Word 97 works. Its x86 though. If we weren't stuck at SP2, there'd be tons of more useful things to install.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
NT on PowerPC Macs!
- Thread starter Rairii
- Start date
-
- Tags
- windows nt
- Sort by reaction score
All non-x86 versions of Windows (other than Windows RT) can run Intel x86 apps in some fashion. Most of the NT 3.5-4.0 ports can only run 16-bit (Windows 3.1) apps, but the DEC Alpha versions could run 32-bit. (The later Itanium and mainstream-Windows-on-ARM-64-bit versions could run 32-bit Intel apps out of the box.) As mentioned above, only Windows RT (stripped down Windows 8 on ARM 32-bit) couldn't run any Intel apps.
One PPC-NT-selling vendor (Motorola, IIRC) released their own 32-bit-Intel emulation layer that installed on NT in addition to the built-in 16-bit; and people recently re-found it and figured out how to install it on non-Motorola computers. So yes, Alpha and PPC versions of NT can run Intel apps just fine.
Heck, long before people rediscovered the Motorola 32-bit emulator, I ran Windows 3.1 versions of Intel apps on my PowerPC ThinkPad. (IE 3 was the latest browser ever released PPC-native, but IE5 was released for Win3.1, so I run the Intel Win3.1 IE5 as my web browser. As well as MS Office for Win3.1.)
That's really cool. Thank you guys. Would have been sweet to run NT on G3s and G4s back in the day.
Knowing what I know now, I'm surprised there wasn't something similar to VMware that let you run NT in a window. We had Virtual PC but it was painstakingly slow, and I suspect that running Windows PPC would have been quite a bit faster since you'd only have been emulating x86 user-mode apps, whereas the OS itself and all its APIs would have still run natively.
It's because PowerPC for Windows was very niche. Combine that with the niche of Macs, it didn't make sense.
Well. You could then run NT PPC in a virtual machine but what then? Where were the apps that would have prompted this effort? If the intention was to increase the use of NT PPC then that software would have had to come from Microsoft at least initally and they would rather you switched to Windows x86.
To clarify, I didn't necessarily mean a commercial product (with its associated cost-benefit justification etc.) but even just a proof of concept from an enthusiast. There were all sorts of wacky apps back then with varying amounts of polish, so I would've thought someone would have given it a go.
I do recall that the topic of booting NT natively would come up occasionally, but the answer at the time was "no", and that's as far as the conversation usually went.
I do recall that the topic of booting NT natively would come up occasionally, but the answer at the time was "no", and that's as far as the conversation usually went.
I see what you mean. Back then Microsoft was a bit of a dirty word so Apple users either shunned it completely or got their MS fix via VPC, SoftWindows etc. There wasn't much call for running it full time on very expensive hardware when you could do it better on a commodity PC box, even for the geekery of it. OS/2 for PPC didn't take off either and never made it to a golden master, let alone retail release, despite IBM's prior readying of a limited range of hardware for it to run on.
wx86 is also an alternative for running 32-bit x86 apps "natively" instead of Motorola Softwindows, it's been linked to earlier in the thread (previous page). AFAIK it was offered by Microsoft, for PPC, MIPS and DEC Alpha, for a very short period of time.
I didn't compare wx86 and Moto SW side-by-side yet, but some things might run on only one or the other, although almost everything should be able to run on either, as long as everything else is compatible with NT4 SP2 (or other installed components, like exact DirectX version and whatnot).
@Hakilo I used it on the Wii because that's what I have at hand, but a Wii U should be better, or one of the compatible G4 Macs. For the latter, you must be able to use ADB directly somehow, as we currently lack USB drivers for mouse and keyboard support. The Wii and the Wii U can get around the USB requirement for mouse and keyboard via some form of leveraging their support for these before booting into the NT OS, but you still have to watch out not to have them lose connection, as there's no plug'n'play for them.
i must've missed the wx86 comment. I found it though, and now I'm curious to compare the two. Might be my next project for this weekend.
