A bunch of the newer PCIe video cards require UEFI support. No way around that unless you replace the BIOS on those video cards.

Had an issue with an Elsa Gloria L videocard (somewhat early OpenGL card for workstations). I got it from a recycler and it apparently was lying under something with a leaky battery, so the acid covered several legs on the main chip.

When I tried to clean it and run on a 440BX motherboard, it didn't POST and I thought the acid was the culprit. Months later I randomly installed it in a Socket 8 board and it worked! Still no idea what could be wrong with PCI on a 440BX board. Basically just a 2 years difference in period.

To give a less obscure example, I tend to have issues with 3C905 cards on early PCI/PnP boards (particularly S5 boards with Opti chipsets). Usually resort to either 3C590 or Realtek 8139 cards. Realtek has proven to be a real compatibility champ for me, the chip is even rated for a 40MHz PCI clock.

"Chinese stuff" in general, back then it was a Boeder USB scanner, in 2008 it was this then-ubiquitous webcam, last year an Avermedia (BT878) capture card - they all barely have drivers for whatever Windows was popular back then...

Of course part of the reason I got the latter is that I almost just picked up an XP system with two practically usable HDD connectors so not a single issue there, but the independent driver that also supports 32 bit Vista/7 has less options and seems to handle interlacing worse, while being unusable on Linux apparently because of wrong logical to physical input mapping - actually the same problem as integrated sound on my most modern PC...

Some more examples:

• Rage Fury Maxx that does not works with popular VIA chipsets from its period
• The Radeon 8500 that was crippled by bad drivers and was said to be inferior to the GF3 Ti500. While it performs better than the GeForce in last drivers.
• Some weirdness in product release dates. Like the fact that PIII Tualatin (2001-2002) came out months after the Pentium 4 (2000). And the tualatin integrates some technologies developped for the P4...

>Boards using earlier AGP chipsets occasionally do not work stable with newer and more power consuming AGP 3.3v cards such as TNT2Ultra and GF3Ti.

Nothing to do with chipset. Early AT AGP motherboards had to use onboard 3.3V regulator, some manufacturers seem to just copy/pasted those designs to ATX boards. Extremely stupid considering 3.3V present on ATX supply. Mod fixing this situation consists of soldering one wire.

My fav is all the confusion during ATA IDE standards being established in mid nineties. Multiple subtle hardware changes to pins and protocols making for a splendid mess. Thats on top of logical differences in later drives, CHS to LBA, LBA24 and LBA48 because idiots could see further than 3 years into the future 😀

You can contribute a lot of VIA chipsets quirks to such list.

One more that comes to mind about hard drives is SATA-II drives occasionally not detected on earliest SATA-I controllers unless you set a compatibility jumper or create a JBOD RAID (VIA, again)

The Serpent Rider wrote on 2023-03-05, 17:36:

You can contribute a lot of VIA chipsets quirks to such list.

You are absolutely right @The Serpent Rider 🤣

VIA ICs in general for me, not only motherboard chipsets (especially Apollo), How many computers using PCI cards equipped with VIA ICs of the same period (USB 2.0, IEE1394 or IDE) had issues I had to fix. The fix was generally to replace the card with something else, newer or not, and then try (with hope) to use the card somewhere else. And sometimes when put elsewhere the cards worked flawlessly. I had a pretty bad opinion of VIA in the end of the 90's. But maybe their chips were OK and the problem was the low-cost cards embedding them or the drivers. I don't know, but the result was this nightmare trying to get one to work with others cards.

That said, period correctness speaking of PC has no real sense for me as the main advantage of the PC is to be expended by nature, cf. the standards. The important thing when I'm building a PC is what I want to do with it. What software will run on it and so I chose the required components because the software dictates, OS and the drivers included. That is why I have several machines of the same period but not build to do the same thing, some having older or newer cards (in a 4 years interval).

