying a 200 kB file via USB 1.1 speed works most times, if you copy several Megabytes the system freezes quite often.

- The NEC USB 2.0 card works stable (although right now, just in USB1.1 mode, but anyway...), but there are quite a few side conditions:
- don't use a Matrox GPU (G400, G450). Combined with the NEC USB card instandly crashes the system during boot.
- don't use a TNT PCI GPU. It crashes the system.
- ATI Rage IIc AGP 8 MB works fine (although muddy display output). Funny: That card was marked defect, as it doesn't boot in my Pentium III.
- the RAM (100 MHz) has to be of SD-Type (no EDO allowed), or maybe you can just use 256 MB of RAM or just two of three sockets. Something of this.
- the CAS-Latency has to be set to 3
- RAM-Speeds have to be set to Normal (not Fast nor Turbo)
- The +1-Wating Tock for the RAM has to be activated. Also disable any setting that increases RAM timings.
With all this, you can copy files to USB and back AND start CPU-hungry applications (latency.exe from the CPU-Z-guys) during that and the system stays stable. It you decrease RAM timings, the file transfer will also run, but the chances of getting freezes during the start of CPU-hungry applications during the transfer increase a lot. There is no point in having a reliable OS with such a plattform.
Luckily I got an AMD K6-III, so slow RAM isn't that much of a problem. Onboard-USB is still not working. Maybe a driver problem. I just deactivated it.

On the other hand the board runs fine and stable with a K6-2 @ 550 MHz @ 2,5 V with fastest RAM timings and even SD- and EDO-RAM mixture. But maybe only because there was no PCI-Soundcard installed, that could stress the PCI bus.

To be honest, that was with distance my bloodiest pc-building-mess so far and I'm so glad, that I got that socket 370 back in that days (which actually still runs).

Getting these USB 2.0 cards working on SS7 motherboards be quite finicky. I'm glad you got it working on your setup.

I have come to the conclusion that the most convenient choice for a versatile retro experience is the S7 or SS7 in terms of how much you can slow it down. Using a Pentium MMX + setmul for example. You can cover from 1981 to the late 90's. Native USB, PS2 mouse, SS7 has AGP...

SS7 is great - using three NEC D720101GJ USB 2.0 cards you can transform your K6-2 into a 386 for a perfect Wing Commander experience.

Skimmed through the thread but could not find a clear answer, so I ask, are there PCI (non-e) USB cards which would work on early-Pentium motherboards? I don't care about the speed so they can be any standard, just to connect USB keyboard and mouse.

you dont care about speed of your pentium after it gets slowed down by USB card?
you can get converters for both keyboard and mouse from USB to serial/AT connectors, or even make one with $5 pico

That's still the kind of device that I'd like to see eventually. A USB 2.0 (or 3.0?) controller that presents devices of certain device classes in a way that older DOS systems support well natively, with compatibility with existing software from the original DOS era.

So for instance:

• a USB sound device that gets presented as a Sound Blaster clone at the register level
• a USB midi device that gets presented as an MPU-401 controller on port 330h
• a USB mass storage or UASP device being presented as an IDE or SCSI controller
• a USB mouse being presented as a serial mouse
• a USB network adapter that gets presented as an NE2000-compatible NIC

Perhaps such a "legacy-native" USB controller could indeed be implemented around a Raspberry Pi Pico or something similar. Something that combines the functionality of projects such as PicoGUS, PicoMEM, ISA Blaster, etc, and takes the idea further.

Hey all.

Question, I just added a nec. Chipset 6 port usb 2.0 card to my g3 mac tower. I will check to see if there is slow down, is anyone else running one that can testify if the slow down is experienced on mac architecture

digger wrote on 2023-07-14, 15:41:

That's still the kind of device that I'd like to see eventually. A USB 2.0 (or 3.0?) controller that presents devices of certain […]
Show full quote

That's still the kind of device that I'd like to see eventually. A USB 2.0 (or 3.0?) controller that presents devices of certain device classes in a way that older DOS systems support well natively, with compatibility with existing software from the original DOS era.

So for instance:

• a USB sound device that gets presented as a Sound Blaster clone at the register level
• a USB midi device that gets presented as an MPU-401 controller on port 330h
• a USB mass storage or UASP device being presented as an IDE or SCSI controller
• a USB mouse being presented as a serial mouse
• a USB network adapter that gets presented as an NE2000-compatible NIC

Perhaps such a "legacy-native" USB controller could indeed be implemented around a Raspberry Pi Pico or something similar. Something that combines the functionality of projects such as PicoGUS, PicoMEM, ISA Blaster, etc, and takes the idea further.