G3, for sure. Particularly the old ones, and particularly particularly a PowerBook or iBook.
If you want to use an emulator, qemu-ppc would have probably done you better.
Wii U is just running it in Wii mode, I tried it. If you're looking for raw speed, you might want to look into a Titanium G4 with the Mac99 firmware build.
I wasn't sure about speed, but apparently @Rairii got to use many times more RAM on the Wii U. Although, again, I don't know if that's the case for the latest release, or just in some unpublished version of the project. There's also the prospect that apparently multi-core support (3 cores) on the Wii U is a possibility: @Rairii shared some progress on his social media platform of choice, in which it seems NT4 identified all 3 CPUs (cores, actually) correctly, although it seems like tasks are all getting done by just one of them. So... I think right now Wii and Wii U are on similar footing, but possibly not forever.
Getting out of vWii mode would be extra awesome, too, if it reaches that point someday.
Here's what I was able to dig out of the website (including some other unrelated NT4 PPC things, I think in reverse chronological order, at least for the most part), sharing for our own reference:
[question] Make Wii U cores act like 3 separate cpus · Issue #56 · Wack0/entii-for-workcubes
This idea comes to me thinking about how Pentium D works but instead of the cpu make the OS think there are 3 cpus, i know Wii u versión, it's using vwii but in the future using a loader could be p...
github.com
Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
the beefiest powerpc nt system in the west?still plenty of things to figure out:after setup, i get bugchecks in cdfs.sys specifically when winlogon/etc comes up, so far only renaming it out the way...
labyrinth.zone
Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
I ported my dxdiag patchset to dxdiag 8.0 because it was the last version to support testing the d3d software renderer It works on the wii! but at ~5-6fps, which if I remember correctly was basica...
labyrinth.zone
Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
and here's a video of the direct3d test running on my wii
labyrinth.zone
Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
@domi all in software, the kernel (win32k) support for d3d hw acceleration didn't get backported from (at that point still in development) win2k until sp3, and PPC never got sp3
labyrinth.zone
Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
oh, that's why my ddraw driver doesn't want to work (on wii) the gdi32!GdiEntry2 thunk in wx86 is present (somehow, given a whole lot of "undocumented" exports in, for example, ntdll, aren't thunk...
labyrinth.zone
Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
lol, I just made 256MB of DDR visible in vwii by a single memory controller register write and yes, there's cafe os stuff in the upper 192MB who needs tweezers?
labyrinth.zone
Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
on the nt bootscreen the kernel only shows the amount of memory it can actually use so, on wii, it subtracts 11MB of RAM from the amount displayed on the bootscreen so, on vwii with extra ram, it...
labyrinth.zone
Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
days since i was forced to agree to the GPL: ...wait a minute
labyrinth.zone
Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
@jmcwhatever hi, i looked quickly at the netbsd/wii init code, seems it's always assuming there's 64MB of DDR only, which may be incorrect - if MEM_RANKSEL is 1 then it's running on a devkit with ...
labyrinth.zone
deneb (@deneb@wetdry.world)
Attached: 1 image @Rairii@labyrinth.zone's port of windows nt 4.0 to the gamecube on live display at the @realraum@chaos.social kiosk #glt25
wetdry.world
Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
does espresso help with longer uptimes? how about latte?
labyrinth.zone
Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
by the way, here's a patched up vwii ios booting in wii u mode with debug output on the gamepad the patches needed: - fix the timing code for the faster clock - hook the hardware cache flush/inval...
labyrinth.zone
Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
getting this far felt like playing minecraft survival multiplayer at times
labyrinth.zone
By the way, I found something very interesting. (Sorry for the double-post, but it's worth it.)
It turns out both Windows NT 3.51 SP5 and 4.0 SP2 had, respectively, "Post-SP5" and "Post-SP2" updates. And guess what, both of these include a few PowerPC updates!
