Pi A+ and 3A+.

I picked the A+ up about the time it came out, but within less than a year we had the Zero. Twice the RAM, USB gadget mode and 1/5 the price. And we lost... the DSI connector that wasn't usable properly for another 5 years?

Then the 3A+ comes out, and A-Ha! I figured there's no way they can put all that on a Zero! And to be fair, we got three years out of that before the 0 killed it. So, that's not so bad.

My problem is, I did all the "retrogame PC builds" when I was between about 9 and 14 years old, and 486 and Pentium chips were "I'll give you money to take this away". I'm sure there were hellish boards I had to deal with back then, but I just don't remember any more; then I got absorbed into legacy UNIX with UltraSPARCs and Indigos, and these days I'd rather emulate CPU's and motherboards, and focus on the parts of this stuff that are directly human interactive. Which reminds me, I really need to make a SUN-USB adapter for my 5c.

The biggest thing I actually remember as being a nightmare are the old blue and grey PowerMacs. Compatibility, reliability, stupid performance bottlenecks - everything from Yosemite to Snakebite. I hate that motherboard so much.

Fic 503. I wanted to upgrade to a kick ass Super7 board with AGP and a high performance K6-II that gave intel a run for its money that I could upgrade to a K3-III some day. Instead I got a cramped layout, jumper hell, instability, and the dreaded Via 4-1 drivers. Install them again! Do it in the right order this time! This next version will fix your problems!

Much of the disappointment came from my high expectations, the fact that I got it after I had used a rock solid 430TX motherboard with jumper free everything, and because it took so long for the K6-III to come out that the platform ended up being a dead end. From a retro stand point, it's not so bad because the drivers matured and I'm not trying to keep up with the Jones for the best frame rates.

Although I still have a certain amount of loathing for 386 vintage PC Chips mother boards, Flakey Nforce boards that almost work, and all the motherboards that used popcorn instead of capacitors.

douglar wrote on 2024-03-11, 20:58:

Fic 503. I wanted to upgrade to a kick ass Super7 board with AGP and a high performance K6-II that gave intel a run for its money that I could upgrade to a K3-III some day. Instead I got a cramped layout, jumper hell, instability, and the dreaded Via 4-1 drivers. Install them again! Do it in the right order this time! This next version will fix your problems!

Much of the disappointment came from my high expectations, the fact that I got it after I had used a rock solid 430TX motherboard with jumper free everything, and because it took so long for the K6-III to come out that the platform ended up being a dead end. From a retro stand point, it's not so bad because the drivers matured and I'm not trying to keep up with the Jones for the best frame rates.

Although I still have a certain amount of loathing for 386 vintage PC Chips mother boards, Flakey Nforce boards that almost work, and all the motherboards that used popcorn instead of capacitors.

In the summer of 2001 I ordered one of those from Tiger Direct, but it was on back order. Then I was given an old ATX case, so I changed my order to a Soyo SY-5EMA+. Tiger Direct mistakenly sent me a Soyo Socket A motherboard instead, which I was both happy and sad about. I was happy because it was better than what I had ordered and they had billed me for the 5EMA+, but I was sad because I now had to order a CPU.

Repo Man11 wrote on 2024-03-11, 21:49:

douglar wrote on 2024-03-11, 20:58:

Fic 503. I wanted to upgrade to a kick ass Super7 board with AGP and a high performance K6-II that gave intel a run for its money that I could upgrade to a K3-III some day. Instead I got a cramped layout, jumper hell, instability, and the dreaded Via 4-1 drivers. Install them again! Do it in the right order this time! This next version will fix your problems!

Much of the disappointment came from my high expectations, the fact that I got it after I had used a rock solid 430TX motherboard with jumper free everything, and because it took so long for the K6-III to come out that the platform ended up being a dead end. From a retro stand point, it's not so bad because the drivers matured and I'm not trying to keep up with the Jones for the best frame rates.

Although I still have a certain amount of loathing for 386 vintage PC Chips mother boards, Flakey Nforce boards that almost work, and all the motherboards that used popcorn instead of capacitors.

In the summer of 2001 I ordered one of those from Tiger Direct, but it was on back order. Then I was given an old ATX case, so I changed my order to a Soyo SY-5EMA+. Tiger Direct mistakenly sent me a Soyo Socket A motherboard instead, which I was both happy and sad about. I was happy because it was better than what I had ordered and they had billed me for the 5EMA+, but I was sad because I now had to order a CPU.

