Why not use the second core for chipset emulation?

thank you @hans you have the patience of a saint

@Hans

second

there enough, more or less active, classic routes anyone could go if they wanted to.

just let us drop the backwards compatibility and move on already

@ferrels

Quote:
Well it lower the CPU usage on main core for one.

but Yeh.. they have half working SMP implementation done, lets say someone start working on this, and when its don, its no longer compatible with AmigaOS4.2 SMP, that be kind a blow… to considering considerable time it take to figure out how to do this, and implement it.

Quote:

it does not mean, nobody is working on it, it might take a long time, and it might not be worth it (As in paying someone to do it, and getting some revenue back in the form of sales), but it possible it can happen, already lot of time is spent on it.

It's just that OS will need to handle L1 Cache, and correct for all public data, and there is lot of list and messages structure, that need some kind of management system. If can be done with current implementation, maybe, but sure it won’t be efficient without, being specifically written to support.

Perhaps L1 can be disabled, but that effect performance. We see what happens, the people who are supposed to work on it are not talking, so who know where we are at. Hopefully get our yearly update at AmiWest.

First time, they talked about it was back 2005, there was talk about dual G4 CPU module for AmigaONE-XE, it’s now almost 20 years after, it was first talked about. So yeh… this was not easy, true has many roadblocks and unnecessary challenges, that has hampered progress.

Quote:

Yeh, it seems that is a likely outcome, of this experimentation.

this why NinjaCyborg perhaps suggested using the other cores in a different way. Having an implementation that allows AMP (Asymmetric Multi-Processing) won’t be bad, but a little disappointing.

I do not know is required for AMP, but I believe this what the OS needed to support for a successful UAE on the second core concept, without SMP.

@Hans

It almost looks like you are one of the last remaining coders holding the flag!
Thank you for that.


You wrote a post a couple of weeks ago, I did draft an answer, but didn't get the time to post it. Now I lost it.

I really welcome your initiatives and the energy you put in the Kea Campus. I wish I had the time to start coding and support you. I can't atm, but the time will come.

The main problem I see with all the effort put in keeping the Amiga "alive" is the mess an politics around the OS.

It really is beyond me, how any newcomer would catch interest if there is so little information or progress to be seen regarding the OS, the development environment and all the messs with the hardware too. The Tabor story is really sad. I know Trevor had the best intentions with it and contrary to many I am grateful he is still around and has not thrown the towel. I can't afford to buy it now, and I am reluctant because of the many imcompatibilities, moreover the x5000, x1000 and the other machines still work well. I really hope the X1000 will get some love in form of CFE or (even a UBoot firmware?) updates.
And

@Tuvok

Your ignore nice working done on clib4, by many developers, and it’s necessary. For progress to be made, on other topics.

@NinjaCyborg

Quote:

The question would be how much additional and useful software would be able to run directly in OS4 with such a solution. I sold my BPPC before OS4 was available for it so I've never been able to make a comparision between PPC hardware with the chipset and hardware without. I do know that WHDload is fundamentally incompatible with OS4, so the vast majority of games that require the chipset simply cant be run without resorting to UAE.

@Tuvok

Quote:
One of the remaining developers, but I'm grateful that there are others. LiveForIt has pointed to clib4 which has 8 contributors to it. There are others too.

We do need more developers, though. And more people doing cool stuff on the Amiga in general. That's one of the motivations behind the Kea Campus' Amiga Corner

Quote:
Thanks.

Just to clarify for everyone: you don't need to be a developer/coder to oin the campus and support my work. Everyone is welcome, and the financial support certainly helps keep it going. If you're not a developer, then you can think of it as a Patreon style sponsorship of Amiga content creation (except Patreon doesn't take a 8-12% cut).

Just make sure you tell me that you're there for the Amiga corner. More info here.

Hans

@Rob

Quote:
Does UAE run the chipset emulation in separate threads from the CPU emulator? If so, then it should be possible to make them run on the unused core(s).

Of course, that would be a lot easier if the OS supported SMP..

Hans

@Hans

it does not.

When you read and write to the chipset, that’s when most chipset is emulated. this how it synchronized clock cycles to the video output. Every read and write, equals a amount of clock cycles, that is used for the video rendering. This who it can get exact output, even when emulation is not real time.

Even the blitter is done like that, even if in that case, should been done in parallel.

So you need to run 99% on the other core, that’s probably fine on X50x0, where emulation is running fine on just one core, the main advantage is that it frees up the main core.

Of course, we can all agree that a bad implementation is better than no implementation.

If anyone is interested in: we have interviewed Matthew Leaman, Managing Director of A-EON and Amiga Kit about the A600GS (and some more:

https://www.amiga-news.de/en/news/AN-2024-10-00040-EN.html

@Primax

Thanks! But perhaps start a new thread, as people may want to comment on the interview, which would further derail this topic.