Rodney Dawn

made the mistake of buying a rx 550 gpu only to discover the drivers are incompatible with windows 7 despite the listing *newegg) saying 7 was supported. Apparently all rx series, and vega series have this problem of no drivers for 7. Had to RMA the card, and am ordering a second r9 390x for crossfire instead. I expected better of AMD. R9 390x is the best AMD gpu that still supports 7. Now I'm wondering if perhaps nvidia has something stronger with full support for 7. If not then I'll just stick with crossfire'd 390's for this fx9350 machine. Seriously, running 7 is getting to be as hard finding compatible hardware as it was running ME (the whole reason I finally gave 7 a try was to use modern hardware). Now the only real point is modern browser TLS support, for the moment at least. Should have figured as much though, 7 is essentially the new Windows ME (or 98se to those who had issues with ME) as in, it's stable, relatively efficient (as windows goes), and power-user friendly (though not as much as 9x). so naturally M$ wants pwople to stop using it, cause they caint make as much money off a stable os. Its all just a stupid moneygrab. ME wasn't broke, and neither is 7, they didn't need replacing. A few service pack updates would be one thing, but 8/8.1 and 10 are the stupidest ideas ms ever made. bloated, spyware ridden, actually built into the core, and just plain garbage. even buggy old 95 is better than 10. 7 (and vista) made sense, as it would have took a complete rewrite of everything to make 9x multicore capable, but everything after 7 is just a waste, (and XP was a resource hog for the hardware of the day. ''most'' reasonably affordable systems that would run XP, would run ME (or 98se) twice as fast. Only once high-clocked multicore chips became the norm rather than the expensive, (and by then Vista was out) was there a practical need for a home user NT based system. (low clocked ''cheap'' multicores were better suited to low end NT workstations than for home users, as even on a multicore capable os, most games and DVD video playback) wersingle threaded-meaning ghz was more important, and a fast AthlonXP or Prescott P4 on ME would easily outrun a early A64 or core2 on XP in those areas, as none of the multicore chips of the time could match the ghz needed for single thread programs like that. (I dont count the prescott's hyperthreading capability as multicore, as it wasn't a true dual-core chip, but even a a prescott making use of hyperthreading on XP was still held back by the lack of apps which benefit from it, and the added weight of the os. (9x being lighter, was still faster except in rare apps which were meant to take advantage of hyperthreading on NT systems). It wasn't until mid vista/early win7 beta era that multithreading really took off, and even then a lot of the apps were only optimized for two cores. Thats actually the one thing that could have made 7 even better, is if they had included a built-in 3d-accelerated VM where you could switch to "Windows ME mode" for compatibility with 9x apps (at least Radeon 9xxx or FX6800 level graphics support) and maybe even 95c/Rage Pro 3d mode as well for really picky dos apps. (the VM creates a ramdrive in the lower 32 bit address space, passes core 0 through to the 9x os directly, core 1 emulates the graphics chip, core 2 emulates any peripherals that need older versions instead of just passed through, (like Soundblaster 16 instead of passing through ac97) while others like usb2 could simply be passed through (ME) or emulated as 1.1 for 95c. and core3 handles emulating floppy support, and either emulating a dvd drive if one is not on the host machine, or passing through (translating SATA data into ATA133 in the process). if one is present. Such an emulator being integrated into 7 would be the only thing that would make it better. (Aside from manufacturers writing drivers for modern hardware for 7 that is.). apologies for rambling, it just really ticks me off to have paid so much for a graphics card, only to have to send it back, but such is the case, as I will never ''downgrade'' to 10.

AMD 785g, Onboard LAN working, in windows ME. USB2.0 ports, PS/2 ports, pci and pcie slots. onboard SATA kinda works (in IDE mode it runs in msdos compatibility mode as a generic storage controller) fix with promise ATA133 pci card and PATA IDE ATA133 SSD. onboard video is sufficient for setup (16 color 640x480) (Fix with X300 series pcie or PCi 7000) onboard audio is hdac97, so fix with soundblaster pci. (DO NOT connect cdrom to the ATA133 PCI card, except during initial install. it slows the card terribly and will horribly bottleneck the SSD.) Complete the install then disconnect cdrom, connect a wireless AC to gigabit ethernet bridge to the LAN port with a length of cat6, and install all updates and hotfixes, then install a usb DVD drive to have a working optical drive. Disable onboard gpu and AC97 hdaudio in bios to free up IRQ's, make sure boot mode is set to legacy BIOS only, and install only a single 1gb stick (or 512m if anyone makes one) of ddr3. use an OPTERON CPU. A FX series chip might work but have not tested that. I only tested this with an Opteron 3380 as I have water cooling and it is a real pain to take it all off and unmount the mb just to swap cpu's. Limit memory with setup switches as usual, then edit system INI appropriately once install is complete. make sure all usb controllers are set to legacy usb support, and 2,0 mode with 1.1 compatibility fallback enabled. use a DVI converter to get HDMI out from the card's DVI port. (video only no audio) and connect external speakers to the soundblaster PCI card. what doesnt work: USB3.0, HDac97 audio, onboard Radeon HD8xxx graphics (except basic 16 color mode) R9 390x (replaced it with PCIe x300 series) SATAIII (set to IDE mode it sort of works, generic IDE controller in msdos compatibility mode). Cool and Quiet. (Use K8 powernow driver to get it working). and of course, only one core is used.