And those early Athlons performed great but were not cheap.

I was very hesitant to buy anything with a Via chipset after the FIC-503. The K6 motherboard was eventually upgraded to a slot-a FIC SD-11 motherboard. The AMD 750 Iron gate was not exciting, but it was stable, which was a relief. I gave it to my Aunt in 2004 and she still uses for her quick books on Win98se. She's 87 now. I should probably offer her an upgrade, yes?

Babasha wrote on 2024-03-11, 19:28:

Eagle Power586 NEC Power TX based on chineese counterfeit chipset.

https://theretroweb.com/expansion-chips/645

You'll be laughing but I had one of those.

It was so awful it wouldn't even boot off any HDD I threw at it. The only CPUs it took were ceramic Pentiums (100 through 150 as that's what I had at the time.). Oh, and you couldn't access the BIOS on it even if you tried.

douglar wrote on 2024-03-12, 11:35:

I should probably offer her an upgrade, yes?

Probably yes. The irongate board finds a new owner.

CharlieFoxtrot wrote on 2023-04-27, 07:59:

Abit NF7-S2G. I still has this board pretty much unused in box. […]
Show full quote

Abit NF7-S2G. I still has this board pretty much unused in box.

Back in the day, I fiddled a lot with Socket A platform. Bought cheap Athlons, tested several motherboards, had fast memory, watercooling and tweaked the hell out of the systems just for fun. Athlons were so much fun, because you could buy low end T-bred Bs and Bartons extremely cheap and overclock the hell out of them. I had a nice EPoX board before that turd and during some of my tests I somehow managed to break the socket. So, I needed a replacement and went to the computer shop relatively near my home. However, they had very limited selection available on the shelf, mostly just generic boards not aimed at the enthusiasts, so there was clearly a problem. Because I didin't have a functioning computer otherwise, I decided to pick this turd because hey, it is NF7-series after all, so how bad it could be?

It was horrible. It had no CPU multiplier controls and CPU support was abysmal (of course this turd pretty much never got any BIOS updates either, not to mention modded BIOSes). You couldn't even get most fast Athlons running at their nominal speed, because it didin't detect the processor correctly and used too small multiplier. You couldn't set CAS2 for memory. Every tweak it had in BIOS, which was not many, usually caused the system becoming horribly unstable.

It was probably after half an hour with that thing when I realised that I had made a huge mistake. I decided to keep the board so I had at least somewhat functioning computer for internet stuff for a time being and ordered a proper enthusiast board. If I remember correctly, it was another turd from Abit, AN7 which was just released at that point. Or I had AN7 before the EPoX, but the point still remains, that it was horrible too. It had all the bells and whistles, but it was disastrously unstable. I just couldn't get it stable with 2 or more RAM sticks even stock and I got it replaced for NF7-S v2.0 after few days of frustration (or EPoX board, if my memory fails me of the order of things)

But NF7-S2G still lingers with me. I couldn't bear to sell it to anyone so I just left it in my bin with a box and accessories as a reminder. Just few weeks ago I browsed the contents of the box. It still had the Tbred-b 1700+ in the socket. Nice processor, overclocked like a beast with water. I think I need to remove it from the board, it doesn't deserve to be tortured with that board.

The AN7 is amazing. They are so good I got a hold of and kept two of them. The NF7 however... well, mine has no stability issues whatsoever, but it does lack any usefull overclocking features. I have no idea why Abit released the thing.

I did however have severe stability issues, back in the day and rather recently, with a couple of Epox boards... 8RDA3 and 8RDA+ to be precise. The 8RDA3 is fine as long as I don't use anything faster then a 2600+ barthon and don't try to use any fast ram (like 400MHZ CL2). The 8RDA is just generally unstable. It can work fine for weeks, then decide to corrupt the hard drive the next day. I do remember the 8RDA3 was finnicky, I had one when they were new and I was so happy when I replaced it with an Albatron KX18D PRO2 witch really surprised me. Cheaper then an AN7, but with similar features and overclocking ability. One of the best boards I've ever had.

