Repo Man11, I tried this BIOS during my testing and it made no difference on detection. See my post where I have linked a newer BIOS version 5.5.0.0 if you want the latest update.

I'm experiencing corruption again on my SiI3112 after booting to an IDE drive running Windows ME and copying files to the SATA drive. Drivers were not installed for the controller to test whether data corruption can be demonstrated in this configuration as it uses the BIOS to access the disk by falling back to real-mode instead of using its protected mode drivers.

The files I copied were InstallShield programs which I ran after the copy and each of them tested fine. I then shut down and tried booting to the SATA drive which also had an installation of Windows ME. The system blue screened with a Windows Protection Error. Subsequent reboots failed to let the operating systems boot and would just hang.

Plugging the IDE drive back in and booting to that yielded an unstable system. Running the InstallShield programs like I had before would cause the operating system to lock up. After a hard reset ScanDisk appeared and the system froze again.

I suspect that there are bugs with the real-mode (BIOS level) disk functions that are triggered by writing to the disk without having the controller drivers installed. This isn't a problem in NT because the drive can't be accessed this way in protected mode without any drivers installed. This is not the case in DOS, Windows 98, and ME, where the drive is accessible by BIOS functions by falling back to real-mode on disk access.

I don't have enough data for protected mode drivers within an NT operating system but using this controller on any DOS or 9x system that might accidentally write to the disk does seem to risk corruption.

Kahenraz wrote on 2021-09-30, 19:57:

Repo Man11, I tried this BIOS during my testing and it made no difference on detection. See my post where I have linked a newer BIOS version 5.5.0.0 if you want the latest update.

Thanks, but it works fine in the system I have it in, and I cannot imagine that there's anymore speed that could be squeezed out of it considering the rest of the hardware, so I'm going to sit tight.

As long as you only access the disk from within Windows with the controller's drivers installed then you'll probably be fine. My testing was not extensive but I think the issue stems from a use case that was not validated during testing and was never advertised as a feature.

I remember that when I bought my a7n8x Deluxe, the Sil SATA controller always corrupted my OS. Even RMA'd my brand new Seagate drive thinking it was the faulty part, but no. After that, PATA only in that board.

I had one of these in my spare parts that I bought for a Socket A board, but it didn't work with it. I thought I'd give it a try with my PCChips 930LMR board. I flashed it to the 5500 non RAID BIOS, and used a spare SSD. I was able to install Win98 to it, but the machine couldn't find the drive on reboot when set to auto, and I had to manually set the first boot device to BBS-0. But no matter what I did, it had a very long delay prior to showing the Win98 splash screen with the card installed. That delay, combined with slightly faster drive read speeds, but significantly lower write speeds when compared to the onboard SIS IDE with a Startech IDE to SATA adapter, meant that I was better off as it was.

The capacitors on my card had gone. It may be the same with yours.

I'm having a similar problem with a Promise controller as well where it works intermittently on reboots or drops out entirely after the system is booted.

I tried the one I just flashed to the 5500 BIOS in my TXP4. Once again, in that board it works perfectly; though it did hang on the first boot up after I swapped the cards, once I hit reset it booted up normally, and went straight into Win98. The newer BIOS didn't change the ATTO benchmark. These cards just seem to have hardware/BIOS issues that restrict their compatibility, so it's a matter of luck of the draw whether they will work in your intended application.

Thanks for testing. I couldn't find any information on what changed in the last 5.5.0.0 BIOS but given the problems these controllers have it's probably best yo have the very latest. The driver page does say that you need 5.5.0.0 if you plan to use version 1.3.11.0.

There are only two non-RAID 32-bit drivers for the SiI3114: version 1.3.10.0 and 1.3.11.0. Version 1.3.10.0 is the only one which has Windows 9x drivers and is certified for NT 4 through Windows Vista. The latest version 1.3.11.0 is NT-only for Windows 2000 through Windows 7.

The latest Win9x driver I've found is the 1.2.0.5, which is what I'm using with my TXP4..

Here is a link to the drivers that were available from Silicon Image:

https://web.archive.org/web/20120503034510/ht … px?pid=28&cat=3

Going far back enough I found your driver 1.2.0.5:

https://web.archive.org/web/20090212115126/ht … px?pid=28&cat=3

This was released back in 2005. The most recent driver version 1.3.11.0 is from 2009.

For what it's worth and since Kahenraz you mentioned a 440BX board, 3112 compatibility with older systems was broken severely with BIOS v.4.2.47. Pentium III and newer is OK, Pentium II is not.

What are the symptoms of this incompatibility?

Kahenraz wrote on 2021-10-02, 19:49:

What are the symptoms of this incompatibility?

Mostly the connected drive is recognized, but it hangs/freezes when accessing it (either attempting to boot from it or simply accessing a file after booting from another media)

I'm pretty sure that I was stable so long as I had the drivers installed even on a pre-Coppermine Celeron.

I've put these away in a bin already so I can't confirm.

I've been testing SSDs in my Pentium 2 with both the IDE to SATA converter and also a Sil3114 card.

I am finding that a lot of the compatibility problems can be traced to specific SSDs rather than the controller.

I am finding that my Intel 320 40gb works just fine, as well as a corsair force 115gb SSD, but I can't get a couple of 60gb sandforce 2 drives to work correctly (one Intel 330, the other Patriot Pyro 60gb)

So definitely try different SSDs if you are not having luck with a configuration. I bet it has something to do with the LBA translations used of older BIOSes.

But are you testing it with or without the drivers installed?

The SSD I used for testing is a 120GB Corsair Force GT. I have a couple of these and have used them for years without issue. Online reviews say that it has a SandForce SF-2281 processor.

Kahenraz wrote on 2021-10-03, 10:56:

But are you testing it with or without the drivers installed?

The SSD I used for testing is a 120GB Corsair Force GT. I have a couple of these and have used them for years without issue. Online reviews say that it has a SandForce SF-2281 processor.

For the Intel 330 60gb, it didn't matter if I was using the IDE to SATA converter or the Sil3114 card (which needs drivers for windows XP install), it would fail to boot after the the first reboot during the Windows XP install.

The Intel 320 40gb worked perfect with the IDE to SATA converter with the Windows XP install, and I suspect there would be no problems if I used the Sil3114 with it either.

Yep, works fine on the Intel 320 SSD - 40gb:

Using the onboard IDE with an IDE to SATA converter:

Using the Sil3114 controller:

SSD choice seems to be critical to getting this to work on old computers. So if you are having an unreasonable amount of trouble with one SSD, definitely try some others before giving up

Kahenraz wrote on 2021-10-02, 00:55:

Here is a link to the drivers that were available from Silicon Image: […]
Show full quote

Here is a link to the drivers that were available from Silicon Image:

https://web.archive.org/web/20120503034510/ht … px?pid=28&cat=3

Going far back enough I found your driver 1.2.0.5:

https://web.archive.org/web/20090212115126/ht … px?pid=28&cat=3

This was released back in 2005. The most recent driver version 1.3.11.0 is from 2009.

OEMs that sell the cards also have their own driver and utility offerings, which can be older or newer than whatever SI has available. That was always the most annoying thing about the SIL controllers, because you had to hunt down the newest drivers and utilities you could find to deal with problems and even the cards shipping now with the ancient controllers still ship with the older drivers and require you to find the latest and greatest.