In seek of a right ICH7 SATA / ATA driver for Windows 9X
My problem is deep rooted and frightening but it is a single problem – Lack of ICH7 SATA driver set for Windows 98SE. I believe that one of software giants has discouraged Intel to further provide Windows 9X drivers for newer chipset. The newer chipset happened to be in my Dell Dimension. With this excessively bitter experience, I may not buy Dell or any newer MS-OS ever again in my life time as longer as Windows 98SE, Solaris and Linux remain strong. In my Dell, Windows 98 is barely surviving without correct driver. USB GEforce and SB Audigy works well because I was able to overcome through cut and paste pci/ven hundreds of time to establish right pci-e root bridge in CPU/motherboard resouce hook.
Windows 9x and Windows NT are built with right and altruistic mind and I can welcome their personalities so to speak and live with them but not with XP or Vista.
I do not want OS itself to get into every operations of computer. OS only should perform launching the programmes and minimal house keeping of memory. I do not expect anything out from it.
I do not want CPU bus bridges (a.k.a., memory controller or multifunction bridge) themselves to get into every operations of computer. Making them emulate sound card, video card and RAID controller are out of question.
My real dream system would pave 10 or so PCI-X slots of 133 or 266MHz bus with at least 18 hardware IRQ allotted. Each card go into these slot has i80303 or better PCI bridge/dedicated IO Controller with embedded CPU with 256 MB on card memory. It means that printer card, mouse card, floppy card, sound card all shoals have own CPU/LPU/MPU and memory subsystem that OS does not even have to know. Give a file name to all such subsystem as memory block.
I did succeed to build a system at leaset a halfway close to this ideal in 2000
Two Xeon CPUs
Intel 80303 on SCSI card 256MB (7 Int13 compatible GDT Logical Volumes formed before OS start)
Motorola 68030 and Emu DSP on Sound Card 64MB
Intel i80960 on Tektronix Printer 64MB
Matrox Millennium 400 (to gain Unix compatibility otherwise GE force GTS is OK)
Another AHA3950 for CD/CDR and scanner
OS supported: Solaris, BSD Unix, Caldera-SCO, Redhat, Suse, O/S2
PC-DOS7. Window 3.1, Windows 95a, Windows 98SE, Windows ME
Windows NT 4.0, Windows 2000, Windows XP 1.0
Supported OS means they can co-exist without modifying metafile structure of legacy form NTFS 1.X. System runs XP and Vista fine but they modifies old Windows NT 4.0 basic data volumes exist in extended partitions if you do not manually pre-mark their EPBR ID from 07 to that of an Unix type such as bf. This invasive personality of Vista and XP 2.0 or 3.0 rules out their qualification of compatibility to any of my system. They can even kill their own sister Windows NT 4.0’s partition.
I am not so much affected by LBA28 LBA32 or LBA48 issues since I never have owned any SCSI disk larger than 280GB (Cheetahs 15.5 Series) nor SATA (Barracuda ES2) larger than 0.954TB per single unit. To install any new OS on SATA platform, I temporarily put an Adaptec 39160 with SCSI drive in channel A and DVD or CD rom on channel B for temporary installation then check if OS and driver correctly mount and recognise file system on SATA. Run series of test from there, then ghost new OS partition from SCSI onto SATA and hide original partitions on SCSI after marking newly ghosted partition active on SATA MBR. Believe or not, it is not too much work. I am used to boot Unix OSes from command line with single or double line strings invoking boot loader. You get used to it. The first, I had to develop a mind to comprehend to different device block mapping expressions pointing out OS kernel location in the system such as c0t0d0p1s:a, /dev/sda1, h0,0,1 or multi(0)disk(0)rdisk(0)partition(1). Then lean primitive commands like map, remap, hide, find, offset, root, boot, kernel and setup.
I never knew anything except SCSI because SCSI was easy and friendly due to being shared bus system with its addressing manually setup by controller slot and jumper pins on target. Of course I was charmed by new generation WD VelociRaptor SATA II disks which boot Windows XP faster than 15000RPM Seagate Cheetah 15.5 series drives. But I trust more familiar SCSI a lot more. SCSI has never given me any headache for past 15 years in any OS arising out from driver issue.
Economic interest was not build into hardware enumeration hierarchy until 486 or even early P5. Pci/ven code started to appear in Windows registry after P5 era. Windows block device mapping has no logical relation to physical device mapping e.g., c0t0d0p0s:b (hba ID, target ID, disk ID, partition ID, slice ID) which makes more sense over Windows device vendor signature based mapping perhaps arising from its economic interests. What a bitter world has it become!
It can possibly mean that if a hardware manufacturer wasn’t nice enough to software giants then the hardware manufacturer can be easily killed by deprived of sales unless the manufacturer was independently powerful as Hitachi, NEC, Panasonic, Pioneer, Sony, Siemens and like. Sadly IBM, Sun and Apple have slightly lost their level of freedom and power comparable to Sony in the past 8 years. Lets give them chances to restore themselves in the next 8 years! Long Live the MSFN to succeed in its unbiased altruistic mission!
Long Live Windows NT and 9X!