@NutsAboutAmiga

Quote:

NutsAboutAmiga wrote: @Karlos Guess it’s a kind yes and no answer to that, when you JIT something you to keep status of guest OS CPU registers in a table, and status of Guest OS CPU flags, its not like true native programs.

Who cares? The most important thing is that the application is running, and very fast. It's the result that counts (and should only count).
Quote:

there some wait time, for JIT to generate code. If and when the JIT cache is flushed, or it hits new code, that again cannot be compared to true native code.

Not even for Petunia or TranceJIT.
Quote:

Also JIT has to complete compilation pretty quickly, while GCC is allowed to spend a lot time thinking about the best result.

Then use AoT compilation and you can apply all optimizations that you like (even at the global, full application level).
Quote:

NutsAboutAmiga wrote: @cdimauro Yes and this why emulations we have today are pretty poor, considering the hardware is almost the same on host as the guest,

Not really: PC hardware evolved; A LOT.
Quote:

only the CPU really is required to be emulated, maybe a few north and southbridge chipsets. Technically you can write OS4 drivers for the host hardware, or the emulation box.

Technically you can completely remove any link/dependency on AmigaOne/SAM machines (included UBoot et similia) and let OS4 immediately boot on a virtual machine which has transparently replaced and emulated all hardware or given direct access to the hardware to boost the performance as much as possible.

@NutsAboutAmiga

Quote:

there some wait time, for JIT to generate code. If and when the JIT cache is flushed, or it hits new code,

That is an implementation detail. A JIT does not have to work that way. Many JITs work like that because JIT is only used for rare components that have never been written as native.
And thus the cost of discarding a translated unit and then re-translating it sometime later is not a big deal and it is not worth the effort to improve it.

On a system, like karlos virtual m68k, where ALL code will ALWAYS be translated, in that situation it does make sense to put in effort to reduce the translation overhead.
The obvious way to do so is to persistently store all translated units as soon as they are created so that next time you need them they are already translated and you just page them in. For example, store the translated units as a separate data-fork in the original binary. A FAT binary if you will that is built on demand.

Eventually all code will have been translated and already stored as units that can be loaded when the application is loaded and from there on there will be ZERO overhead or time spent on translating code going forward.

Many, not all, JITs already do something similar. In that case there is virtually no overhead and no slowdown after the first time you have run the app. You basically have ended up with an almost fully translated application.

Last edited by V8 on 03-Nov-2022 at 12:54 AM.

@Bosanac

Quote:

Bosanac wrote: Quote: In addition calling Qualcomm's latest claims "riddled with inaccuracies," Arm said via email to Fierce Electronics Friday that its original claim against Qualcomm filed in early September is “aimed at protecting the Arm ecosystem and partners who rely on our intellectual property and innovative designs. Qualcomm has violated the terms of the Nuvia license agreement and yet it continues to use the technology, unlicensed. Arm is seeking to enforce Qualcomm’s obligation to destroy and top using the Nuvia designs that were derived from Arm’s technology.” https://www.fierceelectronics.com/sensors/arm-qualcomm-legal-battle-suggests-oems-need-arm-ip-licenses-not-chip-firms

Under SoftBank's management, ARM's business model is changing.

From Dylan Patel,
----------------
According to the updated Qualcomm counterclaim, after 2024, Arm is no longer going to license their CPUs to semiconductor companies such as Qualcomm under technology license agreements (TLAs). Instead, Arm will only license to the device-makers. Arm is allegedly telling OEMs that the only way to get Arm-based chips will be to accept Arm’s new licensing terms. Qualcomm claims that Arm is lying to Qualcomm’s OEM partners about Qualcomm’s licensing terms.

Furthermore, Qualcomm claims that Arm is telling the OEMs that semiconductor manufacturers will not be able to provide other elements of their Arm-based SOCs that Arm also offers as a licensed product. This includes GPUs, NPUs, and ISP. It seems that Arm is effectively bundling its other IP with the CPU IP in a take-it-or-leave-it model. That would mean Samsung’s licensing deal with AMD for GPU or Mediatek with Imagination GPU is no longer allowed after 2024. Furthermore none of these firms could use their in-house ISP or NPU despite it being far superior to Arm's.

If true, it seems Arm is playing very dirty with their threats to Qualcomm and OEMs. Mediatek, Samsung, and other Arm partners should be very scared. This is going to accelerate RISC-V roadmaps rapidly. It also reeks of anti-competitive behavior. Nvidia has a 20-year Arm license secured, so they will be fine. Apple obviously has great licensing terms due to their history with founding Arm. We hear Broadcom also has very favorable terms as well.

If you don’t believe us and think this all seems crazy, we get it. This is what we understand after reading the legal document. Just read this excerpt from Qualcomm’s counterclaim that was filed to the courts yesterday. We have attached the document at the end of the article.

