Sounds like your BIOS only knows ECHS. Sounds like things go sideways when the protected mode IDE drivers load. That's often a sign that the BIOS and the protected mode drivers are using different drive geometries. When you have two different geometries, it's going to be like two different kaleidoscopes. So while you think you are putting your stuff ay the beginning of the drive, and maybe you even did, but it's probably not ordered the same way once the protected mode driver loads.

https://www.vogonswiki.com/index.php/Storage# … age_Limitations

I'd avoid devices larger than 128GB. That opens up the entire LBA48 can of worms. We don't need to go there.

So start with a drive < 128GB
Then boot from an EZ Drive 9.09 install disk http://vogonsdrivers.com/getfile.php?fileid=1900
Then use Win95b when it asks for an operating system disk. Don't use Win95 RTM because it does not know about LBA or drives larger than 8.4GB.
After the EZ drive install completes, reboot from the hard drive and install win95b and you should be OK. Your real mode will be doing LBA and your protected mode will be doing it too.

Edit: The other possibility is that the Windows driver is trying to force DMA mode and the Sata-->Pata bridges often have trouble using DMA modes. They prefer UDMA modes in my experience.

douglar wrote on 2024-07-10, 19:55:

Sounds like your BIOS only knows ECHS.

LBA was already supported by 1994 mainboards. this board is from 1996 or 1997, i would think that LBA support would have been absolutely standard by then. i'd be surprised to see any socket 7 board without LBA support, really. dug up this KB (?) article and it says april 1994 for AMI BIOSes: https://hwiegman.home.xs4all.nl/msdos/122052.htm

douglar wrote on 2024-07-10, 19:55:

Then use Win95b when it asks for an operating system disk. Don't use Win95 RTM because it does not know about LBA or drives larger than 8.4GB.

it says also that LBA is purely a function of the BIOS, which is how i understood it as well. the hardware was already on the market when win95 RTM launched. i've never heard of a suggestion to use ECHS in the BIOS instead when installing win95.

Thanks a lot for your suggestions 😀 !

After some control on the working system with regular IDE HDD :
I can confirm that the Windows 95 version from the Packard Bell Master CD is the 4.00.950 B, so it corresponds with Win 95B.
I can also confirm that when I open the Device Manager, Disk Drives, GENERIC IDE DISK TYPE01, Settings, the "DMA" checkbox is marked by default.
Do you think it can be the problem ? What if just before the first Windows boot during the restoration process, I select the safe mode to uncheck the DMA box ?
How to avoid Windows driver from forcing DMA mode ?
Unfortunately, if the DMA mode is desactivated, performances may be poor isn't it ?

Maybe it worked with the SSD + IDE to SATA converter when I installed a fresh Windows 95 C (not from the Master CD) because the "DMA" checkbox is not marked by default...

Edit :
The motherboard model is this one https://theretroweb.com/motherboards/s/intel- … -orlando,-tampa
It is from mid-96.

bimole wrote on 2024-07-10, 21:08:

Thanks a lot for your suggestions 😀 ! […]
Show full quote

Thanks a lot for your suggestions 😀 !

After some control on the working system with regular IDE HDD :
I can confirm that the Windows 95 version from the Packard Bell Master CD is the 4.00.950 B, so it corresponds with Win 95B.
I can also confirm that when I open the Device Manager, Disk Drives, GENERIC IDE DISK TYPE01, Settings, the "DMA" checkbox is marked by default.
Do you think it can be the problem ? What if just before the first Windows boot during the restoration process, I select the safe mode to uncheck the DMA box ?
How to avoid Windows driver from forcing DMA mode ?
Unfortunately, if the DMA mode is desactivated, performances may be poor isn't it ?

Maybe it worked with the SSD + IDE to SATA converter when I installed a fresh Windows 95 C (not from the Master CD) because the "DMA" checkbox is not marked by default...

Edit :
The motherboard model is this one https://theretroweb.com/motherboards/s/intel- … -orlando,-tampa
It is from mid-96.

That's a cool board. Lots of integrated stuff.

Boot your system in safe mode and uncheck DMA and reboot. If that fixes it, we have a pretty good handle on the issue.

If the performance isn't what you want, I can think of two options that should let you enable DMA:
1) disable the on board IDE controller and install a PCI IDE controller with BIOS that supports UDMA.
2) switch to a CF card or SD-IDE bridge
Both of those would almost certainly work. You could also try a Marvell based Sata-pata bridge, but that's far from a sure bet.

