[Raion]

On the IRC we have had a MAME dev working on Iris Indigo and Indy emulation. While a cycle exact replica in software is a long way off (it can't even boot to an installer or preinstalled disk) he has made progress according to him with getting it into a more usable state. 

There has been as of recent a lot of discussion about this on Reddit and in the Discord and I'd like to step up and say my piece, which can be summed up as:

Emulation of Silicon Graphics hardware is unlikely to revitalize our community any more than traditional efforts. As these are not video game computers, and the platforms being currently targeted are very old instances I do not see it as particularly useful now or in the near future as so much of the higher-end systems is undocumented and unlikely to be replicated in the near future. 

This is why I don't have emulation discussion categories. There's no point in sitting around talking about what is essentially vaporware until it brings something usable to the table. I'm not trying to tread on anybody's efforts I just have reservations and doubts. I don't think it'll bring useful people into the community and I think that emulation is not going to offer very much at the end of the day.

We currently operate on a walled garden model of sorts because you have to have relatively high investment in order to be part of the hobby, and this really keeps a lot of the more obnoxious and annoying people from sticking around. Having functional emulation will bring challenges and it won't be like suddenly we have a superior solution either. 

This is why I don't fund development of emulators I'm more interested in seeing our software transcend hardware in a more haiku type model. Don't get me wrong I love my hardware that I have spent a lot of money on, but IRIX as a free open source OS on modern hardware would do far more for the community than any emulation.

feel free to discuss below based on what I say or what you have to say on the matter I could be completely wrong and be proven wrong in the near future about this and I accept that. Regardless of how I personally feel about emulation once we have some sort of reference implementation that works to a functional degree I will probably end up adding emulation categories.

[Irinikus]

I tend to agree with you.

For me, this hobby is about running the IRIX OS, as well as the applications of the day on the intended Silicon Graphics hardware. The concept of emulation doesn’t inspire me at all.

It will eventually become more and more difficult to keep our machines running, but that’s what the hobby’s about isn’t it? (That’s what the community’s for.)

[CiaoTime]

I fundamentally disagree.

I have a vested interest in seeing the community surrounding MIPS systems grow - both as a hobbyist who personally uses the software and as someone who sells and supports the hardware involved in it. From a software perspective, there are talented coders and interested people out there who simply don't have access to SGI gear, be it for financial or even geographical reasons. If I was to give away every piece of working hardware I had to people who wanted to take a crack at doing work on IRIX, I'd still only be able to add maybe ten or fifteen people to the 'walled garden' that you speak of. Even if an emulator only compels a couple hundred people to fire it up, that's still over an order of magnitude more than I can ever do to get people to experience the system.

A functional emulator would be huge. Having an emulator will not dissuade people from taking care of the real hardware; the Amiga and Macintosh communities didn't throw out everything they had once PC's became capable of running their software faster than physical machines ever could. Brilliant new developments for real hardware (the Vampire boards, A2SERVER, etc.) were made possible because of the existence of good emulation - and while I do doubt that anything remotely similar to the Vampire will be built for MIPS in my lifetime, who knows what else could be?

To me, IRIX itself is interesting, but only part of the reason why Silicon Graphics machines pique my curiosity. The look, feel, and technological design of the hardware is fascinating (which is why I believe that the VW320/540 and Altix-based machines also deserve some love), the history of the machines and what they were used for is fascinating, and the untapped potential still remaining in these old boxes is fascinating.

IRIX on a modern architecture would do nothing for those interested in preserving and using the machines that it was designed for. It would segregate software development between old and new tech, drum up no new reasons beyond sentimentality for anyone to dust off an old Octane, and take away resources from the hardware I'm interested in.

An emulator can bring more interest to the community - and that is never a bad thing, in my opinion.

[Raion]

I'd like to rebuff your point that it would divide the community to have irix on modern hardware. AROS, OS4 and MorphOS exist as ports to more modern hardware than the original 68000-based Amiga design.

We don't have the benefit of video games. Our CGI software was impressive in the 90s and 2000s but is extremely dated by modern standards.

This isn't an, I hate emulation, stance. The hardware is far less interesting to most people compared to the software. I didn't want these machines because they looked cool or because they did cool CGI stuff in the 90s. I wanted them because the desktop experience is among the most pleasant and the HPC aspects.

The other thing I like to bring up is that even if these Indy/Iris Indigo designs take shape they won't help when it comes to the other machines in the lineup, which will have to be started from scratch, developers developing on them will be limited to the MIPS III iteration. This in and of itself is why I refuse to fund emulation. There is no prospects for hire performance machines for many years even if I did so I would be contributing to the black hole even more than I'm already doing for other projects.

[CiaoTime]

Amiga software development was already segregated by the physical switch from 68K to PPC. 'Sides, I have used UAE much more than I have used AROS and MorphOS: OS4 and MorphOS are interesting in that they will still run on (extremely rare and esoteric) real Amigas, but at that point it's still a niche within a niche. I wouldn't be surprised if UAE is used a lot more commonly than the other aforementioned OS's among the hobbyist crowd.

