Velociraptor wrote on 2020-11-21, 10:57:

I thought it would be worth seeing if I could get a SATA PCI card to work in this rather than using the IDE to SATA bridges that people are having problems with.

However I cannot get the PCI card BIOS to work. The boot up process is normal - it goes to VGA BIOS and then motherboard BIOS and then when it's due to go to the PCI SATA card BIOS it freezes with a flashing cursor. Keyboard not responsive. This is with no drives connected to the PCI card.

I'm wondering if anyone has any ideas if there's BIOS settings that can be changed to allow it to work?

IDE to SATA bridges are not all equal and neither are SATA PCI controllers . Which SATA PCI controller are you using ?

Is it possible for you to test this PCI card in another machine ?

The issue could be due to a corrupted BIOS on the PCI card, or simply be an incompatibility with your mainboard . It is trying the card in another PCI slot .

As an example of on incompatibility, consider my Asus P3B-F that has no problems with a Promise PCI SATA controller and a SIL3114 SATA PCI SATA controller, but refuses to work with a Promise Ultra100 .

Hi Darry. Those are all good suggestions and I will get to them in time.

However, I want to ensure before I start taking this system apart and another system to test it that I'm not missing something simple in the BIOS.

Velociraptor wrote on 2020-11-21, 12:26:

Hi Darry. Those are all good suggestions and I will get to them in time.

However, I want to ensure before I start taking this system apart and another system to test it that I'm not missing something simple in the BIOS.

The only (remotely) related options I can think of that you might have in the BIOS are to ignore option ROMs on expansion cards (which would make it impossible to boot from the PCI card, even if it did let the PC boot) and the option to enable/disable NCR SCSI BIOS (which I doubt would have any impact on your issue). Other than that, if some PCI IRQ allocation options in BIOS have been changed, it might be worth trying to reset BIOS to default settings . Speaking of BIOS, are you running with version F3 , F4 or something older ?

Again, knowing what PCI controller card you are actually using would be helpful .

I didn't see any options like that, sadly.

I have reset the BIOS already, sadly too. I'm running the latest BIOS I could find which is F4c. I've not done any looking around to see if there's any modded BIOSes etc.

The card itself has a Silicon Image 3114 IC on it. It wasn't expensive, something like £8. I did know it was a bit of a gamble buying it. I do not think the card itself was made by Silicon Image.

Steps I'll be taking next are - trying it in a different (much more modern) PC to make sure it's not faulty. Removing everything I can, and going with a S3 Trio instead of the FX 5200 in case that makes a difference. Trying different PCI slots.

Going back to SATA bridges, myself and a few friends have been trying various things with them. We had problems with freezing and hanging etc. But it turned out that stopping using a FX 5500 (2 of the 3 of us were and mine was never working anyway) and updating to the latest BIOS seemed to help that. 2 of the 3 of us have UDMA2 working on them, and the 3rd has the bridge at least working in PIO mode. The SATA HBA was another thought and we thought might be better in the long run, it would also get around over 127GB drive sizes.

My experience with SIL3114 based cards has been good, both functionally and compatibility-wise, in 440BX boards (Asus P3B-F and Biostar M6TBD OEM) .

On a board of such vintage, because of LBA issues, I too prefer using either a PCI SATA controller or a PCI IDE controller that works well with IDE to SATA converters (Promise Ultra133 and JM20330 works well together, but not on my P3B-F).

IDE to SATA adapter compatibility varies widely .

The two best ones are, IMHO, those based on
- Marvell 88SA8052 (rather expensive and harder to find but reported as working on Intel ICH4 and Via 686B IDE controllers)
- Jmicron JM20330 (inexpensive and easy to find, but has issues on Intel ICH4 and Via 686B IDE controllers, works fine on Promise Ultra133 and can work fine on ICH2, depending on driver)

I have never tested either of those in DMA mode on a 440BX's 82371EB (PIIX4E). I have done limited testing with a JM20330 in such a setup, but only in PIO mode under DOS . It did seem to work in that scenario .

See mSATA to IDE Trouble

Well that's quite helpful, thanks for the additional information. I've done a bit of looking and that Asus board you mentioned the P38-F seems to be a fairly late in the game 440BX chipset board. I'm wondering if it's partly just luck with what works on boards and what doesn't but also if the newer the board is, the more likely things are compatible.

I'm also wondering if I should maybe just be happy with what I have. I've got a working drive at present, and it's going at UDMA2 speeds which for DOS and Win98SE is plenty fast.

I'm tempted to get that Asus board to try it. I'm tempted to buy other HBAs to see if they work instead... but where would that end, it's naive to think that I can start on a new board and have everything that works on this board work fine without problems and then also get an HBA working too.

Again thanks, it's given me a bit of food for thought.

Edit : And just in case it helps anyone else. All 3 of us that had stability issues with cheap sata to IDE bridges had those issues fixed with an updated BIOS and/or stopping using FX 5500.

Velociraptor wrote on 2020-11-21, 18:38:

Well that's quite helpful, thanks for the additional information. I've done a bit of looking and that Asus board you mentioned t […]
Show full quote

Well that's quite helpful, thanks for the additional information. I've done a bit of looking and that Asus board you mentioned the P38-F seems to be a fairly late in the game 440BX chipset board. I'm wondering if it's partly just luck with what works on boards and what doesn't but also if the newer the board is, the more likely things are compatible.

