[darkhelmet02]

Has anyone ever looked into implementing a mips processor into an fpga to replace dead processor modules? Is there even enough documentation out there to be able to do anything resembling this? I apologize if this is a completely absurd question, but I can't really find much discussion on this topic

---

https://mobile.twitter.com/ashevans81/st...64?lang=en

I was able to find this link, but there is no evidence about whether or not this worked or if he got far with it.

[kubatyszko]

Nothing for SGI machines at the moment.
It may be possible with some simpler machines such as Indy.
Mame supports Indy emulations, so it does have emulation sources for the R4000 series CPU and that makes it possible to implement such a CPU in a FPGA.
Whether this is going to happen I don't know, pretty unlikely as the Indy's and R4000 machines are somewhat abundant and the community is not very big.

[darkhelmet02]

Nothing for SGI machines at the moment.
It may be possible with some simpler machines such as Indy.
Mame supports Indy emulations, so it does have emulation sources for the R4000 series CPU and that makes it possible to implement such a CPU in a FPGA.
Whether this is going to happen I don't know, pretty unlikely as the Indy's and R4000 machines are somewhat abundant and the community is not very big.

I wondered about this, i know mame can emulate indy and indigo2. That would at least be a good place to start, and it would be interesting to instead attempt an R5000 in an fpga seeing as its a much less common cpu. Does anyone happen to know where specifications for sgi cpus could be acquired, or how the developers for mame were able to fully emulate the R4000? This is a project I would be interested in looking further into if more people happened to be interested

[Raion]

The CPUs are well documented, but it's things like proprietary connectors, and undocumented boards/pathways that prevent it. That being said, user solidcore thinks he can do it eventually for the Indy or something simple.

[jpstewart]

There are already soft microprocessors based on the MIPS instruction set.  See the Wikipedia article "Soft Mircroprocessor" for a list.  They're not the right MIPS chips for SGI systems, but it's a start or proof of concept.  So the processor itself is not impossible. 

But none of that will help with the difficulties Raion mentioned with the modules as a whole.

[darkhelmet02]

There are already soft microprocessors based on the MIPS instruction set.  See the Wikipedia article "Soft Mircroprocessor" for a list.  They're not the right MIPS chips for SGI systems, but it's a start or proof of concept.  So the processor itself is not impossible. 

But none of that will help with the difficulties Raion mentioned with the modules as a whole.

I'll check that out, thanks!

---

The CPUs are well documented, but it's things like proprietary connectors, and undocumented boards/pathways that prevent it. That being said, user solidcore thinks he can do it eventually for the Indy or something simple.

I didn't really think about those weird connectors that different cpu modules use, that would be quite an issue.

[Raion]

Yeah, I didn't want to burst your bubble, but it's not like it's easy to adapt a pin-compatible module in place.

[shrek]

Don't the CPU boards in a lot of these old workstations have a bunch of poorly documented support chips? It might not just be the CPU and module connector you'll have to worry about...

[Raion]

That too

[darkhelmet02]

Yeah, I didn't want to burst your bubble, but it's not like it's easy to adapt a pin-compatible module in place.

Hey, no worries! it was just something I was curious about, I mostly wanted to hear some opinions on how difficult it would be