Thanks I would have thought it would take more. Even factoring in disk compression it would still take approximately 290 disks for a ''full'' inastall of ME. (assuming a conservative estimate 2.3mb compressed into each 1.44m disk.)

Another thing that would be interesting to try would be extracting all the files form the setup CD, and creating a spanned volume floppy install for ''really'' ancinet hardware that had no cd drive. Loading 600 floppies would hardly be ''fun'' but it would be neat just to see if it was doable. If possible, you could even create similar spanned volumes on a mere six zip100 disks, or 20 30mb tapes. just interesting ideas to toy with getting ME installed on something like say, a 286, just for fun. It would be a way of seeing the Absolute minimum hardawre you could install it on. Oc course it might be easier to hack something together using a usb CDrom (oldest avalible for compatibility) and a dev board to create a functional usb to serial, paralell, or IDE conversion board. The idea being for machines that the BIOS doesnt know what a cdrom is, so you ''trick'' the BIOS into treating the cd drive as if it were a hsrddrive.

Yeah its one of those oddities, likely from an 8086 machine from the age of it. I have a few IDE and SCSI tape drives, and even one of the floppy-controller driven oddballs (30 meg). I have tapes for most of them but only managed to get one of the IDE models working in windows. Yeah I've seen that early form of ''virus protection'' on a lot of older boards. There should be an option to disable it in BIOS, but undoubtedly some OEM's omitted that function. I've never played Red Baron, but its a shame to lose anything like that, especially difficult to replace software that hasn't been made in ages. Of course having suffered through windows 95's (lack of) stability when running any ''demamding'' apps, running them on ME is truly a blessing. Back in the day I gave up on quake and Doom cause of w95 crashes, and didn't complete them until years later when I discovered they ran beautifly on Win ME, and never had a single crash to desktop. Sure, its not perfect. every os has its flaws, but when properly installed, on sensible hardware (avoid SX chips,in 386 and 486, and celeron/duron/sempron as well, and shoddy hardware in general) it can be even more stable than win 7. As for the game, if your drive supports double-density 5.25 disks, you might be able to fit it on one of those. ME will read them properly, unlike a certain other 9x OS. I've also compressed and subsequently read many 5.25 and 3.5 disks, and even zip disks without any trouble. One thing I have yet to try is assuming the 8" will hold the data for a bootdisk, booting from it. the 486 supports booting through LPT1 and will treat the zip drive as a big floppy for that purpose. I made a windows bootdisk from a zip 100 disk, and loaded it with a few extra dos utilities like memtest for mantinence work. I havent tried booting windows from zip on the 486, as I couldnt get it to recognize an IDE zip750 drive, but I did successfully boot into windows (copied a fresh install from C: onto a zip750 after making the disk bootable by first using it to create a startup disk) under DOS from a startup disk. then disconnected the hdd and let it boot from the zip. This was on an Athlon 1ghz machine, and the slow zip access time took forever to boot, but it did work. sort of a dirty hack way of making a ''portable'' windows ME. No real practical application as it would take forever detecting and installing hardware if booted on any machine other than the original athlon (due to zip speed limitations) which defeats the purpose of portable windows but berhaps someone could adapt the idea to boot from something faster like a modern flashdrive of usb SSD to make it more useful. (there'd still be the issue of hardware detection/reboots, but it wouldnt take forever). I could see that being very useful for testing new motherboard/cpu combinations for 9x compatibility before doing a proper format and install in the normal manner.

Thats what I was afraid of. On the plus side, I already have a soundcard with IDE installed which I had been using for the cdrom connection. Just to test I'll have to swap the cable and try that. Its PIO mode 3 I think. I'll have to double-check though. I'm pretty sure the max drive size it supports is 1.3gb but it runs a slightly newer 4x cdrom just fine (averaging between 2x and 3x speed). I have others to try as well, but I would rather not replace the soundcard if its avoidable. Its one of the better ones, with a lot of memory chips on it. No luck by changing the LPT1 settings in BIOS. The drives behave the same, only working from the DOS startup disk, but disabling bi-directional mode made my zip drive stop working. (I tested it for compatibility reasons). Disabling bi-directional data did get the 8" floppy working however, and the one good disk that I have appears to be some sort of early DOS program as it has .bat files on it. At first I thought it was making Explorer hang again, as attempting to access the drive made the system momentarily unresponsive, but the drive light was on and making access noises so I let it sit a bit and after a minute or so it finally opened an explorer window containing the disk contents. It works, but the old 8" access time makes even the 5.25 look fast by comparison. I copied the disk contents to a 3.5 floppy to explore further without stressing the aged 8" disk, while trying to determine what sort of program it contained.