I'm also wondering if I should maybe just be happy with what I have. I've got a working drive at present, and it's going at UDMA2 speeds which for DOS and Win98SE is plenty fast.

I'm tempted to get that Asus board to try it. I'm tempted to buy other HBAs to see if they work instead... but where would that end, it's naive to think that I can start on a new board and have everything that works on this board work fine without problems and then also get an HBA working too.

Again thanks, it's given me a bit of food for thought.

Edit : And just in case it helps anyone else. All 3 of us that had stability issues with cheap sata to IDE bridges had those issues fixed with an updated BIOS and/or stopping using FX 5500.

If I were in your place, I think I would keep pursuing the HBA angle . Your SIL3114 may be a dud or have a corrupted BIOS . Considering that these things are cheap, I would try getting a another one, based on a different board layout (just to be sure you're not getting a carbon copy of the one you have). Another option, albeit with only 2 SATA ports, is the SIL3112/SIL3512 .

I can't imagine why there is an issue between the cheap SATA to IDE bridges and the FX 5500, but stranger things have happened . That said, I personally would not even consider using SATA to IDE bridges that are not based on an identifiable chip with some kind of reputation. I know some people have been lucky with those cheap ones, but I have only had weird issues when I tried some in the past and have also heard from others having bad luck as well . SATA to IDE bridges based on JMicron, Marvell or Sunplus (Startech uses SPIF223A chips in their converters) can be had for not that much more than the stuff based on unbranded mystery chips .

The FX 5500 was causing other problems for me, it was conflicting with the ACPI in Windows 98 and even if it was the only card in the system it was still unable to work, even after a reinstall. It appears it's just too new a card for the old board. A friend didn't have the conflicts but things stopped freezing for him when he stopped using it.

Interestingly I've just put the HBA in an Asrock 990FX Extreme4 board, this is an AM3+ board so much more recent. It wouldn't even POST with the card in a PCI slot, but when removed it goes back to normal.

Velociraptor wrote on 2020-11-21, 19:04:

The FX 5500 was causing other problems for me, it was conflicting with the ACPI in Windows 98 and even if it was the only card in the system it was still unable to work, even after a reinstall. It appears it's just too new a card for the old board. A friend didn't have the conflicts but things stopped freezing for him when he stopped using it.

Interestingly I've just put the HBA in an Asrock 990FX Extreme4 board, this is an AM3+ board so much more recent. It wouldn't even POST with the card in a PCI slot, but when removed it goes back to normal.

I am currently using an FX 5900 card in my P3B-F and have previously tested an FX 5500 in it with no issues . That said, my FX 5500 does not like an overclocked AGP port (unavoidable if running a Tualatin 1400 with a 440BX), my FX 5900 has no issues with it .

At this point, I think your SIL3114 card is a dud.

Yes, that seems the most likely explanation. A friend also bought one, he's been busy today so hasn't had a chance to properly test it. He could on his Asus 440BX board (can't remember the exact number but it's almost identical feature set and slots to mine) got the HBA BIOS to show up, but hasn't tested beyond that.

FX 5200 worked without any problems and the S3 Trio too. FX 5500 for some reason wanted all the resources.

I remember that of the two SIL3114 cards I bought, one was not working at all. The working one had to have its BIOS reflashed with version 5.5.00. It worked fine, also I suppose if you set up a sufficiently large area of the SSD aside (like 10%) TRIM is probably less needed. My main concern with this card is the 10K of DOS conventional memory it uses which is why I went back to an IDE<->SATA adapter (this also frees a PCI slot).

Another angle might be PCI revision. Some PCI SATA cards require PCI 2.2, where your BX only offers PCI 2.1. I have a few (unbranded cheap SI3114 ones off eBay) that behave like that. For SATA on older boards, I swear by Promise SATA150-TX2Plus as a bullet-proof (and fairly easy to obtain) solution.

dionb wrote on 2020-12-12, 13:03:

Another angle might be PCI revision. Some PCI SATA cards require PCI 2.2, where your BX only offers PCI 2.1. I have a few (unbranded cheap SI3114 ones off eBay) that behave like that. For SATA on older boards, I swear by Promise SATA150-TX2Plus as a bullet-proof (and fairly easy to obtain) solution.

Would you expect the TX2300 to be as good? As far as I can see it looks like it might be a TX2Plus with no IDE.

Velociraptor wrote on 2020-12-13, 02:22:

[...]

Would you expect the TX2300 to be as good? As far as I can see it looks like it might be a TX2Plus with no IDE.

A SATA300-TX2Plus with no IDE. That means one generation newer and no Win9x support, regardless of what it does at PCI level. Promise's SATA150 series was the last one to support Win98.

Got it - I had thought the TX2 Plus was the important thing but obviously not!

I think maybe it comes down to this....

TX is the range. x is the number of SATA ports. Plus indicates if it has IDE as well. And finally 150 or 300 indicates the generation.

So I want a TX of any kind that says 150. If it's a 4 or 2 it doesn't matter and if it's a plus or not, again it doesn't matter. I think ideally I do want the TX2plus.

Have I got that right?

I know the SATA150-TX2Plus and TX4 work with BX and Win98, other SATA150 era cards may also work, but can't guarantee that.

I've now set up some alerts and hopefully I will get one. They don't seem too expensive but they are only sometimes available.

Thanks for the help!