SirNickity wrote:

Never liked Athlons. The chipsets were too hot-rodded for my taste (fast, and crashed even faster), and the design of the chips was questionable. Tiny exposed die. A heatsink retention mechanism that led to "screwing" up your brand new motherboard from trying to latch the retaining clips onto the socket. Thermal margins that required a high-speed fan and really secure HS, else ... smoke. Nah.

Thats surprising to me. Never really had much problem with the Athlon Chips. The tiny exposed die is never a problem if you are careful with heatsink installation/removable.

The motherboards work for the most part, but I admit the following problems:

a) The VIA 686B southbridge was really sucky.
b) The motherboards often have problems with capacitors.

Not that intel were much better. Their entire 810/820/840 line was utterly terrible, having drunken the RAMBUS koolaid and gone completely bonkers. And then Intel released the P4, which was largely a turd design, only usable because they cranked the frequency so hard (with the result of much more power usage and heat)

The entire Athlon/Athlon XP/late P3/P4 situation was a complete comedic show really in terms of reliability and stability. The best boards in this age were the simplest boards simply because they had less random crap to go wrong. Things only really improved once the Athlon 64 came out, and then got even better when Intel pulled their head out of their arse and released the core 2 and delivered the 9xx series chipset (and later the quite brilliant P3x and P4x series chipset)

Edit: I am quite offtopic it seems, but I can't help myself having a good rant. In terms of the VIA P3 chipsets, best thing to do is to leave it at 2x AGP. There is very little performance gain to be found even if you could reliably get 4x AGP working.

Exposed Athlon cores are no more an issue than Pentium III Coppermines, I fail to see the issue.

Never had a problem with KT333 and KT400 boards I used. I know some KT133 implementations were buggy, and I never used nForce2 because it was more headache than its extra performance was worth. I always had a great experience with AthlonXP era builds.

.... erased

I agree with both of you on most points, actually. The CuMine cores were also fragile. I never cracked one myself, but I worked at a computer store during that era and it was something we had to be careful about when dealing with product exchanges and service calls on our white boxes. I don't think the P-III cores were quite as prone to breaking, though, because the HSF assemblies weren't run to the brink like the Athlon XPs. Those took a lot of force to clip on, and the entire thing just seemed like -- is this really how we're doing this? OK...

And really, that's the basis of my opinion of AMD vs. Intel. AMD was obsessed with pushing the envelope. (And good for them, I guess.) Intel was more conservative and generally delivered a more polished product. Not that Intel was perfect. Like already stated here, the RDRAM fiasco was a blunder that would've put a smaller company to bed. I recall waiting out the "RDRAM or SDRAM -- pick your $%@^ sandwich" thing until you could FINALLY buy a P4 board with DDR on an Intel chipset. That coincided with the venerable Northwood cores. And THAT... is when the P4 finally came into its own. Not a great start, but they eventually got it together.

The Core line was a welcome return to sanity though. Northwood and on, P4s were pretty decent. But betting on clock speed was not a great long-term bet. Maybe not even right at the time. Nonetheless, I was pretty happy with my 2.0A for a while, only moving on to a Core 2 Quad 2.4 several years later.

Intel took massive shortcuts to increase their performance, though.
Meltdown and Spectre are a disaster and all of the software security patches now needed make older processors significantly slower.

We still haven't fully recovered from that even with the current generation of CPU's.
And all of that just to win the race against AMD. 😐

There is one big problem with Athlon XP chips (and P3 btw) that make them less attractive to be choosen as the core of a retro rig even when having better performance than their P4 counterparts. Specially if you will use the system in dual boot config with WindowsXP.

And that's the lack of SSE2 instruction set. You can judge if you need that or not. But most of the last apps for XP SP2 and 3 make heavy use of these instructions. Not only media players, but the JavaScript core of these browsers too. Is one thing that wouldn't matter for ones, and be decisive for others.

I'm going to lock this thread now. It's way off target.