It seems to me you're asking several questions wrapped into one verse.
First, you're asking about burning times
Second, you're asking about transcode times (ripping, DVDShrink stuff)
Third, you're asking about machine performance (non-burner performance).
Now, about IDE interfacing. Yes, the interface makes considerable difference in burn time. All other factors being of sufficient speed, the interface can limit your upper reach. Non DMA interfaces might limit your burst interface speed to something below 16Mbytes per second. A burn speed of 8x is going to need 10 Mbytes per second, without interruption. Add to that the time it takes to draw data off the source (your hard drive most likely), and you really need DMA. 16x burns require about 22 Mbytes per second, constantly.
One thing you didn't ask about is the hard disk speed. IDE Drives smaller than 40 Gbytes may only have sustained read speeds in the 15 Mbyte per second range. In the outer regions of the drive, the speed can be half. Drives at or larger than 40 Gbytes, and 7200 RPM rotation, you might find read speeds of 30Mbytes per second in the inner portion of the drive, but drop to about 15 Mbytes per second in the outer portion. This doesn't account for fragmentation, which invokes a seek penalty as the drive moves the head in search of data. There, too, the interface makes a big difference. If your drive could supply 30Mbytes per second or more, yet your interface isn't DMA, you're not going to get that speed.
Assume for a moment that your hard drive and interface are fine. RAM in the region of 256 Mbytes is actually sufficient to run a burn if you don't so much as touch the machine while it's proceeding, AND the software itself is efficient. You can tell if you're starving for RAM by using the task manager for XP or win 2000. If your VM usage (which you may have to turn on to see - using the "select column" option) climbs, and the CPU usage drops to nearly zero while it does, and that exceeds your available physical RAM, you're starving. 512 MBytes is quite sufficient, though - an increase beyond this isn't likely to return any dividend.
Depending on your interface setup, the drive's CPU utilization shouldn't reach more than about 30% or so at high speeds. If it did, you'd starve for processing, and you'll see several buffer underrun cycles. If that's happening to you, you're either starving for source disk speed, RAM (if under 512M) or maybe CPU power, probably in that order.
CPU power is an issue primarily in transcoding (shrinking) the video. DVDShrink appears to be threaded, so a hyperthreaded P4 or a dual CPU machine returns a benefit on that one application. Other transcoders might not be threaded, so P4's and duals wouldn't provide a gain.
A DVDShrink, with the high quality option on, can complete in under 40 minutes for a 2 hour movie on something about the speed of a 3Ghz P4. Without the HQ option, it might only need 10 minutes. At that point you're ready to burn on a 4.7G blank. At 8x speed, that would be another 12 minutes, perhaps - so a 3 hour cycle is quite a long wait.
30-40 minutes is about right for a P4 at 2Ghz, from DVDShrink through burn - if the HQ option is off. It might easily take an hour or so with HQ option on.
If your friend is at 3 hours, I'd suspect CPU speed over all else (but if it's the 1.5Ghz you're talking about - that's not right), with the hard drives coming in at a second option, and configuration issues a very likely third.
I'd think you were on a 700Mhz P3, or less, for a 3 hour DVDShrink to burn cycle.