> most other chipsets of the time were really, really, really bad (*cough* VIA Apollo Pro 133 *cough*) - they were slow, had terrible compatibility issues (including memory compatibility, peripheral compatibility) and as for stability

Maybe they were bad at the time, but the slowness of the Pro133 doesn't matter when the goal is to build a PC that can go slow, and they tend to be more compatible with Via C3 (not surprise) which is the preferred CPU for s370. And as for stability...for the occasional retro gaming PC does it really matter? The main thing that annoys me is how SB Live! can break on this chipset, but there is little reason to use it over an Audigy 1/2.

Never heard of it sorry...😉

There is one instance where I found that having an Intel-based board matters - using a SATA to IDE adapter with a JMicron chip.

From my experience, these work great on any Intel chipset, but try it on a VIA board (KT133A in my case) and you get weird, stuttery and generally unstable behavior. I have heard that adapters with a Marvell chip are better in that regard, but I haven't gotten my hands on one of those yet.

kolderman wrote on 2021-07-25, 18:52:

> most other chipsets of the time were really, really, really bad (*cough* VIA Apollo Pro 133 *cough*) - they were slow, had terrible compatibility issues (including memory compatibility, peripheral compatibility) and as for stability

Maybe they were bad at the time, but the slowness of the Pro133 doesn't matter when the goal is to build a PC that can go slow, and they tend to be more compatible with Via C3 (not surprise) which is the preferred CPU for s370. And as for stability...for the occasional retro gaming PC does it really matter? The main thing that annoys me is how SB Live! can break on this chipset, but there is little reason to use it over an Audigy 1/2.

Yes, I agree that VIA C3 tends to work very well with VIA Apollo Pro 133 (and I do like the more granular ACPI throttle capabilities that VIA offers)... but, man, this chipset can be EXTREMELY bad/fussy (especially on some boards) - even the VIA MVP3 can look like a "mini-440BX" next to it 😀

However, the VIA C3 "Nehemiah" also worked great on most 440BX motherboards that I tried with it (as long as those boards had Coppermine support).
I disagree, though, a Nehemiah is great for its flexibility, so you want that system to be as fast as possible and also as slow as possible. And when it comes to Windows 98 speed, the 440BX wins, hands-down (7400 points in 3DMark2000 with Nehemiah @ 1.53 GHz, compared to 5500 points on VIA Apollo Pro 133 boards - that's a 34% speed difference with the same CPU, same overclock).

Joseph_Joestar wrote on 2021-07-25, 19:05:

There is one instance where I found that having an Intel-based board matters - using a SATA to IDE adapter with a JMicron chip.

From my experience, these work great on any Intel chipset, but try it on a VIA board (KT133A in my case) and you get weird, stuttery and generally unstable behavior. I have heard that adapters with a Marvell chip are better in that regard, but I haven't gotten my hands on one of those yet.

My experience with SATA/CF/SD to IDE adapters is mixed. They can work great on some boards (no matter the chipset), but they can show erratic behavior on other boards with the same chipset (including 440BX).

Quality boards are important just like any other chipset.

I use an Asus CUBX-E with a P3 1.4S (133 fsb) for my retro gaming needs and it's quite nice.

5900XT graphics card.

Asus CUBX-E + Tualatin CPU

I was mainly an AMD dude, but I had zero luck with BX when they were less than 5 years old, used stuff, in the 5 to 10 year frame managed to find/get a running PcPartner BX that is probably still gonna be limited to Katmai only, and a P2-99 ZX version of the P2-B, but still got a box of P2Bs and maybe a couple of Abits somewhere that just wouldn't play nice. Just resurrected 3 out of 4 VIa boards, so on a roll (touch wood) with those right now, so am currently feeling far more kindly disposed to Vias than BXes.... I always wonder when I see the crappy via benches whether they are set up right, having the 4x interleave and AGP turbo enabled, with a later Via 4 in 1 driver installed before the gpu drivers.

I can only say that asus and 440bx are great. Even today it is a rock.

I can not say so much in defense of their competition, even later years did not reach their height.

I'm going to have another crack at them, but the pile of dead Asus P2Bs (various subvariants) is what makes up the bulk of the 80% dead, 20% working of my Asus collection*, vs my 20% dead 80% working PC Chips collection. I have a theory that PC Chips engineers knew they were gonna be using crappy components and designed for crappy components, whereas everyone else's engineers were way too over optimistic about the quality of components the boards would end up with, whether from post release beancounter scrimping or supply line problems or whatever, so in designing for the best available components made the boards super sensitive to anything a little below par.

* collection is a bit too posh a word really, it's just the bunch of old junk I happen to have 🤣

I like the basic strip down Intel BX440 motherboard with NO onboard Audio, Video, or Networking.
This way you can customize it anyway you want with any assortment of cards ( ISA, PCI, AGP )
It can run DOS/Win3x/WinNT351/WinNT40/Win95/WinME/Win2000/Solaris x86/FREE BSD/Linux/Novell/etc…

The AMD k6 CPU’s and Via chipset motherboards are good for gamers. And you can easily down clock these to 386@20mhz
For playing old CPU sensitive DOS games.