--------------

_________________
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB
Amiga 1200 (Rev 1D1, KS 3.2, PiStorm32lite/RPi 4B 4GB/Emu68)
Amiga 500 (Rev 6A, KS 3.2, PiStorm/RPi 3a/Emu68)

@NutsAboutAmiga

Quote:

NutsAboutAmiga wrote: @Hammer You think the ARM or X86 will huge advantage, but they always problem of running something none native on a host, and hardware looks impressive on paper, the emulation environment is not exposing the hardware to its guest os & hw, essential what this means is that accurately support for host os is none existent in guest OS. it does not matter what GL specification your host has, It’s not used.

I didn't have visibility for Linux ARM's UAE maturity.

On WInUAE with Wazp3D and native code execution enabled, hardware-accelerated Wazp3D is available e.g. Quake 68K running on WinUAE with Warzp3D https://www.youtube.com/watch?v=kb05QmY8fLo

_________________
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB
Amiga 1200 (Rev 1D1, KS 3.2, PiStorm32lite/RPi 4B 4GB/Emu68)
Amiga 500 (Rev 6A, KS 3.2, PiStorm/RPi 3a/Emu68)

@Deaths_Head

Quote:

Considering Apple have moved to arm & chip makers such as AMD & intel have got arm licences, appears to signal that x86 is approaching a ceiling in efficiency

The "appears to signal that x86 is approaching a ceiling in efficiency" narrative is not true.

Intel plans to replace Golden Cove with 14th Gen Redwood Cove P-Core architecture while Crestmont replaces Gracemont E-Core.

AMD plans to replace Zen 4 with Zen 5. https://wccftech.com/amd-next-gen-zen-5-cpus-get-early-support-within-hwinfo/ HWinfo has early support for Zen 5.

_________________
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB
Amiga 1200 (Rev 1D1, KS 3.2, PiStorm32lite/RPi 4B 4GB/Emu68)
Amiga 500 (Rev 6A, KS 3.2, PiStorm/RPi 3a/Emu68)

@Hammer

That doesn’t contradict the claim though.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@Hammer

Read more.

http://thellier.free.fr/Wazp3D.htm

It’s not AmigaOS, its AROS, and it’s not 100% supported,
not is OpenGL, it's Warp3D emulation.
its not even a normal Wazp3D install its QuarkTex.

Warp3D: 3Dprog-->Warp3D-->AmigaOS-->Driver-->3Dcard
QuarkTex:3Dprog-->Warp3D-->AmigaOS-->WinUAE-(Windows)->Warp3D.alib-->OpenGL-->Driver-->3Dcard
Wazp3D: 3Dprog-->Warp3D-->AmigaOS-->68040

And as page says many things are not fully supported.
No TCL support, no Multi texture support, lots other stuff missing.
this not full access to the hardware, or full hardware acceleration.

As I wrote emulation for Amiga is has poor hardware acceleration support,
compared to other emulators. and most if not all rendered in FB before its displayed.
even if some GFX libs functions are acclreated by native code.

native code is not the GPU, its often the host CPU.

Last edited by NutsAboutAmiga on 03-Nov-2022 at 09:20 PM.
Last edited by NutsAboutAmiga on 03-Nov-2022 at 09:18 PM.
Last edited by NutsAboutAmiga on 03-Nov-2022 at 09:16 PM.

_________________
http://lifeofliveforit.blogspot.no/
Facebook::LiveForIt Software for AmigaOS

@kolla

Quote:

kolla wrote: @Hammer That doesn’t contradict the claim though.

It does.

@NutsAboutAmiga

Quote:

NutsAboutAmiga wrote: @Hammer Read more. http://thellier.free.fr/Wazp3D.htm It’s not AmigaOS, its AROS, and it’s not 100% supported, not is OpenGL, it's Warp3D emulation. its not even a normal Wazp3D install its QuarkTex. Warp3D: 3Dprog-->Warp3D-->AmigaOS-->Driver-->3Dcard QuarkTex:3Dprog-->Warp3D-->AmigaOS-->WinUAE-(Windows)->Warp3D.alib-->OpenGL-->Driver-->3Dcard Wazp3D: 3Dprog-->Warp3D-->AmigaOS-->68040

What happens after the 68040?

Because the documentation says this:

GLeXcess for OS3 (need stormmesa) in hardware in WinUAE using Wazp3D->OpenGL32.DLL

when talking about WinUAE.
Quote:

And as page says many things are not fully supported. No TCL support, no Multi texture support, lots other stuff missing.

Does Warp3d.library support them? If yes, on which APIs? Because I don't see them on the table of implemented functions.
Quote:

this not full access to the hardware, or full hardware acceleration.