Unless you make it protected mode only where port trapping etc. (i.e how EMM386 can be used for some stuff and how windows drivers work)is possible, such a thing is unrealistic. USB and the controller have no connectivity to the IO ports etc. that these devices use, it is even the whole point of USB to totally break free from such stuff. Only way is with some sort of a middleman like windows driver which can trap accesses to all this (virtualized) hardware and do magic that then is talked over the USB controller to the USB device on other side of it that will let the result be known to rest of the world. So in the end it will essentially be software emulation and it can go a long way, as shown by SBEMU project here that manages to do SB over AC97 and HDAUDIO.

I would avoid usb mouses as much as I can

Tiido wrote on 2023-08-23, 00:10:

digger wrote on 2023-07-14, 15:41:

That's still the kind of device that I'd like to see eventually. A USB 2.0 (or 3.0?) controller that presents devices of certain […]
Show full quote

That's still the kind of device that I'd like to see eventually. A USB 2.0 (or 3.0?) controller that presents devices of certain device classes in a way that older DOS systems support well natively, with compatibility with existing software from the original DOS era.

So for instance:

• a USB sound device that gets presented as a Sound Blaster clone at the register level
• a USB midi device that gets presented as an MPU-401 controller on port 330h
• a USB mass storage or UASP device being presented as an IDE or SCSI controller
• a USB mouse being presented as a serial mouse
• a USB network adapter that gets presented as an NE2000-compatible NIC

Perhaps such a "legacy-native" USB controller could indeed be implemented around a Raspberry Pi Pico or something similar. Something that combines the functionality of projects such as PicoGUS, PicoMEM, ISA Blaster, etc, and takes the idea further.

Unless you make it protected mode only where port trapping etc. (i.e how EMM386 can be used for some stuff and how windows drivers work)is possible, such a thing is unrealistic. USB and the controller have no connectivity to the IO ports etc. that these devices use, it is even the whole point of USB to totally break free from such stuff. Only way is with some sort of a middleman like windows driver which can trap accesses to all this (virtualized) hardware and do magic that then is talked over the USB controller to the USB device on other side of it that will let the result be known to rest of the world. So in the end it will essentially be software emulation and it can go a long way, as shown by SBEMU project here that manages to do SB over AC97 and HDAUDIO.

This could be done totally transparently with no special drivers required and no protected mode required, even on an 8088.

The middleman can be a powerful microcontroller on an ISA card that makes all this legacy hardware appear to the DOS (or Windows or Linux) PC. It can emulate the register and memory accesses so that it's compatible with the legacy hardware, then translate all that for the USB devices. You wouldn't expose direct USB capability to the host PC at all, it wouldn't even know USB exists.

It would be a complicated project, but it's do-able.

And it has to sit in the *ISA* slot, which is what was wanted to be avoided (i.e USB only devices), as I understood. It is very hard to break away from the ISA slot itself if you want compatibility with many of the things relevant to DOS games such sound cards with their need for the ISA DMA controller that simply doesn't work on other connections (excluding some stuff like TDMA on some PCI chipsets that a few back-in-the-day PCI cards used, things that are no longer usable as the methods got removed 🤣).

Tiido wrote on 2023-09-25, 16:45:

And it has to sit in the *ISA* slot, which is what was wanted to be avoided (i.e USB only devices), as I understood. It is very hard to break away from the ISA slot itself if you want compatibility with many of the things relevant to DOS games such sound cards with their need for the ISA DMA controller that simply doesn't work on other connections (excluding some stuff like TDMA on some PCI chipsets that a few back-in-the-day PCI cards used, things that are no longer usable as the methods got removed 🤣).

Ahh, I didn't see that part sorry. Yeah you were correct then.

Though, now I'd love to see one of these ISA cards for my old 8088/286 boxes.

There's one for USB mass storage, USB HID is a bit harder to pull off. You can emulate serial mouse and gameport isn't difficult either but you cannot do anything about keyboard and PS/2 mouse unless you make something that sits in place of the KBC (which would be an interesting thing to try to do).

Yesterday I tried an HP KU-204IN card (the card with only USB 2.0 internal headers) on an ECS P5GX-M with a Cyrix/National GX1. This unfortunately gives slowdown too. Tested in MS-DOS 6.22 with Duke Nukem 3D, DNRATE goes from ~90 to ~20 fps. The slowdown is noticable everywhere.

Tiido wrote on 2023-09-25, 21:49:

There's one for USB mass storage, USB HID is a bit harder to pull off. You can emulate serial mouse and gameport isn't difficult either but you cannot do anything about keyboard and PS/2 mouse unless you make something that sits in place of the KBC (which would be an interesting thing to try to do).

You could go sbemu route and run whole dos session with IO port trapping using JEMM, that way you would be free to inject virtual keyboard/mouse responses.

ibm5155 wrote on 2023-08-29, 09:08:

I would avoid usb mouses as much as I can

All my systems use m590 with logi dongle as standard issue via nec 2.0. otherwise it's recommended sticking to intellimouse optical usb 1.1A for compatibility