The main bottleneck on most computers isn’t CPU speed, RAM speed, or FSB speed. It’s harddrive speed and RAM size, but for different reasons.
The most noticable difference would likely come from either a faster harddrive (say, from 5200 RPM to 7200 RPM) or more RAM. But more RAM will only help if you’re trying to do lots of stuff at one time. (For example, I usually have about a dozen webbrowsers open to various Straight Dope threads, while running the World of Warcraft computer game, while chatting off and on on MSN or GoogleTalk and downloading stuff on BitTorrent. While CPU speed certainly helps out, the amount of RAM is more critical for this kind of multi-tasking.) And a faster harddrive wouldn’t necessarily help programs run faster, unless they required harddrive access on a regular basis. However, it would help programs load at least a little faster (since Windows could read them faster with a faster harddrive).
Since Win98 SE is relatively old, there’s a limit to how much RAM it can usefully have. I’m afraid I don’t recall the exact amount, but I believe anything more than 512 MB is pretty much wasted. On a Windows XP system, you can have (if I recall correctly) 4 GB of memory. Assuming your motherboard supports that much, anyway. If you’re running a 64-bit Windows OS (which most people aren’t), it can handle much more than that.
After that comes CPU speed (although of course it depends on the particular system - a 450 MHz computer would probably be much better off having the CPU upgraded, if possible, than the harddrive or RAM). CPU speed is usually measured in gigahertz or megahertz, with higher obviously being faster when all else is equal. It should be noted, though, that often not all else is equal. A CPU from AMD with a certain speed might have more throughput than a CPU from Intel of the same speed. Think of a highway filled with cars. More cars can get to where they’re going faster if they all move faster. However, you can also get more cars to where they’re going faster by adding more lanes.
L1 cache and L2 cache are essentially RAM that is built right into the CPU chip or motherboard. (I may be wrong here, but I believe some CPUs now have L1 and L2 caches, so any such cache on the motherboard would be called a L3 cache.) What this does is make some operations a bit more efficient. Instead of having to go all the way to RAM to get data to work on, the CPU can operate on L1 or L2 cache instead. This is faster, but we’re talking fractions of fractions of seconds faster. If all you did to your computer were to get a new CPU that is exactly like the old one only with L1 and L2 cache, you probably wouldn’t notice any difference.
FSB = Front Side Bus. This is how fast the motherboard can route information to and from devices. It takes a very small amount of time for, say, the CPU to send information to the modem through the FSB, but that sort of thing is happening constantly so even tiny improvements can add up.
Of the examples you give, the one that would most likely give the most noticable improvement in speed for most people would be the jump in CPU speed from 600 Mhz to 950 Mhz. (I’m assuming you meant Mhz, not Hz, because that would be damn slow. 
) However, if you’re like me and often have lots of different applications open at one time, the increase of RAM from 256 to 512 would have a very noticable affect.
As geeky as I am, I am only of the lower pantheon of geekdom. There are mighty geek gods who roam this board, and I humbly ask that they fill in any gaps I’ve missed and correct any possible errors I’ve made.