It has it, according to the documentation:

Inside WinUAE & AROS can also render in hardware
[...]
So better performances in hardware mode
[...]
This function allow hardware accelerated "Porter & Duff Alpha compositing" (used in OS4 menus/windows transparency)
[...]
Early OPENGL32.DLL support: Like QuarkTex the latest Wazp3D/WinUAE is now hardware accelerated
Now can choose in Wazp3D-Prefs the Renderer: soft or hard (then will use hardware OpenGL)
[...]
+HardwareDriver Lie - Pretend to be have a perfect hardware driver that can do all perfectly
[...]
+Renderer

hard :
WinUAE: draw in WinUAE's BackBuffer then do glReadPixels to the current Amiga-bitmap memory

hard(overlay) : Faster drawing method. Open a double buffered sub-window inside (upside) the current window This mode is recommended for full-screen 3D-Progs.

This hard mode works only inside WinUAE in windowed mode. Need also a 32bits BGRA Workbench or screen
But support windowed & own screen apps
You can select Filtering & AntiAliasing & StateTracker in hardware mode too
Quote:

As I wrote emulation for Amiga is has poor hardware acceleration support, compared to other emulators. and most if not all rendered in FB before its displayed. even if some GFX libs functions are acclreated by native code.

Then there's room for getting better...
Quote:

native code is not the GPU, its often the host CPU.

And what's the difference with Petunia or TranceJIT?

cdimauro Quote:

What happens after the 68040?

The 68040 does the software 3D rendering so it is the end of the line. The 68040 is the last CPU Warp3D is compiled for (Avenger libraries for Voodoo 3-5). They 68040 GCC compiled libraries will crash on a 6888x due to lack of FDop and FSop instructions which GCC uses while the lack of FINT(RZ) instructions make them sub optimal on the 68060 but they do run on the 68060. Some of the older Warp3D libraries including the Warp3D.library itself will work with the 6888x but some of them use a really old and poor compiler (EGCS?). There were never CPU specific versions of the 68k Warp3D libraries, the code quality was poor, there are bugs that were never fixed and v5 of the API was never added. That was during the transition to PPC where the 68k was discontinued like for AmigaOS 4. Maybe it is not much worse than running 16 bit 68000 code on a 32 bit in-order superscalar 68060 without instruction scheduling like Hyperion's 68k AmigaOS 3.x compiled with SAS/C.

cdimauro Quote:

Does Warp3d.library support them? If yes, on which APIs? Because I don't see them on the table of implemented functions.

Warp3D added support for multi-texturing in v5 so the 68k didn't get it but Alain Thellier added support to Wazp3D so it should have that. I'm not sure TCL/T&L support was ever added to Warp3D. Some PPC AmigaOS 4 users are using NovaBridge in place of Warp3D if they have a new graphics card with shaders and can afford Warp3D Nova and NovaBridge after buying a X5000 and gfx card.

cdimauro Quote:

And what's the difference with Petunia or TranceJIT?

I think some PPC AmigaOS 4 users are feeling nervous. Do they not realize that PC hardware that costs half the price of their "PC hardware with a PPC CPU" can emulate their hardware at full performance and has hardware GPU acceleration? They never wondered why only a few thousand and now few hundred AmigaNOne hardware units was being sold? Does it hurt their ego that their hardware is slipping from over priced mid performance to low performance hardware? If making low performance hardware, wouldn't it be better to make it 68k Amiga hardware with a small footprint and mass produce it for all Amiga users?

Last edited by matthey on 04-Nov-2022 at 01:18 AM.
Last edited by matthey on 04-Nov-2022 at 01:12 AM.

@Hammer

Quote:

Hammer wrote: From Dylan Patel ---------------- According to the updated Qualcomm counterclaim, after 2024, Arm is no longer going to license their CPUs to semiconductor companies such as Qualcomm under technology license agreements (TLAs). Instead, Arm will only license to the device-makers. Arm is allegedly telling OEMs that the only way to get Arm-based chips will be to accept Arm’s new licensing terms. Qualcomm claims that Arm is lying to Qualcomm’s OEM partners about Qualcomm’s licensing terms

PATEL?

_________________
I HAVE ABS OF STEEL
--
CAN YOU SEE ME? CAN YOU HEAR ME? OK FOR WORK

@matthey

Quote:

matthey wrote: cdimauro Quote: What happens after the 68040? The 68040 does the software 3D rendering so it is the end of the line. The 68040 is the last CPU Warp3D is compiled for (Avenger libraries for Voodoo 3-5). They 68040 GCC compiled libraries will crash on a 6888x due to lack of FDop and FSop instructions which GCC uses while the lack of FINT(RZ) instructions make them sub optimal on the 68060 but they do run on the 68060. Some of the older Warp3D libraries including the Warp3D.library itself will work with the 6888x but some of them use a really old and poor compiler (EGCS?). There were never CPU specific versions of the 68k Warp3D libraries, the code quality was poor, there are bugs that were never fixed and v5 of the API was never added. That was during the transition to PPC where the 68k was discontinued like for AmigaOS 4. Maybe it is not much worse than running 16 bit 68000 code on a 32 bit in-order superscalar 68060 without instruction scheduling like Hyperion's 68k AmigaOS 3.x compiled with SAS/C.