So if you wanted just ONE computer for Win98/DOS gaming I would choose the AMD K6-2 or K6-3+ With a Super socket 7 motherboard with VIA 133 chipset. If you don’t care about ISA cards then maybe an AMD Duron 600/800 motherboard.

BitWrangler wrote on 2021-07-25, 20:51:

I'm going to have another crack at them, but the pile of dead Asus P2Bs (various subvariants) is what makes up the bulk of the 80% dead, 20% working of my Asus collection*, vs my 20% dead 80% working PC Chips collection. I have a theory that PC Chips engineers knew they were gonna be using crappy components and designed for crappy components, whereas everyone else's engineers were way too over optimistic about the quality of components the boards would end up with, whether from post release beancounter scrimping or supply line problems or whatever, so in designing for the best available components made the boards super sensitive to anything a little below par.

* collection is a bit too posh a word really, it's just the bunch of old junk I happen to have 🤣

I'm glad it is your gunk and not my junk mate...😉

Nothing wrong with Asus P2B boards (unless you mistreat them). Have a few and they still all work great and never replaced the caps yet (was pre-bad cap era). just my opinion 😀

It's a myth, there are a lot of 440BX boards that have issues with high power draw AGP cards for example.

The BX boards struck me as reliable and high performance. I still have a few which are my go to option for testing various cards.

A good board that doesn't work with a series of drivers, but is perfectly fine after a certain release - so what are we trying to "destroy"? Straws being clutched ... 😀

kolderman wrote on 2021-07-25, 18:52:

> most other chipsets of the time were really, really, really bad (*cough* VIA Apollo Pro 133 *cough*) - they were slow, had terrible compatibility issues (including memory compatibility, peripheral compatibility) and as for stability

Maybe they were bad at the time, but the slowness of the Pro133 doesn't matter when the goal is to build a PC that can go slow, and they tend to be more compatible with Via C3 (not surprise) which is the preferred CPU for s370. And as for stability...for the occasional retro gaming PC does it really matter? The main thing that annoys me is how SB Live! can break on this chipset, but there is little reason to use it over an Audigy 1/2.

One of my motherboards has the Apollo Pro 133A at maximum memory capacity theres a slight pause on post after counting.
But I also know it accepts server ecc ram that the bios doesn't support so essentially a server chipset.
(I ran a ecc module bios didn't realise system booted normally)

Byrd wrote on 2021-07-26, 07:11:

A good board that doesn't work with a series of drivers, but is perfectly fine after a certain release - so what are we trying to "destroy"? Straws being clutched ... 😀

No, a good board will work with any card, any driver I throw at it. In fact, I have quite a few of these boards (either with the 440BX chipset, or with newer VIA chipsets like the KT400/KT600/KT880).
The fact that it works fine after a certain driver release, just tells me that nVIDIA probably applied a dirty fix (or even more plausible is that they just did a random driver change that, for some reason, acts as a workaround to this specific problem).

Also, it's easy to say this now, when we have all possible drivers readily available, but let's not forget that the driver that actually fixes these issues (45.23) was released in August 2003. By that time, most people that had used a 440BX platform had already upgraded to a Pentium 4, Athlon XP, etc.

Intel486dx33 wrote on 2021-07-25, 22:18:

a Super socket 7 motherboard with VIA 133 chipset. If you don’t care about ISA cards then maybe an AMD Duron 600/800 motherboard.

There's no such thing as a 'VIA 133' chipset for SS7. Also a Duron 800 can be used with chipsets with an ISA slot. Could you please stop posting incorrect 'advice'.

BX was an all in chipset, but those 800 series crap were crippled because of impeeding P4 or well that RDRAM fiasko. And yes I even had a Athlon XP machine with a DMA functional ISA slot. can not remember the model, but something from Asus.

Cyberdyne wrote on 2021-07-26, 12:18:

BX was an all in chipset, but those 800 series crap were crippled because of impeeding P4 or well that RDRAM fiasko. And yes I even had a Athlon XP machine with a DMA functional ISA slot. can not remember the model, but something from Asus.

Those RDRAM 800-series chipsets were actually pretty good if used with RDRAM. They offered similar (i820) or better (i840) performance to i440BX, introduced useful new stuff (133MHz FSB, ATA-66/100, PCI 2.2, AGP 4x, AC'97, and moving PCI controller to the southbridge with a faster interconnect between the bridges) and even if we don't like it, ditching legacy like ISA made sense looking forward, not back, particularly if you consider the way legacy driver code was the main driver of Windows instability. If RDRAM had been affordable, Rambus' business practices less revolting and the whole Caminogate and MTH-mess had been skipped, they would have been a fine evolution. Even today I like them for early 2k/XP builds.

Yes, i810 was hopelessly crippled and i815 was also limited in terms of memory (although max 512MB was hardly an issue in 2001), but they were low-end chipsets and no worse than contemporary Via or SiS alternatives (SiS 630 or Via PLE133 anybody? 😦 )

Intel's memory strategy was a complete mess at the time, but where they let engineers do their best without the politics (so no i815 hobbling, i820 with MTH or i845B with single-channel SDR-SDRAM) the results were excellent.