As for my "worse boards", oho, there is a list:

- Abit KV7 - I have a few of these, and really wanted to build a KT600 + Athlon 2600+ (AXDA2600DKV3C, toroughbread, 2133MHz) PC. My boards are unstable, at least with this CPU. One of them worked fine with a 2200+, did several benchmark loops for 6 hours streight with that chip. Put the 2600+ in it and it locks up in 30-60 minutes. Dissapointing, as I remember these were among the better KT600 boards back in the day. Also picky about ram.
- Asus P4P800 Deluxe - I have a box of these. Issues with the EEPROM and/or CPU socket. Fast when working, they were well priced back in the day, but boy do they like to die.
- Asus A7N8X - very picky about ram. Had issues running some board with FSB200 CPU's. Issues getting the systems stable with dual channel ram configurations.
- Asus P5K - had one back in the day - would not say it's a horrible board, but it is slow, and tends to get unstable when using faster more power hugry chips.
- Asus P5A - any revision. Again, I have a box of these. Some even post, but throw a "Gate A20 error" when loading himem.sys.
- Epox 8RDA3 - generally unstable board.
- MSI MS6167 - slot A board. I love the fact that it has two ISA slots - BUT, out of 3 boards (got a hold of another one last weekend) none worked properly. One even let out magic smoke. Mosfets near the CPU slot like to cook themselves.
- Gigabyte X79-UD7. Good board apart from a wierd bios issue. It would randomly loose CMOS settings. Otherwise ran fine and is a great overclocker. Had one of these when it was new, paired with an i7 3930k. It could overclock the snot out of that chip, but once in a while it would fail to post, loop 2-3 times, and I'd get a "cmos settings invalid" error. Good thing the BIOS had a "save configuration" option. Found another one last summer - pretty cheap, picked it up. Played with it for a few days, them BAM - CMOS configuration invalid all over again. My brother in law had a Z77 board with the exact same issue (can't remember the exact model).
- Gigabyte Z77-UD3M - the most pointless Z77 board ever made. It did not have any voltage settings in bios, severly limiting overclocking potential and as such negating the whole reason for buying a Z77 in the first place.
- Matsonic socket A boards. These were fine if you didn't try any sort of tweaks and stuck with cheaper CPUs. I remember trying to get a 1700+ running on some KT133A matsonic board - could not get it running. Came across the board recently again - by some accident the orange heatsink fell of the northbridge - and surprise - it's a regular KT133 - not an "A" model. No wonder it had so many issues with FSB133 chips. It was marketed as a KT133 tough. Hell, even the BIOS string sais "KT133A".
- Most if not all socket 478 SDRAM boards. These things are HORRIBLY slow. I don't know if it's a fault of the boards themselves, since I've tried Asus, DFI, Gigabyte, MSI, OEM stuff from IBM and HP, and they're all dog slow - even compared to a regular 1GHZ pentium 3. Some are even unstable, and down-right unusable, particularly models from PCChips and Matsonic.

Generally most MSI or ASUS 2001 to 2004 motherboards. In fact the first good MSI board I've seen was the MSI P35 Neo2. Budget asus boards in that period were horrenodus. Started with mediocre boards in 2001 and just kept getting worse and worse. Their high end stuff was OK tough. I have a P4C800 that's been kicking since 2006, and it was used as a file server at one point.

Socket3 wrote on 2024-03-12, 21:15:

- Most if not all socket 478 SDRAM boards. These things are HORRIBLY slow. I don't know if it's a fault of the boards themselves,

More the fact that the Pentium 4 (Willamette particularly) was a horrible CPU design that was slower than the Pentium III and was designed for much faster RDRAM and DDR memory? 😉

I used one for years, (1.7 Willamette, SDRAM) before finally scrapping it around 2007. I was honestly shocked when around the same time I picked up a couple of trash servers for about £10 each - dual Tualatin 1.26 (eventually with 2GB SDRAM, think they were less on arrival) - and found they significantly outperformed my old P4 as an Ubuntu server.

I have no nostalgia for Netburst.

Aye, lemme dig:

- ABIT IP35 Pro - rare for me to mention one. But GOOD GOD I've seen myriads of these. Specifically one of the RAM FETs runs at alarmingly high temperatures.
- any ECS-made ABIT - yes, these exist. They're only any better than the original ECS boards.
- P5K, as @Socket3 mentioned - these are the worst to exist ever - none I have ever POST'd, and have had bad CPU socket BGA. The latest one is even worse with bad solder under southbridge as well and the northbridge was smoked already.
- A7N8X of any kind - NONE to this day have ever ran a Barton stable. Same chips used on the A7N8X boards would work fine on EP-8RDA6+ and K7N2 Delta-ILSR. Heck even a puny Jetway N2PAP-LITE board ran these properly.
- A8N-SLi - bumpgate, nuff said. No, FYI nForce 4 isn't safe from bumpgate either.
- GB H55M-USB3 - underspec'd VRM. I have one and fixed the VRM flaw with a few parts off a dead HD5850M MXM GPU. Works fine but boy does that NEC USB3 chip run hot.
- PCChips M810 - either I'm severely dumb or these things don't even have a working sound chip whatsoever. The red one barely works properly (rev 7.1 w/ socket, surprise surprise) and has a VIA soundcard, while the green one works fine but has an craptacularly abysmal ALS100, to the point I wonder if it's a wise choice to go Aureal.