Understood, but this belongs to Warp3d, whereas the 68040 was mentioned on Wazp3d:

Wazp3D: 3Dprog-->Warp3D-->AmigaOS-->68040

which could be hardware-accelerated. Hence my question. But maybe it was a typo.
Quote:

cdimauro Quote: Does Warp3d.library support them? If yes, on which APIs? Because I don't see them on the table of implemented functions. Warp3D added support for multi-texturing in v5 so the 68k didn't get it but Alain Thellier added support to Wazp3D so it should have that.

That was my idea. It would be interesting to have a comment from Alain about the correct status of Wazp3D.
Quote:

I'm not sure TCL/T&L support was ever added to Warp3D. Some PPC AmigaOS 4 users are using NovaBridge in place of Warp3D if they have a new graphics card with shaders and can afford Warp3D Nova and NovaBridge after buying a X5000 and gfx card.

Indeed. TCL/T&L is Stone Age technology.
Quote:

cdimauro Quote: And what's the difference with Petunia or TranceJIT? I think some PPC AmigaOS 4 users are feeling nervous. Do they not realize that PC hardware that costs half the price of their "PC hardware with a PPC CPU" can emulate their hardware at full performance and has hardware GPU acceleration? They never wondered why only a few thousand and now few hundred AmigaNOne hardware units was being sold? Does it hurt their ego that their hardware is slipping from over priced mid performance to low performance hardware?

I think that they are aware of it, but they'll not change their mind because they want to have the physical machine. Either it's a psychological thing (the idea to have a system which has a PowerPC inside) or they think that buying such machines is still of service for their "cause".

For Nuts is different, because he's one feeling nervous: he sees all of this like a threat for his Totem. It's clearly visible, especially when he says things that applies also to PowerPC (Petunia, TranceJIT), I highlight it, and he systematically drops the discussion not answering.
Quote:

If making low performance hardware, wouldn't it be better to make it 68k Amiga hardware with a small footprint and mass produce it for all Amiga users?

Why not, if it's profitable.

But the PowerPC aficionados are used to spend a lot of money for their hobby and getting obsolete hardware. So, they don't care if something like what you're proposing could be made.

However in the last years some re-discovered the 68k, which is flourishing again: they are likely interested on it.

cdimauro Quote:

Understood, but this belongs to Warp3d, whereas the 68040 was mentioned on Wazp3d: Wazp3D: 3Dprog-->Warp3D-->AmigaOS-->68040 which could be hardware-accelerated. Hence my question. But maybe it was a typo.

I expect it means that software rendering is an option. Maybe hardware rendering doesn't work in all setups? If the Warp3D API maps to the host 3D API, shouldn't hardware acceleration work on more hardware than AmigaOS 4 hardware that requires drivers for each GPU?

cdimauro Quote:

That was my idea. It would be interesting to have a comment from Alain about the correct status of Wazp3D.

The aminet docs say Wazp3D supports v5 functions.

http://aminet.net/package/driver/video/Wazp3D Quote:

MOST IMPORTANT FEATURES o Compatible with the famous Warp3D.library for AmigaOS o Inside Aros & WinUAE & Morphos can also render in hardware o StormMesa & MiniGL compatibility o The 97 functions from the latest Warp3D.library v5 are presents o Sources included (...so Wazp3D has been ported to AROS) o Can be software only (=slow) : no need to have 3D hardware o Work in UAE so can serve to compile/test your 3D softwares inside UAE o Wazp3D can lie and pretend to have a "perfect hardware driver" so the 3D-softwares will not fail due to a lacking feature. o Full Documentation, faq and grabs at http://thellier.free.fr/Wazp3D.htm o Unlike QuarkTex you can run multiple Warp3D applications concurrently o Built-in debugger o Demo-program http://aminet.net/dev/src/Cow3D.lha to use with Wazp3D

You could trust me that Warp3D v5 added multi-texturing. It was the major feature added. I have a pdf called w3d_v5.pdf with the title when opening of "Warp3D V5 API changes". I couldn't find the pdf anywhere online. So much for support. Is that because Warp3D is abandoned or because the PPC AmigaNOne market is too small to provide support?

cdimauro Quote:

I think that they are aware of it, but they'll not change their mind because they want to have the physical machine. Either it's a psychological thing (the idea to have a system which has a PowerPC inside) or they think that buying such machines is still of service for their "cause". For Nuts is different, because he's one feeling nervous: he sees all of this like a threat for his Totem. It's clearly visible, especially when he says things that applies also to PowerPC (Petunia, TranceJIT), I highlight it, and he systematically drops the discussion not answering.