One more that comes to mind about hard drives is SATA-II drives occasionally not detected on earliest SATA-I controllers unless you set a compatibility jumper or create a JBOD RAID (VIA, again)

That's pretty much only VIA problem. Again.

These incompatibilities are a reality, but also not what "period correct" means. For example it's not period correct to use a voodoo1 on a PIII just because the motherboard has a PCI slot, but to use it on a motherboard of around the same time.

SB Live on Via Apollo Pro is a constant annoyance.

Motherboards keyed for 3.3v AGP when they don't actually support it.

Sound cards that fail to initialize with anything faster than a 486.

When the somewhat ubiquitous 3 pin fan connector started being used, the pin out was not standardized leading to fun surprises.

P6 era motherboards, especially with multiple processors, can suck a lot of amps from the 5v rail at power up leading the short circuit detection being tripped on newer PSUs.

RussD wrote on 2023-03-06, 00:32:

When the somewhat ubiquitous 3 pin fan connector started being used, the pin out was not standardized leading to fun surprises.

P6 era motherboards, especially with multiple processors, can suck a lot of amps from the 5v rail at power up leading the short circuit detection being tripped on newer PSUs.

Compaq used this 3 pin fans on their computers, both on computer and their servers, that used them, no, it is for fault detection when fan is not turning at each power on, when fan is properly spinning the 3rd wire is kept grounded by the fan. Dell with their 3 pin black plugs same thing. Just ground the fault detection wire and use standard fans under-volted for quietness.

Cheers,

konc wrote on 2023-03-05, 20:45:

These incompatibilities are a reality, but also not what "period correct" means. For example it's not period correct to use a voodoo1 on a PIII just because the motherboard has a PCI slot, but to use it on a motherboard of around the same time.

Many people back in the day were using a wide range of 3d GPUs on an even wider range of PC configurations, is a Voodoo1 on a PIII unusual ..sure, does it work not really but a Voodoo2 would be just as applicable here "Period Correct" is a broad term and you shouldn't be trying to strictly limit what is and is not correct. So long as the hardware is of similar age a couple of years either side isn't an issue.

Back then you used what you had available to you, even older GPUs got use far beyond their use by date simply because Quake/Quake2/Unreal/Quake3 could run on potato video hardware, if all you had was a TnT on your P3 then you used it.

Bruno128 wrote on 2023-03-05, 14:11:

A few examples of such issues I'd like to bring up: […]
Show full quote

A few examples of such issues I'd like to bring up:

• Boards using earlier AGP chipsets occasionally do not work stable with newer and more power consuming AGP 3.3v cards such as TNT2Ultra and GF3Ti. Presumably voltage regulator is to blame.
• ABit KT7 family of some revisions do not work with newer AGP cards such as 6800Ultra despite having 1.5v support. It is thought that the reason behind is signaling levels.
• ASRock 775Dual and 4CoreDual series won't post with newer PCIe gen.3 cards despite implied backward compatibility between all PCIe generations including 1.0a used.
• Earlier DDR2-533 boards occasionally won't post with DDR2-800 modules despite JEDEC timings present.

Well there were many common issues back in late 90's and early 2000's with certain backward compatibilties.
The AGP thing was very common, is why later they added a Molex connector..
PCIe Gen 3 was never meant to be used on PCIe Gen 1 boards no matter what anyone says....
Yes not just DDR2 but DDR and DDR3 had such issues, mostly cured by a BIOS upgrade IF they released one...
No offense but did you grow up playing with computer parts (not playing with computer but the parts) during the 90's ?
Just curious, as there were many more issues back then than what you describe but am glad you did. Thanks !