I'm thinking its an I/O error, or driver timing issue. I'll have to try setting the port to non bi-direcctional and try that. Might have to try dropping the fsb as well, although I believe the paralell port, being an extension of the ISA bus, should be clock locked to 8.333 mhz on this board. I can try 25x4 to maintain the chip stock freq of 100, and see if that changes anything. What makes me suspect a driver timing issue, or missing/incorrect driver is they all seem to work fine being accessed from a DOS startup disk. I belive windows is trying to treat them as floppy drives, and then the driver gets ''confused'' by the capacity. Perhaps forcing the drives to run in ''msdos compatibility mode'' would fix the issue, if changing the port to non bi directional, and fiddling with the busclock fails. Unfortunately I don't currently have a drive with windows 95 on it to hook up as a slave, and point the new hardware wizard at to see if it finds a more suitable driver, so I can't easily try that. (probably would not be a very stable solution even if it worked though). I may have an old drive with 98 (original non SE) on it around here somewhere to try, but its unlikely as I have always upgraded every machine I get to ME as soon as possible. Its definately a paralell port connection, not 25-pin serial. The machine does have a 25 pin serial port, which used to connect to an old dial-up modem but is currently unoccupied. This computer ''can'' have a SCSI port too, as I have two free ISA slots and a handful of old SCSI and MFM ISA cards I could try to see which have drivers for ME. THe USB to IDE adapter idea would likely work well, except the 486 has no usb. If such a thing existed, I would need either an ISA, VLB, or rs/232 to usb add-on in order to try that on the 486. I doubt they make any of those as I have never seen one, of course. THe ATA6 controller to support larger drives would be great, but I am hindered by the lack of PCI. The only bus with enough bandwidth for such a controller would be the VLB, assuming anyone made an ATA6 VLB card, that is.

I can write to the old drives in DOS fine. I tested them afoter formatting, by using the bootdisk to ''start computer with cd rom support'' and then copying a 50 meg .zip archive from a cd by changing to cd directory and then "copy test.zip F:\test

Yeah, the Mach64 was one of the better graphics cards for the 486, and even came standard as integrated graphics on some socket 7 boards (no ''real'' VLB slot, but had the VESA bus internally connected as part of the chipset, essentially emulating the 486 memory subsystem as part of the northbridge on those boards.) This was an early attempt at decent integrated graphics which was later superceded by the Rage II which internally connected to the PCI bus on later socket 7 and eventually Rage Pro which internally used a 1x agp bus on Super 7 boards. Good to know its just a bios limitation, as I've got some sizeable PATA drives that would work great with ATA133 controllers (on machines with PCI slots) Unfortunately the fastest bus on the 486 machine is the VLB, and the best card I have found is an ATA33 controller that supports a max of 32gb. I got the Trident working in 256 color 800x600 by installing the vram from the broken card, and put the ATA33 card in the VLB, the 3.2 caviar SE is working perfectly with two partitions of roughly 1.5gb (minus a bit for formatted capacity) and is noticeably faster than using the old PIO controller. Yeah the 3.2 is a replacement for the original 400 meg drive as I needed more room to install ME to begin with. the original drive actually formatted fine, but was just under the minimum size (583mb according to setup) by a couple hundred mb so I had substituted the wd and configured with jumpers to limit capacity to 850 megs for the old pio controller (no longer necessary with the VESA ata33 card. The 3.2 was pulled from an old socket 7 machine, and fully supports ATA33 speeds. I may end up swapping it for a 10 gig MAXTOR ATA66 drive (which would downclock to ata33 of course, but because of the larger cache the Maxtor gives slightly better throughput (about 450kb/s better read/write) when downclocked to ata33 than what the wd (a true ata33 drive) gets. The old paralell port drives are mostly MFM, with a couple early PIO models (the larger ones). The trouble I am having, is using a bootdisk, I can get DOS to see them, and format with FAT, but in windows they either dont show at all or just show as a generic ''removable disk'' which locks up Explorer when I try to open it. I think its likely a driver issue, and need to find a modded driver from win95 tweaked to work with ME, assuming such is possible. I know it can be done with printer/scanner drivers, also using LPT1 so theoretically it should be possible with HDDs as well. I assume these were probably win3.11 or early win95 drives, and maybe even older than that. I would think they would work using ''msdos compatibility mode'' since the DOS bootdisk sees and can write to them, same for the 8" floppy (I have 1 good disk for it, and one which is magnetically degraded to the point of being unusable), but for whatever reason windows doesn't. LPT1 is set to allow bi-directional data of course, and my paralell port zip drive works flawlessly. Thanks for the tip about the orientation of the floppy cable, now I have both the 3.5 and 5.25 working as they should. That's what I was afraid of with the serial ps/2 adapter. I'll just stick with the serial mouse for now, though I miss the scroll wheel, and the ball is nowhere near as precise as the ps2 optical. Eventually I hope to find an ISA card with PS/2 connectors to solve the problem properly.