Hypex also didn't seem to realize that PPC hardware could be emulated with full performance on PC hardware that costs half as much as AmigaNOne hardware. The 68k Amiga users have faced the reality that emulators are hundreds of times higher performance but rev6 68060s still go for more than an A1222 will sell for and likely have more demand. Someone was asking in a thread about a ReAmiga 1200 like the ReAmiga 3000+. A ReAmiga 1200 or THEA1200 Maxi with an empty 68060 CPU socket and AA+ would smoke the A1222 in sales. When the unbranded little guys do what the Amiga "PC hardware with a PPC CPU" guys won't, it will be too late for an "official" Amiga rebirth done right.

@matthey

Quote:

matthey wrote: cdimauro Quote: Understood, but this belongs to Warp3d, whereas the 68040 was mentioned on Wazp3d: Wazp3D: 3Dprog-->Warp3D-->AmigaOS-->68040 which could be hardware-accelerated. Hence my question. But maybe it was a typo. I expect it means that software rendering is an option. Maybe hardware rendering doesn't work in all setups?

It only works on WinUAE or AROS, according to the Wazp3D documentation.
Quote:

If the Warp3D API maps to the host 3D API, shouldn't hardware acceleration work on more hardware than AmigaOS 4 hardware that requires drivers for each GPU?

I assume so.
Quote:

cdimauro Quote: That was my idea. It would be interesting to have a comment from Alain about the correct status of Wazp3D. The aminet docs say Wazp3D supports v5 functions. http://aminet.net/package/driver/video/Wazp3D Quote: MOST IMPORTANT FEATURES o Compatible with the famous Warp3D.library for AmigaOS o Inside Aros & WinUAE & Morphos can also render in hardware o StormMesa & MiniGL compatibility o The 97 functions from the latest Warp3D.library v5 are presents o Sources included (...so Wazp3D has been ported to AROS) o Can be software only (=slow) : no need to have 3D hardware o Work in UAE so can serve to compile/test your 3D softwares inside UAE o Wazp3D can lie and pretend to have a "perfect hardware driver" so the 3D-softwares will not fail due to a lacking feature. o Full Documentation, faq and grabs at http://thellier.free.fr/Wazp3D.htm o Unlike QuarkTex you can run multiple Warp3D applications concurrently o Built-in debugger o Demo-program http://aminet.net/dev/src/Cow3D.lha to use with Wazp3D You could trust me that Warp3D v5 added multi-texturing. It was the major feature added. I have a pdf called w3d_v5.pdf with the title when opening of "Warp3D V5 API changes". I couldn't find the pdf anywhere online.

Yes, I saw it, but if you take a look at the the table at the bottom ("IMPLEMENTED FUNCTIONS") it looks that some Warp3D features (included some v5 ones) aren't fully implemented.

I don't know if this hurts compatibility or causes issues for the applications using the library. That's why it was interesting to see what Alan could tell about it.
Quote:

So much for support. Is that because Warp3D is abandoned or because the PPC AmigaNOne market is too small to provide support?

I think that Warp3D is here to stay because it's the oldest and most used library for 3D applications / games.

Then if it's remapped to Nova or something like that for OS4, it doesn't matter, since it's a transparent thing.

For all other platforms it's enough that Wazp3D is mapped to the host 3D APIs. OpenGL is ok for Windows because I think that it's easy to remap Wazp3D APIs calls to some equivalent OpenGL ones; but it's sub-optimal. Nowadays we have the DirectX12 APIs which allow to better use the underneath GPU hardware, but porting to them requires A LOT of work which isn't justified for a one-man project.
Quote:

cdimauro Quote: I think that they are aware of it, but they'll not change their mind because they want to have the physical machine. Either it's a psychological thing (the idea to have a system which has a PowerPC inside) or they think that buying such machines is still of service for their "cause". For Nuts is different, because he's one feeling nervous: he sees all of this like a threat for his Totem. It's clearly visible, especially when he says things that applies also to PowerPC (Petunia, TranceJIT), I highlight it, and he systematically drops the discussion not answering. Hypex also didn't seem to realize that PPC hardware could be emulated with full performance on PC hardware that costs half as much as AmigaNOne hardware.

Even much less than half.

On April I've spent around €1000 for a new PC with 12900K, with 32GB of good DDR4 (memory latency is very important for emulators) and 1TB of NVMe/PCIExpress4. The last month I've completed it with a GeForce 3060 which costed me €300. But you don't need so high specs for a PC which has to emulate an OS4 machine.
Quote:

The 68k Amiga users have faced the reality that emulators are hundreds of times higher performance but rev6 68060s still go for more than an A1222 will sell for and likely have more demand. Someone was asking in a thread about a ReAmiga 1200 like the ReAmiga 3000+. A ReAmiga 1200 or THEA1200 Maxi with an empty 68060 CPU socket and AA+ would smoke the A1222 in sales. When the unbranded little guys do what the Amiga "PC hardware with a PPC CPU" guys won't, it will be too late for an "official" Amiga rebirth done right.