@Socket3 - do you still have any of those 6167s around? I'd be interested in testing out if I can get one going again by moving it from THT MOSFETs to SMD. I got a LuckyTech P6BX2 that has had its original FETs+heatsinks harvested (see photos) and used SMDs to get it up and running - it's still working and they seem to run much cooler.

InTheStudy wrote on 2024-03-12, 22:16:

More the fact that the Pentium 4 (Willamette particularly) was a horrible CPU design that was slower than the Pentium III and wa […]
Show full quote

Socket3 wrote on 2024-03-12, 21:15:

- Most if not all socket 478 SDRAM boards. These things are HORRIBLY slow. I don't know if it's a fault of the boards themselves,

More the fact that the Pentium 4 (Willamette particularly) was a horrible CPU design that was slower than the Pentium III and was designed for much faster RDRAM and DDR memory? 😉

I used one for years, (1.7 Willamette, SDRAM) before finally scrapping it around 2007. I was honestly shocked when around the same time I picked up a couple of trash servers for about £10 each - dual Tualatin 1.26 (eventually with 2GB SDRAM, think they were less on arrival) - and found they significantly outperformed my old P4 as an Ubuntu server.

I have no nostalgia for Netburst.

Willamette with SDRAM was about 1GHz slower clock for clock than a PIII. For example a 1.8GHz was like a 850MHz PIII in overall benchmarks. I had customers that insisted on the P4 because BIGGER NUMBERS!!! I tried to tell them the PIII could likely be faster and cost way less. They never listened.

PcBytes wrote on 2024-03-12, 22:46:

Aye, lemme dig: […]
Show full quote

Aye, lemme dig:

- ABIT IP35 Pro - rare for me to mention one. But GOOD GOD I've seen myriads of these. Specifically one of the RAM FETs runs at alarmingly high temperatures.
- any ECS-made ABIT - yes, these exist. They're only any better than the original ECS boards.
- P5K, as @Socket3 mentioned - these are the worst to exist ever - none I have ever POST'd, and have had bad CPU socket BGA. The latest one is even worse with bad solder under southbridge as well and the northbridge was smoked already.
- A7N8X of any kind - NONE to this day have ever ran a Barton stable. Same chips used on the A7N8X boards would work fine on EP-8RDA6+ and K7N2 Delta-ILSR. Heck even a puny Jetway N2PAP-LITE board ran these properly.
- A8N-SLi - bumpgate, nuff said. No, FYI nForce 4 isn't safe from bumpgate either.
- GB H55M-USB3 - underspec'd VRM. I have one and fixed the VRM flaw with a few parts off a dead HD5850M MXM GPU. Works fine but boy does that NEC USB3 chip run hot.
- PCChips M810 - either I'm severely dumb or these things don't even have a working sound chip whatsoever. The red one barely works properly (rev 7.1 w/ socket, surprise surprise) and has a VIA soundcard, while the green one works fine but has an craptacularly abysmal ALS100, to the point I wonder if it's a wise choice to go Aureal.

@Socket3 - do you still have any of those 6167s around? I'd be interested in testing out if I can get one going again by moving it from THT MOSFETs to SMD. I got a LuckyTech P6BX2 that has had its original FETs+heatsinks harvested (see photos) and used SMDs to get it up and running - it's still working and they seem to run much cooler.

As a matter of fact I do. I'll PM you with details. Maybe you could help me get an ECS P6S5AT going as well...

Lots of pc chips boards pushing here!
I'm gonna add two:
- pc chips M520. Mentioned before. The VX Pro chipset is a dud, failing to compete with anything else on the market. Fake cache next to real cache (?) and horrible layout near the power connector. It just misses winbios to complete the list.
- pc chips M530. Let's bad than the previous. It has sdram slots just for the looks bit doesn't actually boot with sdram. Have to stick with edo. With it's i430VX it should be an ok performer but on all edges it lacks about 10% compared to decent boards.