Hi,

Here are some news...
I can't uncheck the DMA box in safe mode: when I click OK to validate I have an error message, somthing like "registry error, you have to free space or reboot..."
I also tried some tweaks in the BIOS but there are very few options. The IDE translation mode was in "auto detected" mode. I moved it to "Logical Block" after trying without success "Standard CHS" and "Extended CHS" (invalid disk when booting...) I also fiddled with "Multiple sector settings" (disabled, 4 sectors/block, 8 sectors/block, auto detected) and "Fast programmed I/O modes" (auto detected or disabled) --> no success.

However I found that if I delete the hard drive controller in the device manager (in safe mode), it boots correctly and try to find a new driver.
The problematic driver is "Intel 82371SB PCI IDE".
Instead of using the drivers from the MasterCD, I found Intel drivers https://theretroweb.com/expansion-chips/2751#driver

Unfortunately, even if the driver installation process seems to go well, the same freeze is happening when booting Windows.
And if I disable the driver in the device manager, it works.
Now I know where is problem (PCI IDE chip driver and/or PATA-SATA converter) but I can't find a trick. Maybe I will try with an other PATA-SATA converter.

Anyway, the HDD solution remains functionnal and I'm now testing with a 4GB CF...

Stay tuned!

Hi,

Sorry for the double post but I think I have found the culprit... DMA !

I wrote above that the IDE to SATA converter worked with a vanilla Windows 95 OSR2.5 install.
But by default on a fresh Windows 95 install, DMA is not activated. As soon as I mark the DMA checkbox in the Device Manager, it can't boot anymore. And no way to uncheck the box even in safe mode. I'm going to restore the disk (hopefully with disk clones it's very quick!).

Actually, the answer was here on the forum (Windows 98 on SSD - No DMA mode) but I didn't make the connection with it...
There seems to be better converters than others. The one with "no name" chip may be the best candidate. Mine is equipped with the JMicron JM20330 and it turns out to be prone to DMA problems...

I also checked on my PIII i815e with Windows 98SE on which this converter worked flawlessly. DMA is also NOT activated!!

I 'll try to lay hands on this "no name" chip converter...

One more question :
Does the DMA make really the difference regarding read/write performance ? Or does it only make the machine faster since CPU is no longer involved in data transfers (goal of DMA if I have clearly understood...) ?

Thanks 😉

disabled udma in bios?

bimole wrote on 2024-07-14, 15:39:

There seems to be better converters than others. The one with "no name" chip may be the best candidate. Mine is equipped with the JMicron JM20330 and it turns out to be prone to DMA problems...

I also checked on my PIII i815e with Windows 98SE on which this converter worked flawlessly. DMA is also NOT activated!!

I 'll try to lay hands on this "no name" chip converter...

All of the no-name pata-sata bridges that I've seen have been Jmicron clones.

I have not had any luck getting DMA to work with any of the Jmicron or the the Marvell or the SunplusIT chips either.

Matze79 found a fm4-4383 bridge from Canon that has a chip labeled "bu19503kv" that might work, but I've never seen that.

Might be cheaper to buy a PCI Sata controller w/ bios. Those are pretty cheap and plentiful.

ux-3 wrote on 2024-07-14, 16:30:

disabled udma in bios?

Hmmm... Nothing like this in BIOS. As I wrote above, this is a Packard Bell machine with a quite lean BIOS...

douglar wrote on 2024-07-14, 17:32:

All of the no-name pata-sata bridges that I've seen have been Jmicron clones. […]
Show full quote

bimole wrote on 2024-07-14, 15:39:

There seems to be better converters than others. The one with "no name" chip may be the best candidate. Mine is equipped with the JMicron JM20330 and it turns out to be prone to DMA problems...

I also checked on my PIII i815e with Windows 98SE on which this converter worked flawlessly. DMA is also NOT activated!!

I 'll try to lay hands on this "no name" chip converter...

All of the no-name pata-sata bridges that I've seen have been Jmicron clones.

I have not had any luck getting DMA to work with any of the Jmicron or the the Marvell or the SunplusIT chips either.

Matze79 found a fm4-4383 bridge from Canon that has a chip labeled "bu19503kv" that might work, but I've never seen that.

Might be cheaper to buy a PCI Sata controller w/ bios. Those are pretty cheap and plentiful.