It's already too late: since YEARS. But if you (not you: it's a general statement) don't want to see it, you will not see it.

The A1222 was already dead when the geniuses decided to use a PowerPC CPU which has a completely incompatible FPU.
Now it's even much worse, since GCC already dropped support for those CPUs some time ago. So, OS4 cannot use a modern compiler. It means that this o.s. either should be stuck to the latest version which supports this crappy CPU (thus, hurting the modernization of the platform) or at certain point in time OS4 will go for a newer GCC and A1222 will not be supported anymore.
How many OS4 users that are interested on buying an A1222 (because it's cheap) know that they might end up having a piece of junk after some time?

Regarding the 68k and Amiga, the Amiga Replacement Project opened a big door on getting a more modern Amiga-compatible hardware keeping full compatibility:

"This is an attempt to make clean Verilog sources for each chip on the Amiga. Unlike MiniMig and related implementations, this seeks to maintain perfect compatibility with the whole chip and the whole chip only to serve as a second-source to the slowly disappearing stock of Amiga chips. These will be drop-in replacements."

It looks a very good starting point for a renewed Amiga platform. However that's only for the chipset, whereas something simil is missing for the CPU.

@cdimauro

Quote:

But you don't need so high specs for a PC which has to emulate an OS4 machine.

Trust me, you do, unless something drastically is done to support hardware virtualization techniques in the emulator.

Quote:

Now it's even much worse, since GCC already dropped support for those CPUs some time ago. So, OS4 cannot use a modern compiler. It means that this o.s. either should be stuck to the latest version which supports this crappy CPU (thus, hurting the modernization of the platform) or at certain point in time OS4 will go for a newer GCC and A1222 will not be supported anymore.

When we are talking about GCC/C or C++

We have like preprocessor, the compiler, and assembler and the linker, and clibs.

You can pick one or the other or use part of it, if you wan to you just use preprocessor, for example.
of cause if you’re going to pick and choice the build process becomes more complicated. but be good why overcome incompatibility in the preprocessor for example.

Most of changes are in clibs, not in the compiler, on 680x0 amiga you using a really old compiler, geektools package, for many projects it does not matter in most cases.

For A1222 that different FPU, you have some of CLibs that use compatible integer opcode, and some libs that uses float, that’s not so compatible, in some case building your own libs, can be a good idea for optimal performance.
The point anyhow the doors are closing on AMCC460 CPU, this why A1222 will be the only option, with exception of the X5040 and ACube-Systems laptop project. In my eyes the laptop project is more interesting then even the X5040 because it’s a different formfactor when we have had before.

Last edited by NutsAboutAmiga on 05-Nov-2022 at 12:16 PM.

_________________
http://lifeofliveforit.blogspot.no/
Facebook::LiveForIt Software for AmigaOS

@matthey

I don’t remember exactly, why Warp3D support was dropped, but I believe it was because API was old and outdated, and they wanted new features.

They did add a few features into Warp3D, but many old games suffered from bugs, and problems. If don’t remember wrong their issues where that newer graphics cards where so backwards compatible. and got worse with etch newer model.

Nova basically starts over from a blank page.

Last edited by NutsAboutAmiga on 05-Nov-2022 at 12:26 PM.

_________________
http://lifeofliveforit.blogspot.no/
Facebook::LiveForIt Software for AmigaOS

@NutsAboutAmiga

Quote:

NutsAboutAmiga wrote: @cdimauro Quote: But you don't need so high specs for a PC which has to emulate an OS4 machine. Trust me, you do, unless something drastically is done to support hardware virtualization techniques in the emulator.

I can't trust you because there are emulators for PowerPC-based consoles which draw circles around the original machines.

The PS3 used a PowerPC core which is roughly equivalent to a 1.6Ghz G5 which is a top notch processor for post-Amiga PowerPC machines.

So, there's nothing that prevents to achieve something similar by emulating OS4 machines.
Quote:

Quote: Now it's even much worse, since GCC already dropped support for those CPUs some time ago. So, OS4 cannot use a modern compiler. It means that this o.s. either should be stuck to the latest version which supports this crappy CPU (thus, hurting the modernization of the platform) or at certain point in time OS4 will go for a newer GCC and A1222 will not be supported anymore. When we are talking about GCC/C or C++ We have like preprocessor, the compiler, and assembler and the linker, and clibs. You can pick one or the other or use part of it, if you wan to you just use preprocessor, for example. of cause if you’re going to pick and choice the build process becomes more complicated. but be good why overcome incompatibility in the preprocessor for example. Most of changes are in clibs, not in the compiler, on 680x0 amiga you using a really old compiler, geektools package, for many projects it does not matter in most cases. For A1222 that different FPU, you have some of CLibs that use compatible integer opcode, and some libs that uses float, that’s not so compatible, in some case building your own libs, can be a good idea for optimal performance.