Horun wrote on 2023-03-06, 03:20:

Well there were many common issues back in late 90's and early 2000's with certain backward compatibilties. The AGP thing was v […]
Show full quote

Bruno128 wrote on 2023-03-05, 14:11:

A few examples of such issues I'd like to bring up: […]
Show full quote

A few examples of such issues I'd like to bring up:

• Boards using earlier AGP chipsets occasionally do not work stable with newer and more power consuming AGP 3.3v cards such as TNT2Ultra and GF3Ti. Presumably voltage regulator is to blame.
• ABit KT7 family of some revisions do not work with newer AGP cards such as 6800Ultra despite having 1.5v support. It is thought that the reason behind is signaling levels.
• ASRock 775Dual and 4CoreDual series won't post with newer PCIe gen.3 cards despite implied backward compatibility between all PCIe generations including 1.0a used.
• Earlier DDR2-533 boards occasionally won't post with DDR2-800 modules despite JEDEC timings present.

Well there were many common issues back in late 90's and early 2000's with certain backward compatibilties.
The AGP thing was very common, is why later they added a Molex connector..
PCIe Gen 3 was never meant to be used on PCIe Gen 1 boards no matter what anyone says....
Yes not just DDR2 but DDR and DDR3 had such issues, mostly cured by a BIOS upgrade IF they released one...
No offense but did you grow up playing with computer parts (not playing with computer but the parts) during the 90's ?
Just curious, as there were many more issues back then than what you describe but am glad you did. Thanks !

The PCIe thing I thought most people understood that PCIe 3.0 cards for the most part are not fully backwards compatible, they might use the same name but PCIe 1/2 and PCIe 3 are very different beasts under the hood. I guess a few of the early PCIe 3 cards might be ok on a PCIe 2 board but I wouldn't bet on it unless they were hybrid and could fall back to the slower speed.

Same for the DDR compatibility, since it was all run through the north bridge compatibility was solely reliant on that and even a BIOS update may not fix it if the NB couldn't handle the higher speeds or voltages required. Nforce 780i/790i boards were bad for this with DDR3 support as later DDR3 modules actually used a much lower voltage 1.35v than the early DDR3 modules 2.1v did, thus compatibility was all over the place and it was hit or miss if your 1.35v DDR3 would even post in a board that expected DDR3 2.1v modules.

I will say this, I have a large number of the older 1.8v - 2.1v DDR3 modules and they are far more reliable than the low voltage modules, and they don't mind being pushed hard. That said I do have to keep them stored apart from the LV DDR3 as sticking 2.1v modules into a board that only wants 1.35v modules is no bueno.

AGP was honestly a headache at the end of its life .. you could see they were trying hard to standardize it or get it to a point where you had compatibility regardless of board or GPU ..8X was a good step but SLI/PCIe came in and pretty much pushed it into the obsolete bin.

There's a reason they sarcastically call leading edge "bleeding edge"... a neophile type PC enthusiast who had to have the latest and greatest always had a flaky science experiment of a machine waiting for drivers to catch up. Many enthusiasts were on the "rolling upgrade" plan, meaning CPU this year, GPU next year, Mem and HDD after, so they practically never had a whole period correct machine, right in the period where it would be "correct".

BitWrangler wrote on 2023-03-06, 04:07:

There's a reason they sarcastically call leading edge "bleeding edge"... a neophile type PC enthusiast who had to have the latest and greatest always had a flaky science experiment of a machine waiting for drivers to catch up. Many enthusiasts were on the "rolling upgrade" plan, meaning CPU this year, GPU next year, Mem and HDD after, so they practically never had a whole period correct machine, right in the period where it would be "correct".

And then you have the uhhh I'm not upgrading till they pry this machine from my cold dead hands types, where they refuse to upgrade for 10+ years and expect their dinosaur to not be extinct. The type who expect their Core2 Quad Q6600 to be able to handle the modern internet and 1440p HD gaming as well as a modern PC and bitch about it constantly when its like watching a gif slide show.

BitWrangler wrote on 2023-03-06, 04:07:

Many enthusiasts were on the "rolling upgrade" plan, meaning CPU this year, GPU next year, Mem and HDD after, so they practically never had a whole period correct machine, right in the period where it would be "correct".

That, I think, it still true today.