Yes, it has several empty sockets. I've also got an identical card with a broken D-sub (VGA) connector which I could pull chips from to populate the empty sockets on the working Trident card. Hopefully that will bring it up to either 640/480 16 bit, or 800x600 256 color, either of which would be a nice improvement. Then I could use the VLB slot for a nice (for its time) Ultra ATA33 card (the HDD is a WD32000 caviar SE 3.2gig capable of running ultra ATA33 mode so would definately benefit from running at ATA33 rather than PIO mode. (and could remove the jumper limiting capacity to 850 megs (needed by the motherboard PIO controller) and take advantage of the full drive. the VLB ATA33 controller supports drives up to 32 gb) and of course windows ME can theoretically handle up to 137 (or more with patches) so the full 3.2 being usbale would be no problem. Going to use partitionmagic to resize the C: partition to 1.5 gb, then partition the remaining space to hold the swapfile, and expanded data storage, assuming I get the trident ISA gpu sorted out so I can free up the VLB slot to run the ATA33 controller, that is. Probably going to have to play with the jumper settings to get the ATA controller ''happy'' though. while the Mach64 gpu runs fine with 40mhz FSB/VLB the HDD controller likely needs the busclock at 33mhz (which is standard for the chip anyway, 33x3. but I was running it at 40x2.5 (still 100mhz cpu clock) for the improved system throughput of 40mhz fsb. 50x2 is also workable, as I have really good EDO dram, but not practical as even the MAch64 VLB card has major issues with the VLB running at 50mhz. I dont never run the chip above rated 100mhz, but I have tried many combinations of fsb and multi to get to 100 in order to optimize throughput while maintaining full function of all components. The chip originally had no real cooling, so I used some thermal adhesive, and a small northbridge cooler off a junk p4 board (bad caps and toasted mosfets) and a small chipset fan with 3-pin to FDD power addapter (tapped for 5v so it stays silent) just to help keep it cool under the heavy (for a 486) workload of running windows ME. Granted even this simple mild active cooling is overkill for a 486, but it was free, runs quiet, and will no doubt prolong the life of the system. ON the software aspect, the old 486 has only had two real issues with ME. One being it doesnt like to use the 5.25 and 3.5 floppy together (backing up old 5.25's to 3.5's) or reading from either floppy while writing to the paralell port attached zip drive. (likely a bug in the old floppy controller and not windows itself) This happens regardless of busclock. The other is a sound glitch, where if I use a regular serial mouse, the startup and shutdown sound plays properly, but if I use a serial to ps/2 adapter to use my optical mouse it sometimes causes crackling in the audio. I'm thinking this is due to the adapter itself, as the same optical mouse in a standard ps/2 port doesn't cause this crackling on K6, pentium, or newer systems also running ME. Speaking of paralell port drives, I also have several tape drives, a good working 8" floppy, and a few hdds in the 80 to 130 meg range which I have been trying to get working with ME.

Thanks That will certainly make things easier.

Yeah I just gave it a try, and it does indeed run on ME if you replace Psapi. I made the mistake of running the machine out of resources trying to find all calling modules, but it does indeed work. For reference, the equivalent directory to C:/winNt, on ME (and 98/se) on a default install would Be C:/Windows

opteron updae, still looking for a simple way to update win7 bootloader to recognize ME (installed on a DOM SSD in a PCI to ATA133 add in card) AM3+ Opteron 3380 if that helps. Primary GPU R9 390x PCIe, secondary GPU ATI 7000 PCI bios set to display first pcie, then pci then onboard (radeon HD6200 series) legacy BIOS boot mode, SATA set to IDE mode. I can boot either OS by selecting boot drive in bios, but would be simpler if someone could help me edit win7 boot file to recognize ME (D:/windows)

was actually wondering about purchasing that package myself. got a nice quad opteron I'd like to be able to use at least two cores of for some apps in ME.

Sorry I was unable to relpy sooner. RLoew, You have done a great service keeping win ME(9X) alive with your ram patch, and various other programs. I think I speak for the community in general when I say, R.I.P, and your work will be sorely missed. Thank you for all your help, and rest well wherever you are now.