But the point was/is that you cannot update the GCC compiler if you still want to support the A1222 with its incompatible CPU. Do you agree?
Quote:

The point anyhow the doors are closing on AMCC460 CPU, this why A1222 will be the only option, with exception of the X5040 and ACube-Systems laptop project. In my eyes the laptop project is more interesting then even the X5040 because it’s a different formfactor when we have had before.

And there's nothing else that you could have in future. Unless you go for some POWER processors, which embraced little-endian since some years (IBM officially supports only this endian).
Quote:

NutsAboutAmiga wrote: @matthey I don’t remember exactly, why Warp3D support was dropped, but I believe it was because API was old and outdated, and they wanted new features. They did add a few features into Warp3D, but many old games suffered from bugs, and problems. If don’t remember wrong their issues where that newer graphics cards where so backwards compatible. and got worse with etch newer model. Nova basically starts over from a blank page.

And since it's blank you've to fill it: how many games/applications only support Nova? How many Warp3d? And how many both?

NutsAboutAmiga Quote:

I don’t remember exactly, why Warp3D support was dropped, but I believe it was because API was old and outdated, and they wanted new features.

The Warp3D API may have had some flaws but the primary reason Nova was introduced was that the GPU hardware changed to shaders and much of the API would have been deprecated to support shaders so a new API was introduced instead. Warp3D is a low level API like 3dfx Glide which makes surviving a hardware transition difficult. Even high level APIs like OpenGL did not transition well.

NutsAboutAmiga Quote:

They did add a few features into Warp3D, but many old games suffered from bugs, and problems. If don’t remember wrong their issues where that newer graphics cards where so backwards compatible. and got worse with each newer model.

I don't think the games were as much to fault as compiler and Warp3D bugs and issues. I know because I fixed several of them for the 68k libraries.

Avenger libraries
o wrong stride type calculation for z-buffer writes trash memory big time (Warp3D bug)
o a very common hardware function was so slow that it caused timing issues (likely a compiler issue)

Permedia library
o floating point rounding problems would result in black rendered objects (maybe a compiler issue)

Warp3D.library
o horrible EGCS? code optimization (indirect rendering works better than direct rendering with indirect queue code optimized as it multitasks better)

Vbcc
o Wrong colors in GLQuake fixed by adding FPU initialization

I expect the newer problems are bugs or timing issues as well. The developers should have had Alain Thellier with his Wazp3D debugging features to debug it. Driver code is not easy to debug but we were able to do it. It's likely another case of too small of market to properly develop software. Only Amiga makes it possible.

@cdimauro

Hello guys
(I havent been online for months )

much things to unpack...

hardware acceleration for waZp3D only exist in WinUAE68k, Arosx86, MorphosPPC
It respectively call opengl32.dll, mesa,tinygl

The v5 functions that appeared on OS4 are only implemented in the OS4 version but mostly they do nothing (just update flags,states,etc....) so it allow the OS4 progs to run but multitexturing will not be drawn (more precisely multixture is managed but only first tex is drawn)

There is no point to have v5 functions on 68k as there is no programs using it (multitex) as WaRp3D v5 was never released for 68k
(even if WaZp3D could eventually be made via winuae native calls to (a updated) soft3d.dll and the native opengl)

On WinUAEppc there is no "native interface" to call x86 opengl32.dll so no hardware support (and no hard multitexturing)

Last edited by thellier on 06-Nov-2022 at 10:45 AM.
Last edited by thellier on 06-Nov-2022 at 10:44 AM.
Last edited by thellier on 06-Nov-2022 at 10:42 AM.
Last edited by thellier on 06-Nov-2022 at 10:41 AM.
Last edited by thellier on 06-Nov-2022 at 10:41 AM.

@thellier

Quote:

thellier wrote: @cdimauro Hello guys (I havent been online for months ) much things to unpack... hardware acceleration for waZp3D only exist in WinUAE68k, Arosx86, MorphosPPC It respectively call opengl32.dll, mesa,tinygl The v5 functions that appeared on OS4 are only implemented in the OS4 version but mostly they do nothing (just update flags,states,etc....) so it allow the OS4 progs to run but multitexturing will not be drawn (more precisely multixture is managed but only first tex is drawn) There is no point to have v5 functions on 68k as there is no programs using it (multitex) as WaRp3D v5 was never released for 68k (even if WaZp3D could eventually be made via winuae native calls to (a updated) soft3d.dll and the native opengl) On WinUAEppc there is no "native interface" to call x86 opengl32.dll so no hardware support (and no hard multitexturing)

Many thanks for the punctual clarification, Alan.

This also explains why it doesn't make sense to have a fully v5-compliant implementation of waZp3D for 68k games/applications.

@matthey

Quote:

AMD Ryzen 7 3700X emulates the PPC based Nintendo Wii at 60fps.