NT 3.51 Post-SP5 Updates were easily and conveniently archived at WinWorldPC, look for "Windows NT 3.51 Post-SP5 Hotfixes".
NT 4.0 Post-SP2 Updates were not as comfy nor as easy to locate and get, but nothing undoable: there was a complete file list here in which the files were not downloadable, and another one here in which they are. I guess I could have used curl or whatever, but instead I painstakingly downloaded every single file and created each folder manually, without skipping anything. I have attached a .ZIP of that effort here in this post for convenience.
Notice that you can check out the parent directories of these sites for more such fixes for other service packs if desirable. I found nothing in native PowerPC on Post-SP3+ hotfixes, as expected, but they are there, as well. Of note might be NT 4.0 Post-SP1 fixes in case we ever get MIPS NT 4.0 on hardware like the N64, PS2, PSP, PSV (PSP CPU is included) and PS3 (PS2 CPU is included in early models or, if not, an emulator thereof). These all share the MIPS III CPU architecture which is targeted by NT 3.x and 4.x, which AFAIK is an architecture that is both 32-bit and 64-bit compatible.
Imagine a PS2 and a Wii networking with each other via their respective native Windows NT 4.0 SP1/SP2 OSes someday...
@Rairii has at the very least ported most of his ARC firmware to MIPS almost just now, with which he got somewhere on his MIPS-based laptop (chronological links below):
labyrinth.zone
labyrinth.zone
labyrinth.zone
labyrinth.zone
I'm way more excited about THIS, though, but that one certainly deserves its own topic!!
It turns out both Windows NT 3.51 SP5 and 4.0 SP2 had, respectively, "Post-SP5" and "Post-SP2" updates. And guess what, both of these include a few PowerPC updates!
NT 3.51 Post-SP5 Updates were easily and conveniently archived at WinWorldPC, look for "Windows NT 3.51 Post-SP5 Hotfixes".
NT 4.0 Post-SP2 Updates were not as comfy nor as easy to locate and get, but nothing undoable: there was a complete file list here in which the files were not downloadable, and another one here in which they are. I guess I could have used curl or whatever, but instead I painstakingly downloaded every single file and created each folder manually, without skipping anything. I have attached a .ZIP of that effort here in this post for convenience.
Notice that you can check out the parent directories of these sites for more such fixes for other service packs if desirable. I found nothing in native PowerPC on Post-SP3+ hotfixes, as expected, but they are there, as well. Of note might be NT 4.0 Post-SP1 fixes in case we ever get MIPS NT 4.0 on hardware like the N64, PS2, PSP, PSV (PSP CPU is included) and PS3 (PS2 CPU is included in early models or, if not, an emulator thereof). These all share the MIPS III CPU architecture which is targeted by NT 3.x and 4.x, which AFAIK is an architecture that is both 32-bit and 64-bit compatible.
Imagine a PS2 and a Wii networking with each other via their respective native Windows NT 4.0 SP1/SP2 OSes someday...
@Rairii has at the very least ported most of his ARC firmware to MIPS almost just now, with which he got somewhere on his MIPS-based laptop (chronological links below):
Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
so I was just informed about these things and I put in an offer for one on ebay https://en.wikipedia.org/wiki/Skytone_Alpha-400 MIPS laptop circa 2008? yes please
labyrinth.zone
Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
the MIPS laptop arrived, can i call it a bnuuy because of the SM64 rabbit named after the architecture
labyrinth.zone
Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
arc firmware codebase intended to be for powerpc, now starting to run on mips
labyrinth.zone
Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
MIPS NT on Skytone Alpha-400 turns out to be harder than previously thought, due to Ingenic JZ4730 lacking a MIPS FPU, and MIPS NT's RtlFillMemory and RtlMoveMemory using the FPU (because R4000 was...
labyrinth.zone
I'm way more excited about THIS, though, but that one certainly deserves its own topic!!
Last edited:
I stumbled upon this PPC NT app completely by accident...
