My XP Pro SP2 is totally smooth on a Pentium II 400 MHz laptop with merely 160MB RAM and a 5GB IDE drive..

Kahenraz wrote on 2021-09-28, 03:48:

I'm installing to an SSD so it's not the disk. 256MB is sufficient as I used this amount with XP back in the day. The only other bottleneck here is the processor.

I wouldn't say that. When I was an early adaptor of SSDs back in the early 2010s,
XP ran like a slug on a 16GB SATA SSD, despite correct alignment etc etc.
On the same hardware, but with Win 7 32-bit, it was smooth as silk.

XP is still installing. I'll be investigating this further when it completes.

My installation disc has SP3 slipstreamed, if it makes a difference.

Actually, it just blue screened. I wonder if there is an issue with the PCI-to-SATA card I'm using.

Hours later it's still installing. This is slower than I would expect even at 533Mhz.

Kahenraz wrote on 2021-09-28, 04:48:

XP is still installing. I'll be investigating this further when it completes.

My installation disc has SP3 slipstreamed, if it makes a difference.

Actually, it just blue screened. I wonder if there is an issue with the PCI-to-SATA card I'm using.

might be, one way to test is to the use a regular hdd to see what happens

Yeah, SP3 is very bad news for older computers running XP 😀 Hardware requirements increased with SP3.

Use SP2 or earlier if you actually want it to sort of perform OKish.

I do remember using Windows XP on a Celeron 333 @415mhz. It was bad, but not as bad as you seem to be experiencing.

I remember XP been noticeably slower on C500 vs P2 400 in the early/mid 2000's.
I wouldn't call either fast but good enough as basic low end office PC's.
And we definitely had Windows 2003 running as a Dev box on our P2B-DS with just the 1 P2 400 installed, Can't remember the ram 256? 512?

I'd be blaming the slipstream media, I'm not sure if any of these machines saw SP2 and had definitely retired before SP3

It will be slow. Probably shouldn't be that slow however.

If I might ask though, why XP?
That system would be far more suited to Windows 98.

XP installs, especially low end hardware, optical to spindle, are best done when you're in the mood to pointedly ignore something for 4 hours or so.

In 2004 to 2006 I installed and used XP on lower spec hardware than that, such as setting my mother up with an Asus TXP4 and a K6-2+ @500, and I don't remember it being that slow. By the time SP3 came along, I had upgraded her to substantially faster hardware. Since it shouldn't go online, I'd agree that you shouldn't go any further than SP2 with that hardware. And the BSOD does point to some underlying issue other than just the slow CPU and low memory.

Jasin Natael wrote on 2021-09-28, 13:33:

If I might ask though, why XP?

I need to test this hardware on XP.

Does anyone have a GeForce 6600 AGP card to help test for a graphical glitch?

Windows XP setup kept blue screening on my P2B during setup so I moved over to my Gigabyte 775 with a Pentium D 3.4Ghz and setup is still taking hours to complete. This may be caused by the SiI3112 PCI-to-SATA controller I'm using. Further testing is required... if XP ever reaches the desktop that is.

I have several different BIOS versions to test but I need to get a working system first.

Kahenraz wrote on 2021-09-28, 17:15:

Jasin Natael wrote on 2021-09-28, 13:33:

If I might ask though, why XP?

I need to test this hardware on XP.

Does anyone have a GeForce 6600 AGP card to help test for a graphical glitch?

Gotcha.

Do you have the option of using a conventional IDE hard disk and skip using the adapter?

Yes but I wanted to test these adapters and see what they are capable of and what their compatibility is like. So instead of installing to an IDE drive I tried using a SATA card. That was a mistake.

After finally booting into Windows, the SiI3112 (the card I was installing with) seemed to work fine and performance benchmarked well. I don't know what the problem was.

I tried swapping to my SiI3114 which should be the same thing but with 4 SATA ports but performance completely tanked. This is with the latest non-RAID BIOS and F6 drivers installed.