That's interesting. Looks easier than other emulation needing a string of QEMU parameters just to get going. Anything that can take on a Sam Flex with a HD7890?

Quote:

While the Wii Broadway CPU only operates at 729MHz and the OoO integer performance is likely the same as the Gekko predecessor at 2.31DMIPS/MHz which may be slightly inferior to the P1022 CPU used in the A1222 "Tabor", there is a full standard PPC FPU and VMX with 50 added 3D SIMD instructions. The Wii GPU is also being emulated. It's obvious that the low end A1222 system could be emulated easily on a "Ryzen". While the X1000 and X5000 would be more challenging to emulate, they mostly gain CPU performance which is easily emulated. I expect an AMD Ryzen 7 3700X based system that costs half of an X5000 system could emulate the X5000 system.

The Broadway looks like a sensible CPU. So do they just emulate the GPU? I would imagine that would be slow and the bruteforce speed of the CPU makes up for it as without GPU drawing operations are slowed down. I thought it may translate the GPU instructions or bytecodes into OpenGL or use Vulcan to assist if it can.

By the time an A1222 comes out it will be so old my X1000 could emulate it and run rings around it.

The X1000 uses DDR2 at 800 FSB while the X5000 uses DDR3 also at 800 FSB I read. X1000 from 2011 CPU at 1.8 Ghz and X5000 from 2017 CPU at 2.0 Ghz slightly faster. Both are over double the Broadway CPU clock and just under triple. So matching the performance would be getting close.

Quote:

Perhaps in response to this thread. Trevor Dickinson wrote the following.

I think it's reasonable what he said in regards to there being a dedicated machine. But the problem is when the dedicated machine has obvious faults in firmware or can easily crash. It gets even worse when OS4 continues to have bugs in the OS or some component that wasn't tested and sits in a broken state. When people pay a premium people expect quality and I think it should be delivered. There's no making excuses about how it cost millions to produce so the end price couldn't match expectations. However in the Amiga world it's customary to sell a below par product because it's Amiga and it would go for 68K products as well. When you buy a Tesla you get a rich mans car powered by a fancy battery. But you don't get a rich mans car that breaks down with bugs all the time because it was worth it to buy it just for this exclusive battery.

What didn't make sense is an audience member asked Trevor Dickinson about non PowerPC hardware. I had to listen a few times as it was at first unclear what he was asking. Then Trevor responded about OS4 being emulated under WinUAE. I didn't understand this. I thought the question was about OS4 running on hardware that wasn't PPC based. Emulation is irrelevant here and makes as much sense as asking what PC motherboard supports the ZZ9000 video card. 😃 Aside from that there is the ALICE project which emulates OS4 and 68K with Rabbit Hole on a PC laptop. While purists can buy a PPC board. Alice? Alice? What the #$*? Is Alice? 🤔 Merged with all this is a silly sounding comment about not wanting OS4 on a PC. Hang on, A/Eon produced the Alice which is Trevor's idea. Then he suggested WinUAE as how OS4 does run on non PowerPC hardware. Did Trevor suddenly change his mind in the space of a few minutes and go from PC guy to purist? Now that is silly!

Quote:

Emulation is the result of noncompetitive hardware. Is an Amiga label on PC hardware with a PPC CPU really an Amiga anymore than the same with an x86-64 CPU? Many people like AmigaOS 4 but the hardware is nowhere close to competitive. We don't need high end hardware that can't compete. Lower end hardware is cheaper, the AmigaOS is more competitive down there and the Amiga user base can be expanded faster. There is demand for inexpensive small footprint computers like the 68k Amiga below where standard Linux distros scale. How about creating Amiga hardware that plays to the Amiga's strengths before the Amiga is nothing but a virtual machine running on PC hardware?

Well, it's not exactly an Amiga, but this AmigaOne contraption that Amiga The Inc came out with whatever it means. But, I would say no. The only Amiga thing about it being able to run Amiga code natively that is a a direct descendant of the original code. Even before then people plugged PC cards in with a Mediator and to me that was no less Amiga than you could get. The AmigaOne dumped the Amiga but kept the cards!

However, as expensive as they may be, they are still cheaper what they could have cost. Imagine if every card inside was a custom designed card. For example, consider the ZZ9000 at €399.00. Someone come along and ask why they'd spend so much when they can get a better PC video card for less. Except at todays prices really. And technically they'd be right. I don't even see 3d listed as a feature which even Amiga cards had 20 years ago so what's going on? Now the ZZ9000 is a multi function device. With Ethernet and USB. So people can go on, does it have Gigabit Ethernet? It it USB 2 and 3 with DMA? If we can criticise the expensive AmigaOne boards surely we can criticise expensive 68K expansion boards? After all, both are selling expensive products to a niche market, where they are simply duplicating what already exists in the commodity market in an inferior form.

Madness!