Dependency Walker
(Scroll down for PPC version. A version is also available for MIPS and both 32 and 64-bit DEC Alpha. And IA-64, as well as x86 32 and 64-bit as usual. No ARM or Windows-CE-only CPUs, though?)
Very famous app among developers apparently, particularly good at figuring out what other files/methods a program depends on. Seems to be commonly used alongside CFF Explorer (although that is x86-only AFAIK, but maybe this can work on NT4 PPC with Moto SW and/or wx86?).
Dependency Walker
(Scroll down for PPC version. A version is also available for MIPS and both 32 and 64-bit DEC Alpha. And IA-64, as well as x86 32 and 64-bit as usual. No ARM or Windows-CE-only CPUs, though?)
Very famous app among developers apparently, particularly good at figuring out what other files/methods a program depends on. Seems to be commonly used alongside CFF Explorer (although that is x86-only AFAIK, but maybe this can work on NT4 PPC with Moto SW and/or wx86?).
I noticed the NT4.0 Post-SP2 Hotfixes / updates use some archive format that causes WBTD (Windows Binary Type Detector) to report:
But it turns out there's nothing wrong with the binaries: my app was just overly-cautious about the reporting, and in this situation it should report it exactly as what it is:
A 16-bit x86 DOS app like any other.
In this case, the ".EXE"s are just self-extracting compressed files, by "PKLITE Copr." or "PKWARE Inc.", based on the binary data.
So essentially the bug is that my tool misreports an "error" in this particular situation when there's no error. By the point the analysis reaches this part of the program, if it "errors" there, it's actually just that it's just one of the many formats found for DOS apps.
To further confirm this, you may notice that if you try to run them, even the Intel version, on a 64-bit version of Windows, it will refuse to (if your system is stock), as 64-bit systems are barred from running 16-bit apps. Yet the executable runs under 32-bit and 16-bit Windows, confirming its 16-bit nature.
I fixed the messaging in my tool for this specific situation, never encountered this problem with all the binaries I had tested so far. This was a good test case to catch and address!
WBTD 1.1 is attached as a ZIP to this post, like my original release in the previous page. With the same binary, it reports:
Of course, it is distributed as I did with the original release: a PPC-native version compiled in PPC VC++ 4.1 on a Wii, a 32-bit x86 version compiled in VS2022 on Win7, the original source code as a single ".c" file, plus screenshots and relevant text documents.
But it turns out there's nothing wrong with the binaries: my app was just overly-cautious about the reporting, and in this situation it should report it exactly as what it is:
A 16-bit x86 DOS app like any other.
In this case, the ".EXE"s are just self-extracting compressed files, by "PKLITE Copr." or "PKWARE Inc.", based on the binary data.
So essentially the bug is that my tool misreports an "error" in this particular situation when there's no error. By the point the analysis reaches this part of the program, if it "errors" there, it's actually just that it's just one of the many formats found for DOS apps.
To further confirm this, you may notice that if you try to run them, even the Intel version, on a 64-bit version of Windows, it will refuse to (if your system is stock), as 64-bit systems are barred from running 16-bit apps. Yet the executable runs under 32-bit and 16-bit Windows, confirming its 16-bit nature.
I fixed the messaging in my tool for this specific situation, never encountered this problem with all the binaries I had tested so far. This was a good test case to catch and address!
WBTD 1.1 is attached as a ZIP to this post, like my original release in the previous page. With the same binary, it reports:
Of course, it is distributed as I did with the original release: a PPC-native version compiled in PPC VC++ 4.1 on a Wii, a 32-bit x86 version compiled in VS2022 on Win7, the original source code as a single ".c" file, plus screenshots and relevant text documents.
Should be straightforward enough with the
Here is a little bit of documentation https://wiki.qemu.org/Documentation/Platforms/PowerPC#IBM_RS/6000_7020_(40p)
I'll take a look at it deeper later.
If I had to guess, nobody's really tried this because NT4 x86 exists and is superior anyway in virtual environments.
