VOGONS


First post, by stanwebber

User metadata
Rank Member
Rank
Member

i have a kt133a setup and am being seduced by the option of having more than 1 isa slot. i skipped the whole k6 to early athlon era going straight from p2/3 to athlon xp. i searched a bit online, but the answers i'm looking for will probably take considerably more digging. i'm sure someone who lived thru this era could rattle off the answers in short order.

- i presume amd 751 officially only supports a 100mhz fsb. are there boards that allow overclocking to 133mhz with the requisite pci/agp dividers?
- do any amd 751 boards have cpu microcode updates available to recognize athlon xp processors (even at 100mhz fsb) for mmx/sse?
- i presume all athlon cpus up to xp palomino are compatble voltage-wise, but what about multiplier? i have a selection of xp palomino cpus using 11, 12, 12.5, 13x multipliers.
- i have a radeon 9800 non-pro and x1050. would there be any issues running these late-stage agp cards?
- is there a go-to amd 751 motherboard with 2 isa slots if you want to run an athlon xp?

Reply 1 of 13, by paradigital

User metadata
Rank Oldbie
Rank
Oldbie

Back in the day my first Socket A board (and first custom build) was a Gigabyte GA-7IXE4.

This was a 2x ISA slot board that I had no AGP 2x issues with, AND it supported overclocking the FSB up to 115MHz.

I originally used a TNT2 Ultra, but later a Geforce 2 Pro and finally a Radeon 8500 before retiring the system.

Looks like there is official support for the Athlon XP (Palomino) 1500+, which was a 200fsb part. That would suggest that you should be OK with any Palomino, albeit clocked down to a lower FSB.

Reply 2 of 13, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

Does anyone know why AMD didn't push the Irongate from the get go? They seemed perfectly content allowing shitty companies like VIA to dominate their platform. I avoided Athlon like the plague because I knew I'd get stuck with VIA, which had already burned me with their crappy Socket7 products. I also had roommates with Athlons using VIA chipsets, and it seemed like more of the same. Between shitty chipsets, bad caps and thermally inefficient CPUs, it was just a no win scenario in this era.

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 3 of 13, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++
Anonymous Coward wrote:

Does anyone know why AMD didn't push the Irongate from the get go?

But they did. That's original Slot A north bridge.

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 4 of 13, by Socket3

User metadata
Rank Oldbie
Rank
Oldbie
Anonymous Coward wrote on 2022-05-22, 04:59:

I avoided Athlon like the plague because I knew I'd get stuck with VIA, which had already burned me with their crappy Socket7 products. I also had roommates with Athlons using VIA chipsets, and it seemed like more of the same. Between shitty chipsets, bad caps and thermally inefficient CPUs, it was just a no win scenario in this era.

That was far from my experience.

My first VIA chipset board was a PC Partner MVP4 with a K6-2 400. On board trident blade 3D video witch was not the best thing for gaming but it was opengl and d3d compatible, and performance was more or less on par with a Riva 128. I had a great experience with that system in my teenage years. No lockups, no instability. It was this platform that thought me about overclocking (with the help of the internet and PC magazines of course). As such I managed to run my 400MHz K6-2 @ 450MHz by changing the multiplier and bumping the voltage to 2.4V. It did 500Mhz as well, but I was afraid of killing the chip so I'd only run it at that speed for limited periods of time - usually when playing Homeworld or Dungeon Keeper 2.

What followed was a string of AMD Socket A computers with VIA chipset boards. When I upgraded from my K6 in 2002 (I think) to a 850MHz duron, I kept the ram (192MB PC100 and 133 SDRAM), sound card (yamaha ymf-719), the hard drives, the optical drive and the Voodoo 2 I had purchased a year or so before, and moved them into the new PC. The mainboard was a Matsonic KT266 board with both SDRAM and DDRAM slots, and I had to buy an ATX case and PSU, as well as an AGP video card (a radeon 7500).