Good to know. The mobo has a riser with 2 PCI slots, but on this old machine, would it be possible to boot from a PCI device ? Have to check...

bimole wrote on 2024-07-14, 15:39:

I wrote above that the IDE to SATA converter worked with a vanilla Windows 95 OSR2.5 install.
But by default on a fresh Windows 95 install, DMA is not activated. As soon as I mark the DMA checkbox in the Device Manager, it can't boot anymore. And no way to uncheck the box even in safe mode. I'm going to restore the disk (hopefully with disk clones it's very quick!).

Hello,

I have messed with these JMicron SATA-IDE converters too. I had the exact same problem with the DMA checkbox on Windows as well as messing with DMA on Linux. I was doing it on a 430HX, quite similar to your 430VX.

There is praise elsewhere for Marvell-based converters (e.g. Startech) but they don't work quite right for me either. When I connect one to the primary interface of an early PCI Pentium board onboard IDE, it interferes with the secondary channel, even though they should be independent. I'm not sure if removing a pin or cutting a trace somewhere might help (I vaguely recall such a suggestion). I just mention this as it's often suggested to use a Marvell one instead, so you know it won't necessarily help.

I have a 430TX (which adds UltraDMA) coming that I got at a low price on eBay, and I intend to play with these converters on it and see how well it works.

At least Win95 defaults to DMA off. When I boot certain Linux, NetBSD, etc. on this board, I have to mess with kernel command line or recompile to block IDE DMA.

Did you find this thread:
DMA Mode check does not stick in Windows 98SE?

ux-3 wrote on 2024-07-15, 07:11:

Did you find this thread:
DMA Mode check does not stick in Windows 98SE?

That was an interesting thread, and while the BIOS issue is certainly a potential problem for many Pentium motherboards, I'm not sure the BIOS issue is the issue here.

1) OP was able use DMA modes in Windows with his original spinning hard drive
2) OP is using Win95b, not Win98se.

Seems like his problem is that he cannot uncheck DMA and he's using a SATA - PATA bridge that doesn't use DMA and his motherboard doesn't do UDMA.

ux-3 wrote on 2024-07-15, 07:11:

Did you find this thread:
DMA Mode check does not stick in Windows 98SE?

Very interresting, i'll give a try.
There is also this trick in this video from Phil's computer lab (at the end) :
https://www.youtube.com/watch?v=gyY6WdaXnxU

I've a PCChips M520 (VX chipset) with a Spectra 400 upgrade running a K6-3+ @ 450 MHz. This is one of the few motherboards I have that works properly with a Sil 3114 PCI SATA controller, and that made it very easy to use a SATA hard drive with this board.

bimole wrote on 2024-07-15, 19:33:

There is also this trick in this video from Phil's computer lab (at the end) :

not sure what you mean?

He shows how to force PIO when you are stuck with the DMA checkbox marked and frozen Win boot.
Then you can boot properly to uncheck the DMA in the device manager.
Don't know if it works on all PCs, I havn't tried yet...

Repo Man11 wrote on 2024-07-15, 19:39:

I've a PCChips M520 (VX chipset) with a Spectra 400 upgrade running a K6-3+ @ 450 MHz. This is one of the few motherboards I have that works properly with a Sil 3114 PCI SATA controller, and that made it very easy to use a SATA hard drive with this board.

This is a very good news since my mobo is also built around the 430VX!
Sil3114-based controllers seem to be affordable and much more commun than the Promise SATA150 PCI controllers

bimole wrote on 2024-07-15, 21:05:

He shows how to force PIO when you are stuck with the DMA checkbox marked and frozen Win boot. Then you can boot properly to unc […]
Show full quote

He shows how to force PIO when you are stuck with the DMA checkbox marked and frozen Win boot.
Then you can boot properly to uncheck the DMA in the device manager.
Don't know if it works on all PCs, I havn't tried yet...

Repo Man11 wrote on 2024-07-15, 19:39:

I've a PCChips M520 (VX chipset) with a Spectra 400 upgrade running a K6-3+ @ 450 MHz. This is one of the few motherboards I have that works properly with a Sil 3114 PCI SATA controller, and that made it very easy to use a SATA hard drive with this board.

This is a very good news since my mobo is also built around the 430VX!
Sil3114-based controllers seem to be affordable and much more commun than the Promise SATA150 PCI controllers

No promises though - they are known to be picky about what motherboards they work with, and I've had very poor luck using them with any Via chipset boards. But they have worked with no issues on my Asus TXP4 and the M520.