Also had a HP Vectra 486 which was easy too slow, probably because it had no cache at all and couldn't even be added.

Once I had a Dell Celeron s775 system. I was planning to upgrade it in time, but ended up not to. Terrible board, terrible company, because they didn't add the AGP slot (!!!) The PCB was prepared but they saved a whole euro.

Last but not least: my Multitech MPF-PC does not seem to cooperate, i may choose between booting from hard disk with no keyboard working or booting from fdd while not accepting the boot sector but keyboard keeps working.

As much as I bash PCChips, they have some okay-ish boards as well. M577 and M726MRT (the combo slot version w/ 370 and Slot1) seem to be fairly stable, although a bit slower than usual boards (MVP3 and Aladdin Pro 2)

waterbeesje wrote on 2024-03-13, 19:19:

Also had a HP Vectra 486 which was easy too slow, probably because it had no cache at all and couldn't even be added.

Let's all hail HP for them inventing the idea of the Covington Celeron core before the Pentium Pro was even a thing!

Running a 486 system with contemporary RAM without L2 cache is clearly a bad idea, especially as most 486 chipsets can easily provide a 2-1-1-1 burst at 33 MHz FSB clock on L2 cache hits.

my worst boards i got

ECS P5VX-A - no built in L2 cache, USB ports don't do anything, BIOS is too lackluster, and no MMX support
Young Micro Systems 486 VEGA Green - no L2 cache seems to work at all on it, even took known good ICs from a SiS 471 board and no change, no 3.3v support, and seems to lag behind even a OPTi 895
PCChips M919 V3.4B/F - Requires proprietary L2 cache modules, VLB slot is finicky, and also quite unstable and would randomly lock up

A7N8X-E Deluxe: Lacking the 4-pin CPU power connector, Silicon Image 3112 SATA controller stability issues when using RAID.

PD2JK wrote on 2024-03-13, 21:43:

A7N8X-E Deluxe: Lacking the 4-pin CPU power connector, Silicon Image 3112 SATA controller stability issues when using RAID.

The -X version isn't any better (I run one at the moment of writing this post). I do wonder, does the 3112 and 3114 have this issue on anything else besides ASUS?

PcBytes wrote on 2024-03-13, 20:52:

As much as I bash PCChips, they have some okay-ish boards as well. M577 and M726MRT (the combo slot version w/ 370 and Slot1) seem to be fairly stable, although a bit slower than usual boards (MVP3 and Aladdin Pro 2)

Ah, the M577 is actually my super 7 testing board of choice! It's a bit slower than you'd expect from mVP3 but it literally eats any CPU you throw at it 😁

And the pcb isn't made from cardboard as well, like my M537dma33 for example

(Yes, I do possess my fair share of pc chips randomness)

mkarcher wrote on 2024-03-13, 20:53:

waterbeesje wrote on 2024-03-13, 19:19:

Also had a HP Vectra 486 which was easy too slow, probably because it had no cache at all and couldn't even be added.

Let's all hail HP for them inventing the idea of the Covington Celeron core before the Pentium Pro was even a thing!

Running a 486 system with contemporary RAM without L2 cache is clearly a bad idea, especially as most 486 chipsets can easily provide a 2-1-1-1 burst at 33 MHz FSB clock on L2 cache hits.

In all my personal machines L2 cache only makes like a 5% difference overall with a 486 that has 16K L1. Maybe all my machines have shitty chipsets.

My most shitty PC Chips motherboard I have is an M741LMRT .. it has defied me... I've got about 7 or 8 other PC Chips that are fine for what they are, even a couple that impress me. This one though, has all the frequently complained of faults. Thin board, bad soldering, general jank.

The shitty to working ratio of PC Chips though is HIGHER that that of Asus boards here, with a dead A7V333 A7V133 P2B P3B among those number. So Asus actually gets first place for shitty boards, because 30% of their boards I have don't work right.
Gigabyte comes second with 20% Though on a lot fewer boards and I'm angrier at them because some were bought new enough that warranty replacement should have happened but they were all deny deny deny.
MSI comes third with 15% shitty... though they would tie with gigabyte if I dinged the neo board for it's lack of upward CPU support.
PC Chips actually comes in at 4th most shitty overall at 12%

Maybe I should put noughties Dells in zeroth position though, but it's a whole system thing, the boards are artificially brain damaged for memory and graphics support and PSUs don't have enough drive connectors. Work in supplied config, but real throwaway engineering.