Yokemate of Keyboards
Posts: 4967 from 2009/1/28
From: Delaware, USA
>Why would an x64-native MorphOS need to run on "a software based hypersvisor on a X64 system"?
Because I use VM boxes to run multiple OS'.
>> Power 9 [...] could run existing MorphOS with no alteration
>using QEMU's mac99 virtualization with KVM-PR, that is.
Or other virtualizations. I'd like to see something emulating a P5040 based X5000.
>> while providing a multitude of additional threads that could run a more
>> advanced little endian 64 based box (better than a X64 version)
>Why not run this potential little-endian PPC64 version of MorphOS on bare metal (PowerNV) instead of boxed?
Everything on Power4-9 runs with the hypervisor enabled, even if its only running a single thread.
>> we could even devote a few threads to emulating X64 [...],
>> for simultaneous Power and X64 threads via the hypervisor
>I'm not sure the hypervisor can help with emulating x64 on POWER.
Niether am I, I haven't had the time to study it well enough, but what little I have covered gives me the impression it will be the core of ourability to run X64 and PPC code concurrently.
>>no doubt OS4 as well as all we would need would be P50X0 emulation
>OS4 already runs fine on QEMU's Sam460 emulation. Btw, QEMU already has P50X0/e5500 emulation but that >isn't enough to run OS4 as there's more to a board/system than just the CPU.
Frankly, I not that concerned as to what the emulated system is, I'd just like to beable to freak out OS4 fanatics by being able to run their OS as part of the type of system I've described.
>> or better yet e6500 core emulation
>OS4 doesn't run on e6500.
No, but MorphOS could, and Linux does.
>> PCIe slots that are four generations ahead of Tabor
Correct, PCIe v1.0 vs v4.0, so not four generations, but only three. However, if the specs I have are correct, PCIe v1.0 has a transfer rate of 250MB/sec vs PCIe v4.0's transfer rate of 1969MB/sec.
Or about eight times the bandwidth per lane, with Tabor (and Samantha) only supplying 4 PCIe lanes to the video card while Blackbird's expansion slots provide 8 or 16 lanes.
In other words, a potential of almost 16 times the bandwidth.
>> Take existing MorphOS elements, enhance them for PPC64le,
>> and then just rework the code for X64.
>Code that has been cleaned from big-endian assumptions and adapted to modernized MorphOS API should >just need a recompile for a potential x64 MorphOS.
Yes, this step would be relatively simple.
>> What would really make it cool is if we could develop a cut and paste
>> function via the hypervisor that would allow us to transfer data from one
>> OS session/Window to another. Think about it. You pull something from
>> legacy AmigaOS or MorphOS, and could paste it directly into a more
>> advanced environment for further development, again, on one system
>QEMU supports clipboard sharing between guest and host (and thus also between guests). There needs to >run a special service in the guest OS for this to work, though, which doesn't exist yet for AmigaOS or >MorphOS.
So the facility is there if we hook into it? Cool!
To me, this looks like the superior platform, not X64.[ Edited by Jim 01.01.2019 - 17:55 ]
"Never attribute to malice what can more readily explained by incompetence"