Once you add any device that uses ATA33, the other drive on the chain is forced to use the same speed. In which case its not an ATA100 being shared but instead its ATA33 being shared. Thats the sacrifice of backwards compatibility.
If you use the proper cables (ie 80 ribbon cable) you can mix ATA33 and up drives without a problem. Modern IDE controllers support independent reading of each device, so the ATA33 drive will read at its max speed, as will the ATA100 drive. Try it out. If you put an ATA33 and ATA100 drive on the same 80 ribbon cable Windows XP will report one device using UDMA2 and the other UDMA5. Test performance if you like, it will be the same either way if using the proper cable.
Dan, although what you say makes sense. You really need to do a real world benchmark.
You are arguing with someone who said:
Burn from CD to CD on same channel. Or burn from hard drive to CD on same channel.
Now compare that to burning either on different channels.
Now ocme back, and repeat the numbers - I dare ya.
I am going to decline your dare. Why? Because frankly, benchmarks don't mean jack sh!t to me. Regular use is the only "real worl benchmark" that matters to me. I use my computers both ways, and there is no noticable difference in my view. When I make backup CDs on my file server box. The CD-ROM and CD-RW are on the same channel, CD-ROM is master, and CD-RW slave. My times are about a minute, minute and-a-half slower than doing a CD to CD copy on my main rig. The main rig has each optical drive on a seperate channel. Of course, I will also mention the main rig has a 48x burner while the file server has a 32x burner. That
might account for some of the difference... It was different when ATA33/66 was the norm, this isn't the case with modern drives. Also, I apologize for failing to be more clear. I said
You will notice little, if any "real" increase in speeds.
That implied benchmarks may show a difference. But those few seconds are hardly critical. I should have been more clear.
\Dan