Why Pentium 501

On Wed, 5 Jul 2006 14:08:54 +1000, "Rod Speed"

Sadly you're right, this is far too common.

Actually it does happen quite often in my experience. By far the most common cause of a heatsink coming lose from a CPU, not counting improper installation, is when it does so in shipping (either via a courier company or simply someone moving their computer). Most of the time the end result is that the heatsink tears totally lose from the motherboard and is sitting in the bottom of the case. The processor often comes along with it because cold thermal paste is pretty darn sticky stuff, more than sticky enough to yank the CPU out of the socket.

Interesting side note: I haven't seen this happen often enough with the new LGA-775 sockets to know how this would affect things vs. the old pin-on-CPU style sockets. I suspect the new pin-on-motherboard design would be better for surviving such a crash. Of course, the new LGA-775 sockets also hold the CPU in place MUCH more securely, so the chips probably wouldn't get yanked out in the first place.

I never had any issues with the CPU top on AMD chips, but I didn't much like how tough it was to clamp the heatsinks down. I was always more worried about accidentally scratching up the system board with the screwdriver used to mount the thing. AMD and Intel's newer solutions are much better.

I've been using nVidia chipsets for about 4 years now, and in my books their track record easily matches Intel's for reliability and drivers. Even if I were buying an Intel CPU I would strongly consider one of nVidia's chipsets for some of their features.

1GB of memory is cheap these days, no need to limit a system to any less. Beyond 1GB the performance advantage for most users is VERY small, much smaller than a faster CPU.

I think you're underestimating the average computer user. People DO notice differences in performance. They might not *care* that much, but that's not to say they won't notice it. This is true even for fairly simple tasks like web browsing, e-mail, word processing, etc. Note that when I mention the 20% number I'm talking about actual application performance, not some synthetic CPU benchmark or anything like that. Often times it takes much more than a 20% improvement in CPU performance to equal a 20% improvement in overall system performance.

------------- Tony Hill