The system did have some issues, but not in the configuration listed above. The first problem reared it's head when I upgraded from the 850MHz duron to a Athlon XP 1800+, and I narrowed it down to the SDRAM it inherited from the K6-2. One of the sticks was PC100 - the 64MB one, while the other was PC133. The 1800+ uses 133MHz FSB, and the board also clocked the ram to 133MHz witch caused instability. After removing the PC100 module the PC worked perfectly - and I was doing some heavy gaming back then. Another problem arose when I upgraded to DDR memory. The motherboard was a cheap budget upgrade part, and I knew that going in. It wasn't very happy running 256Mb of PC2100 DDR.

So I sold the Matsonic and bought an Epox KT400A (8K9A if memory serves) motherboard, witch I kept for a bit - up until I upgraded to a Barton (2600+) core CPU. I had no problems with the Epox board. it was extremely stable, very compatible (took the 7500, a 8500, then later an FX 5200 Ultra). Loosing the ISA slot meant I had to get a PCI sound card, and since I couldn't afford a creative card I got a Genius YMF724 sound card - and the Epox ran with that sound card flawlessly as well.

I encountered no heat issues, no instability with the epox or later via chipset boards I used - I did again have an issue with a Albatron Nforce 2 motherboard witch i got together with the 2600+. I could not figure out the cause of the issue, but I traded the Albatron for an Asus A7V880 (KT880) and enjoyed two more years of trouble free computing.

By the time I was done with socket A, my configuration was: Athlon XP 3000+, Asus A7V880 Deluxe, 512MB of DDR400 (PQI brand) in dual channel, Leadtek A360 FX 5700, 80GB Matrox SATA HDD + 40GB Matrox IDE HDD and a 350W Antec PSU.

As for heat, I had no such issues. For socket A I used Titan brand coolers with 80MM fan and copper base and had never seen my CPUs go over 65C. I was a big fan of SpeedFan at the time and had CPU temps down in the taskbar near the clock at all times.

titan.jpg
Filename
titan.jpg
File size
45.84 KiB
Views
1082 views
File license
CC-BY-4.0

During one of my upgrades I had to live without my computer for a couple of weeks because I sold the motherboard and CPU and was waiting for new parts to arrive - so I borrowed a Pentium 4 machine from my dad's work. It was a 2GHz northwood running on a Intel D845WN desktop board with 256MB of PC-133 SDRAM. It came with a Matrox G450 video card and a horrible ADI or Analog devices (can't remember exactly) on board sound.

And... Impressions were not good. Sure it had a nicer case then the PC I had back then, it had a CD-RW drive (witch I nicked and replaced with my 48x Creative before returning it) but boy was it DOG SLOW compared to my 2100+ and Epox KT400 + DDR.... It took forever to boot into XP, although it was never really used at my dad's work - he was computer illiterate at the time 😁 - so it had a fresh win2k install with office 97 and that's it. Out of curiosity, since I'd never used an intel system before, I installed windows XP (big mistake) and replaced the Matrox G450 with my radeon 8500. It performed well enough in games - apart from some minor stuttering in GTA Vice City - but it got extremely HOT. I remember seeing 80 to 82C in speedfan while gaming. I also had problems getting my YMF724 to work correctly in dos as I still played some DOS games and would dual boot 98 and XP for that exact reason. The i845 would on that intel board would not play nice with my PCI yamaha, and I couldn't find DOS drivers for the on-board ADI codec. In fact this computer was the reason I kept away from Intel builds for as long as possible, way up until the Core 2 Quad "era".

It's true that I didn't experience the early VIA chipsets - the VP and VP3 for socket 7, or the dog slow 486 chipsets at the time. I was a teenager when socket 7 was current tech, and because of that my PCs were always one or several generations behind. PCs were either a tool for work or a luxury, one most could not afford. While young adults and IT specialists were wrestling with socket 5 and socket 7 PCs I was stuck on the 486 platform with a custom built (not by me) point of sales PC running a 133MHz cyrix 5x86 on a late PCI UMC motherboard - possibly a Biostar MB-8425UUD-A or some variant of the PCCHIPS M919 - I was to young to remember, but it had PS/2, it came with EDO ram not FPM and the layout and color I remember are very close to the Biostar. Don't remember if it had VLB but I know it didn't have a coast module like the M919 does, and it had a Dallas RTC. But I digress. Said IT specialists were complaining about VIA back then (socket 5-7 era), as well as OPTi and later SIS in the socket A era.

