Daniel Greenberg
Thanks Nick. I downloaded the ide patches for 2.2.19 from www.kernel.org/pub/linux/kernel/people/hedrick/ide-2.2.19.
After applying the patch, "make xconfig" showed new options under "Block Devices." I selected the VIA82CXXX driver and recompiled and . . . drive performance *declined* by another 2 MB/SEC or so as measured by hdparm -t. (So I saw a drop from around 11.5 MB/SEC to around 9.5 MB/SEC with the via driver. I ran the test 5 or 6 times and took the averages.)
The following is taken from 2xPIII (866 MHz), 694D Pro (MS-6321) ATX mainboard (chipset VIA VT82C694X), IBM-DPTA-372730 HD, customized 2.2.16 kernel (no patches applied), SuSE-7.0. # hdparm -tT /dev/hda Timing buffer-cache reads: 128 MB in 0.94 seconds =136.17 MB/sec Timing buffered disk reads: 64 MB in 2.86 seconds = 22.38 MB/sec # hdparm -v /dev/hda multcount = 0 (off) I/O support = 0 (default 16-bit) unmaskirq = 0 (off) using_dma = 1 (on) keepsettings = 0 (off) nowerr = 0 (off) readonly = 0 (off) readahead = 8 (on) geometry = 3328/255/63, sectors = 53464320, start = 0
I do find myself thinking though that at some point it might be easier to try another motherboard/chipset combination (with the same hardrive) and see what happens to drive performance.
I would definitely do it like this. When I buy a new computer I expect some performance levels while running Linux (HD speed is one of them). If the computer doesn't meet them then I require a different model. Not all computer companies are willing to do it like this but some local assemblers usually are. -- Alexandr.Malusek@imv.liu.se