So far I am not impressed with either of these controller cards but more testing needs to be done. Note that the above screenshots are on a Socket 775 with a Pentium D 3.4Ghz not the 533Mhz Celeron as mentioned in the title.

The SIL chips are crap from my experience. They were crap when they were built onto motherboards and they were crap when they were on PCI cards.

Back in the day, it was a very well known issue that you would get data corruption if you used them.

https://www.wilderssecurity.com/threads/fix-f … 112-sata.67812/ - possible fix for at least some things.
https://bugzilla.kernel.org/show_bug.cgi?id=3861
https://www.cdrinfo.com/forum/printable.asp?m=143300
https://club.myce.com/t/whats-largest-ide-har … drive/229360/24 - possibly fixed in 3114 and maybe better on 3112 when using 3114 drivers.
https://www.techspot.com/community/topics/lon … freezing.24920/
https://en.wikipedia.org/wiki/Talk%3ASilicon_Image
https://forums.tomshardware.com/threads/probl … araidev.700696/
https://linux-ide.vger.kernel.narkive.com/DZV … a-sil-3114-chip

But this does not explain the horrendous data corruption seen when using it in plain DOS... Boots a few times and then starts corrupting random files. Tried multiple cards with SIL chips and they all have one major issue or another.
Maybe a firmware update on the RAID versions could help but I am not optimistic at all.

Reminded me of why I hated them back when they were current tech.

cyclone3d wrote on 2021-09-28, 20:29:

The SIL chips are crap from my experience. They were crap when they were built onto motherboards and they were crap when they we […]
Show full quote

The SIL chips are crap from my experience. They were crap when they were built onto motherboards and they were crap when they were on PCI cards.

Back in the day, it was a very well known issue that you would get data corruption if you used them.

https://www.wilderssecurity.com/threads/fix-f … 112-sata.67812/ - possible fix for at least some things.
https://bugzilla.kernel.org/show_bug.cgi?id=3861
https://www.cdrinfo.com/forum/printable.asp?m=143300
https://club.myce.com/t/whats-largest-ide-har … drive/229360/24 - possibly fixed in 3114 and maybe better on 3112 when using 3114 drivers.

But this does not explain the horrendous data corruption seen when using it in plain DOS... Boots a few times and then starts corrupting random files. Tried multiple cards with SIL chips and they all have one major issue or another.

Reminded me of why I hated them back when they were current tech.

I agree all of the dealings with them I've had have been poor.

I found that I was using the wrong driver (RAID instead of non-RAID) for the Sil3114. I installed the correct one and it's about the same as the Sil3112 but a little slower on writes.

I have been reading a lot of negative information about both the Sil3112 and Sil3114 but I wanted to see it for myself. It's hard to tell but in the links you provided most of the people with issues mention that they are using the RAID feature. I'm using the non-RAID BIOS.

After the correct drivers are installed they seem to work fine in Windows XP on my Socket 775. I don't understand why setup took forever though. That was very strange.

What alternatives are there for PCI-to-SATA controllers for older systems like the 440BX?

Promise SATA1-150 controllers are the best I have used so far. They just work. There are a few models that don't have Win9x drivers though but those were mostly the super high end ones with a RAM slot for cache.

I have a few others I have managed to procure as well that have Win9x drivers but have not had a chance to test them out yet.

I installed Windows XP again on the same board with the same configuration but using the onboard ATA33 controller and it was much faster. I forgot to mention this before but with my first attempt I was using a Silicon Image SiI0680A IDE controller. I had assumed that it would be as good if not better than the onboard controller but it is actually very slow without any drivers installed.

So I can confirm that installing Windows XP on a 533Mhz Celeron does not happen particularly fast, it's not abysmally slow either. I would say that it probably takes about twice as long as a 1.4Ghz Pentium 3 but I didn't time it. Maybe an hour and a half or so to reach the desktop I'm guessing.