The point is I skipped socket 3 and socket 5 VIA chipsets. I only came upon the VP and VP3 chipsets when I started tinkering with old hardware, and I have to agree with what people were saying at the time, the VP is a mess. The VP3 is a bit better but it's relegated to bottom of the barred motherboards with poor build quality and cheap components. I loved the MVP4, and I bought my first MVP3 motherboard in 2004 or 2005 after upgrading to socket 939 for my main PC. It was then when I decided I needed a DEDICATED DOS games computer - so I got a second hand complete AMD K6-2 550 with a TNT2 M64 and a Lucky Tech P5MVP3. I still have the PC in the exact configuration it was back then. Same CPU, same (shitty) TNT2 M64, same case, same PSU and optical drives. Only the HDD and fans were replaced. Haven't used it in at least two years, as it was superseded by my K6-III build - witch is also based on an MVP3 board, the Aopen AX59.

Over the years I did tinker with some KT133 KT266 and VIA pentium 4 chispets witch did give me lots of trouble, but this has been in the last 5-10 years, and the boards I had trouble with were cheap and poor quality to begin with - not to mention old. A VIA KT133 from a reputable vendor like ABIT is a thing of beauty. Chaintech and Soltek made good KT266 and KT333 mainboards. Epox made great KT400 boards. Gigabyte made excellent KT600 and KT880's. The A7V880 from Asus is unfortunately not a great board, they did not age well - capacitor plague and whatnot. The MOSFETS Asus used are of questionable quality as well, so if you ever want to try VIA's answer to the nforce2, skip the Asus and look for a Gigabyte GA-7VT880 PRO or Abit KW7.

So there we are - my experience with VIA chipsets was a very good one.

Reply 5 of 13, by mockingbird

User metadata
Rank Oldbie
Rank
Oldbie
Anonymous Coward wrote on 2022-05-22, 04:59:

Does anyone know why AMD didn't push the Irongate from the get go? They seemed perfectly content allowing shitty companies like VIA to dominate their platform. I avoided Athlon like the plague because I knew I'd get stuck with VIA, which had already burned me with their crappy Socket7 products. I also had roommates with Athlons using VIA chipsets, and it seemed like more of the same. Between shitty chipsets, bad caps and thermally inefficient CPUs, it was just a no win scenario in this era.

Wasn't Irongate essentially a VIA chipset?

Look at the Wikipedia entry for "List of AMD chipsets". AMD 645 was a licensed VIA chipset, it stands to reason that their engineers used VIA designs later on as well. This is speculation.

mslrlv.png
(Decommissioned:)
7ivtic.png

Reply 6 of 13, by luk1999

User metadata
Rank Newbie
Rank
Newbie
stanwebber wrote on 2022-05-21, 18:39:

- i presume all athlon cpus up to xp palomino are compatble voltage-wise, but what about multiplier? i have a selection of xp palomino cpus using 11, 12, 12.5, 13x multipliers.

Palominos have (usually) locked multiplier, so even 13x shouldn't be an issue.
The question is that if such old board will run Palomino as even some early KT133A boards are not supporting them...

Pentium 4 2.4C, ASUS P4C800-E Deluxe, 1 GB RAM, GF FX5700 128 MB AGP, SB Audigy, Chieftec GPS-400AA-101A, Win XP SP2
Celeron 400, Compaq Garry, 128 MB RAM, Voodoo Banshee, ALS100 Plus+, Compaq 200 W, Win 98SE

Reply 7 of 13, by paradigital

User metadata
Rank Oldbie
Rank
Oldbie
luk1999 wrote on 2022-05-22, 16:57:
stanwebber wrote on 2022-05-21, 18:39:

- i presume all athlon cpus up to xp palomino are compatble voltage-wise, but what about multiplier? i have a selection of xp palomino cpus using 11, 12, 12.5, 13x multipliers.