40p machine type.Here is a little bit of documentation https://wiki.qemu.org/Documentation/Platforms/PowerPC#IBM_RS/6000_7020_(40p)
I'll take a look at it deeper later.
If I had to guess, nobody's really tried this because NT4 x86 exists and is superior anyway in virtual environments.
The smp implementation on the wiiu is actually kind of broken, so getting it working could be a bit of a challenge. There's an issue on the wiiu linux git that explains it better than I could, but the assumption is that official nintendo sdk's worked around the issue- https://gitlab.com/linux-wiiu/linux-wiiu/-/issues/20I wasn't sure about speed, but apparently @Rairii got to use many times more RAM on the Wii U. Although, again, I don't know if that's the case for the latest release, or just in some unpublished version of the project. There's also the prospect that apparently multi-core support (3 cores) on the Wii U is a possibility: @Rairii shared some progress on his social media platform of choice, in which it seems NT4 identified all 3 CPUs (cores, actually) correctly, although it seems like tasks are all getting done by just one of them. So... I think right now Wii and Wii U are on similar footing, but possibly not forever.
Getting out of vWii mode would be extra awesome, too, if it reaches that point someday.
Here's what I was able to dig out of the website (including some other unrelated NT4 PPC things, I think in reverse chronological order, at least for the most part), sharing for our own reference:
[question] Make Wii U cores act like 3 separate cpus · Issue #56 · Wack0/entii-for-workcubes
This idea comes to me thinking about how Pentium D works but instead of the cpu make the OS think there are 3 cpus, i know Wii u versión, it's using vwii but in the future using a loader could be p...Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
the beefiest powerpc nt system in the west?still plenty of things to figure out:after setup, i get bugchecks in cdfs.sys specifically when winlogon/etc comes up, so far only renaming it out the way...Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
I ported my dxdiag patchset to dxdiag 8.0 because it was the last version to support testing the d3d software renderer It works on the wii! but at ~5-6fps, which if I remember correctly was basica...Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
and here's a video of the direct3d test running on my wiiRairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
@domi all in software, the kernel (win32k) support for d3d hw acceleration didn't get backported from (at that point still in development) win2k until sp3, and PPC never got sp3Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
oh, that's why my ddraw driver doesn't want to work (on wii) the gdi32!GdiEntry2 thunk in wx86 is present (somehow, given a whole lot of "undocumented" exports in, for example, ntdll, aren't thunk...Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
lol, I just made 256MB of DDR visible in vwii by a single memory controller register write and yes, there's cafe os stuff in the upper 192MB who needs tweezers?Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
on the nt bootscreen the kernel only shows the amount of memory it can actually use so, on wii, it subtracts 11MB of RAM from the amount displayed on the bootscreen so, on vwii with extra ram, it...Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
days since i was forced to agree to the GPL: ...wait a minuteRairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
@jmcwhatever hi, i looked quickly at the netbsd/wii init code, seems it's always assuming there's 64MB of DDR only, which may be incorrect - if MEM_RANKSEL is 1 then it's running on a devkit with ...deneb (@deneb@wetdry.world)
Attached: 1 image @Rairii@labyrinth.zone's port of windows nt 4.0 to the gamecube on live display at the @realraum@chaos.social kiosk #glt25Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
does espresso help with longer uptimes? how about latte?Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
by the way, here's a patched up vwii ios booting in wii u mode with debug output on the gamepad the patches needed: - fix the timing code for the faster clock - hook the hardware cache flush/inval...Rairii :win3_progman: :win3: (@Rairii@labyrinth.zone)
getting this far felt like playing minecraft survival multiplayer at times
It's a shame, the WiiU is the easiest multi-core PowerPC to buy right now. Although, of course, a single core is enough
Can anyone tell me about the possibility of porting to Xbox 360? There's also a PowerPC, and I think there were homebrew versions.
Although, of course, the PowerPC G5 would be a good choice