ernstoud

Hi all, Great thread. Thanks to Tech Inferno Fan for this. I own a number of old HP nc6000 systems with two HDD's with Windows XP on one and Windows 7 on the other (still using some software that only runs on XP). Windows 7 runs from the HDD in the multibay and speed was capped at 30 Mb./sec at UDMA2. From this thread I learned from experimenting with Bar-Edit (great tool!) and reading the ICH4M datasheet that for some reason the BIOS programs the secondary IDE controller for UDMA2 instead of UDMA5. I am now using a batch file in Startup that contains the proper setpci and hdparm commands to enable UDMA5 and an AutoHotkey script that detects the Windows message that a resume occurred which triggers the same batch file to be run after a resume. I now enjoy more than 50 Mb./sec, full UDMA5 on the drive in the multibay. Next step will be installing a Transcend 128 Gb. PATA SSD I just bought in the multibay. From Tech Inferno Fan's post I don't understand a number of issues: 1. He says you need to program the 54h, 48h, 4Ah and 4Bh register by: setpci -s 0:1f.1 54.l=3033 setpci -s 0:1f.1 48.l=10001 However, since configuration register 48h is 8-bits according to the Intel datasheets, the last setpci command sends a 32 bit long word to a byte wide register... effectively also overwriting 49h, 4ah and 4bh. The same for the first setpci command; register 54h is 16-bits so the command should be: setpci -s 0:1f.1 54.w=3033 I use: setpci -d 8086:24CA 4B.b=1 setpci -d 8086:24CA 54.w=5055 Which works fine. (Note that in a nc6000 the multibay drive is secondary master, not slave so register 54 is not to be 3033 but 5055). 2. Why do we actually need HDPARM? Why is setting the PCI registers not enough? What does HDPARM do? I guess it triggers the Windows IDE driver to reset to the newly changed PCI configuration? Indeed without the HDPARM -X udma5 /dev/hdb command the new PCI registers won't work... Anyways. I learned a lot and now have full speed in the multibay. Thanks!