SGI's were certainly involved in game development, and I'd love to see some of the period-accurate development suites dug up and released to the public. Plenty's known about the N64 side of things, but there's also PlayStation and arcade-related tech still in the underbrush. Still, Macs were never extraordinary gaming machines either, and they're still quite popular all the same.

The R5000 core in MAME is relatively robust - a few arcade games like San Francisco Rush and NFL Blitz run on R5000-based hardware already - so I don't see MIPS IV support being a problem once it's ready. The R5000 Indy in MAME can boot to PROM and do the same things that the R4000 Indy currently can.

[Raion]

As of right now from what I understand all of the development is focused on the MIPS III variants. The other thing to consider here is that while you can get into the firmware that's nowhere near accurate enough emulation to boot kernel. It's going to be several months of kernel panics and debugging before anywhere near & installable or bootable system happens. Unlike Amigas we are very much in the technical Stone age of emulation. And while one may have a case that an R5000 Indy is not far off, I would wager a guess of it being five to ten years before we have a working Odyssey graphics based computer.

When it comes to development I'm looking for quality not quantity, if we get 100 developers who are only interested in porting Linux that does nothing for us. That's why I try to find people who actually are interested in the desktop usage of these systems rather than having them look pretty.

In any case I believe this will still be a niche for many years to come even if we have a miracle tomorrow. There is no such thing as a lifeboat in a community like this

[commodorejohn]

Emulation is always a good thing to have, because hardware is mortal, and while an IRIX port to modern commodity hardware would be interesting in a lot of ways, it wouldn't be much more than a nicer, more consistent UI on top of commodity *nix if you couldn't run any actual IRIX software on it.

That said, yeah, emulation is rarely a perfect solution either. (And that goes doubly for MAME, where just trying to follow the basic structure of the source code is like looking into the Macro-nomicon.) Plus, hardware is just cooler

[LarBob]

The hardware is far less interesting to most people compared to the software. I didn't want these machines because they looked cool or because they did cool CGI stuff in the 90s. I wanted them because the desktop experience is among the most pleasant and the HPC aspects.

I don't think this is true, it seems like a lot of people got into the hobby due to the hardware behind these machines. After all, SGI pioneered a lot in the realm of graphics and high performance computing. Their advancements weren't all due to some magic software advancements. For example, their pioneering in the fields of dedicated 3D graphics hardware, accelerated geometry pipelining (they arguably brought it to the "mainstream" but didn't invent it), clustering (NUMAlink), etc. Personally, I have a copy of the "RealityEngine Graphics" paper by Kurt Akeley that I printed out and have in a school binder because I think a lot of what they did with hardware is fascinating. The engineering that went into many of these machines is beautiful. I'm sure others can chime in on this, but to me it seems inaccurate/biased to say that SGI hardware is far less interesting than the software to most people. It's fine if you didn't get into this for the hardware, but I don't think it's fair to say that most did.

[Irinikus]

The hardware is far less interesting to most people compared to the software. I didn't want these machines because they looked cool or because they did cool CGI stuff in the 90s. I wanted them because the desktop experience is among the most pleasant and the HPC aspects.

I don't think this is true, it seems like a lot of people got into the hobby due to the hardware behind these machines. After all, SGI pioneered a lot in the realm of graphics and high performance computing. Their advancements weren't all due to some magic software advancements. For example, their pioneering in the fields of dedicated 3D graphics hardware, accelerated geometry pipelining (they arguably brought it to the "mainstream" but didn't invent it), clustering (NUMAlink), etc. Personally, I have a copy of the "RealityEngine Graphics" paper by Kurt Akeley that I printed out and have in a school binder because I think a lot of what they did with hardware is fascinating. The engineering that went into many of these machines is beautiful. I'm sure others can chime in on this, but to me it seems inaccurate/biased to say that SGI hardware is far less interesting than the software to most people. It's fine if you didn't get into this for the hardware, but I don't think it's fair to say that most did.

I agree! I got into the hobby more for the hardware initially, as it's really cool and laid the groundwork for many systems which we take for granted today, but I'm now interested in both.

[LarBob]

I agree! I got into the hobby more for the hardware initially, as it's really cool and laid the groundwork for many systems which we take for granted today, but I'm now interested in both

Agreed, though I do think that emulation will be super nice if we get it as well. Even though we don't have video games to boost the popularity of SGI machines, it will at least help to get people into the scene. It's one thing to look at a video something and another thing to experience and play with it yourself. It'll be nice for people to test out IRIX, mess with it, etc. without needing a machine, and a major thing in my opinion is that it'll allow developers to tinker with IRIX and get more software ported without needing a machine! After all, we're here because we love these machines, and I think emulation will overall have more pros than cons.