Palominos have (usually) locked multiplier, so even 13x shouldn't be an issue.
The question is that if such old board will run Palomino as even some early KT133A boards are not supporting them...

As I said above, the Gigabyte board I had officially supports Palomino at 100fsb. That would suggest that microcode wise you are golden, even if you end up clocking down due to most being 133fsb parts.

Reply 8 of 13, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t
mockingbird wrote on 2022-05-22, 14:41:
Anonymous Coward wrote on 2022-05-22, 04:59:

Does anyone know why AMD didn't push the Irongate from the get go? They seemed perfectly content allowing shitty companies like VIA to dominate their platform. I avoided Athlon like the plague because I knew I'd get stuck with VIA, which had already burned me with their crappy Socket7 products. I also had roommates with Athlons using VIA chipsets, and it seemed like more of the same. Between shitty chipsets, bad caps and thermally inefficient CPUs, it was just a no win scenario in this era.

Wasn't Irongate essentially a VIA chipset?

Look at the Wikipedia entry for "List of AMD chipsets". AMD 645 was a licensed VIA chipset, it stands to reason that their engineers used VIA designs later on as well. This is speculation.

That's entirely possible. I actually suspected that at the time, but I couldn't find anything to back it up.
In any case, it would have been a smart idea to copy the intel model in capturing the chipset market for their own platform to better control the quality.
It would be interesting to know if the Irongate was a 1:1 copy of a VIA chipset, or modified in some way. The biggest problem with VIA was that they left the customers to do the beta testing for them. Sometimes with the right revision (usually a later one) you could get lucky. Maybe AMD had permission to modify the design to iron out bugs?

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 9 of 13, by jazper

User metadata
Rank Newbie
Rank
Newbie
luk1999 wrote on 2022-05-22, 16:57:
stanwebber wrote on 2022-05-21, 18:39:

- i presume all athlon cpus up to xp palomino are compatble voltage-wise, but what about multiplier? i have a selection of xp palomino cpus using 11, 12, 12.5, 13x multipliers.

Palominos have (usually) locked multiplier, so even 13x shouldn't be an issue.
The question is that if such old board will run Palomino as even some early KT133A boards are not supporting them...

Palominos do have a locked multiplier, but you can cut/trace bridges on the actual chips to change the multiplier - I remember wiring a power supply with some sewing pins to blow out bridges and set higher multipliers.

https://hubpages.com/technology/Nostalgia-Ove … Pencil-required

As an aside, I'm 90% sure that the Irongate chipset wasn't Via developed- it didn't suffer from any of the via issues of that generation, and via had many of them.

Reply 10 of 13, by mockingbird

User metadata
Rank Oldbie
Rank
Oldbie
jazper wrote on 2022-05-22, 23:29:

As an aside, I'm 90% sure that the Irongate chipset wasn't Via developed- it didn't suffer from any of the via issues of that generation, and via had many of them.

There's only one way to find out... De-cap and photograph the die.

mslrlv.png
(Decommissioned:)
7ivtic.png

Reply 11 of 13, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++
jazper wrote:

As an aside, I'm 90% sure that the Irongate chipset wasn't Via developed- it didn't suffer from any of the via issues of that generation, and via had many of them.

*laughs in forced AGP 1x mode on early Slot A boards

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 12 of 13, by swaaye

User metadata
Rank l33t++
Rank
l33t++

Yeah AMD 750 does have some AGP issues. All GeForce cards will likely run at AGP 1x and may still be unstable for example. There are a few steppings of the northbridge as well.

I wasn't even aware of Socket A AMD 750 motherboards until now.

I'm not sure when the AMD platform finally had rock solid AGP but it probably wasn't until AGP 8x. Though I have little experience with SIS Athlon boards.

Reply 13 of 13, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++

AMD AGP 4x is rock-solid implementation. Hell, it also predates KT266 boards.

I must be some kind of standard: the anonymous gangbanger of the 21st century.