@Hammer

I don't need your link as I know what is happening but thank you.
YOU guys are the ones clogging up this thread with useless information and bickering

Just stop and leave this thread for news about the A1222.
Thank you

@outlawal2

So there are new videos demonstrating the SPE FPU emulation/optimisation then? Please share with us, as there is no other marketing to get hyped about!

_________________
"Art challenges technology. Technology inspires the art."
John Lasseter, Co-Founder of Pixar Animation Studios

What is not always taken in consideration in "future investment" is that P1022 CPU in A1222 is dual core. This is something that will instantly outperform other single core Amigas when released for AmigaOS4. Even very generic multicore implementation where some task is on first core and the other on the second can boost things considerably. Its not here yet but it is in hardware and ready to use.

@dooz

That makes sense, if that is the direction of the CPU you are going. Thanks for explaining that!

How close is the dual core support to release for AOS4?

@Matt3k

Quote:

What is not always taken in consideration in "future investment" is that P1022 CPU in A1222 is dual core. This is something that will instantly outperform other single core Amigas when released for AmigaOS4. Even very generic multicore implementation where some task is on first core and the other on the second can boost things considerably. Its not here yet but it is in hardware and ready to use.

That's certainly true, but I'd caution folks on buying anything Amiga-related based on 'what will be' rather than 'what is' at the time of shipment. Make sure the A1222 (or SAM460 for that matter) makes sense for you based on what it can do at the time of purchase to avoid disappointment if promised features never make it for whatever reason.

Thankfully all the HW is supported on the A1222 unlike the X1000 at launch-time, but the FPU situation is something folks need to understand and be OK with; perhaps multicore support will come, and that'll be great. But we don't have it now and there has been no public demonstration of it working, either.

I love my A1222, and I think others will too, but I can also see where folks might not.

By the way this is a great reason to visit AmiWest or Amiga38 this year. You can try out either system for yourself. That way you can make up your own mind rather than being bombarded with the opinions of the same six guys on AWN arguing over the same hogwash they've flooded every AOS4-related thread with for the past five years.

I don't suppose you'd be interested in taking the trip out to Cali this autumn?

-- eliyahu

_________________
"Physical reality is consistent with universal laws. When the laws do not operate, there is no reality. All of this is unreal."

@eliyahu

Seems like solid advice. Nice that you are enjoying yours!

What is the issue with the FPU?

Thanks for explaining.

@Matt3k

it has something else instead.

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

@Matt3k

Han's gives a good explanation.

Link.

@Matt3k

Quote:

What is the issue with the FPU?

I have a high-level explanation in my guide:

http://eliyahu.org/tabor/intro.html

-- eliyahu

_________________
"Physical reality is consistent with universal laws. When the laws do not operate, there is no reality. All of this is unreal."

@Matt3k

Quote:

What is the issue with the FPU?

It is not really a FPU and it is not compatible with anything else.
You have two options:

* Software trap and emulation of the FPU instructions which is enormously slow. So slow it makes a SAM440 look like a speed monster.

* Recompile all applications that use FPU. Great, now all software needs to be distributed in multiple versions. A normal version and a normal a1222 version. If you don't have the source code and can not recompile the software for the a1222 then, well I guess shit will suck.

Such is daily life in NG world.

@V8

Shouldn’t shit stink instead of sucking? I guess if you had a straw it would suck too?

_________________
Sent from my Quantum Computer.

Who gives a shit about those lame Intel fanboys anyway.

@V8 plus A-EON, AAA ACube or anyone else who cares!

Quote:

It is not really a FPU and it is not compatible with anything else.

I suppose FPUs are over-rated and SPEs created no issues for the PS3 whatsoever with everything working, simple to code for as expected and on par with the XBox 360! -

We need some real world tests because obviously SPEs HAVE created issues in the past for the PS3, FPUs ARE an important requirement on the Amiga ever since the 030 in my view and the A1222+ IS now too expensive to justify an impulse buy on something that may be substandard in real world tests!

SO SHOW US ALL THIS TRAPPING AND OPTIMISING ACTUALLY WORKS OR WHY WOULD ANYONE RISK IT BEING A BIG DISAPPOINTMENT?

Last edited by BigD on 18-Jul-2023 at 11:26 PM.
Last edited by BigD on 18-Jul-2023 at 11:15 PM.

_________________
"Art challenges technology. Technology inspires the art."
John Lasseter, Co-Founder of Pixar Animation Studios

@BigD

I’ll be super excited when the Amiga X12000 comes out. Almost 3X better than X5000!

:)

_________________
Sent from my Quantum Computer.

@BigD

Quote:

SO SHOW US ALL THIS TRAPPING AND OPTIMISING ACTUALLY WORKS OR WHY WOULD ANYONE RISK IT BEING A BIG DISAPPOINTMENT?

_________________
Fulfill newlight's Elite Narcissist Demands NOW Or He Will Send You To H3LL!

A ten year old PPC with incompatible FPU requiring something conceptually similar to the trap and emulate that crippled FPU requiring applications on 040 and 060 and was only really remedied with patching. I wonder how this emulation works this time.

The FPU is needed more for PPC software than it was on 68K if for no other reason than the fact all (or almost all, I'm not aware od any exceptions) PPC ever used in Amiga and NG had them as standard, and they've generally been quite performant, even the humble 603e.

_________________
Doing stupid things for fun...

@V8

Quote:

V8 wrote: @Matt3k Quote: What is the issue with the FPU? It is not really a FPU and it is not compatible with anything else. You have two options: * Software trap and emulation of the FPU instructions which is enormously slow. So slow it makes a SAM440 look like a speed monster. * Recompile all applications that use FPU. Great, now all software needs to be distributed in multiple versions. A normal version and a normal a1222 version. If you don't have the source code and can not recompile the software for the a1222 then, well I guess shit will suck. Such is daily life in NG world.

The e500v2's custom FPU is a dead-end path since the successor T1024's e5500 CPU cores have a PowerPC's standard scalar double-precision FP64 FPU. It's PS3's CELL SPE time-wasting activity.

PowerPC e5500 CPU has a four-instruction, out-of-order pipeline with a double-precision FPU, and three integer units. On scalar code, the e5500 CPU should rival ARM Cortex A78's four-instruction issue and out-of-order pipeline processing.

PS; ARM Cortex A78 has 128-bit NEON SIMD support like A53 and A72.

_________________
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)

@Karlos

Quote:

Karlos wrote: A ten year old PPC with incompatible FPU requiring something conceptually similar to the trap and emulate that crippled FPU requiring applications on 040 and 060 and was only really remedied with patching. I wonder how this emulation works this time. The FPU is needed more for PPC software than it was on 68K if for no other reason than the fact all (or almost all, I'm not aware od any exceptions) PPC ever used in Amiga and NG had them as standard, and they've generally been quite performant, even the humble 603e.

Unlike other sensible CPUs, e500v2's custom FPU shares the same registers as general-purpose registers (GPRs).

FPU emulation on e500v2 will require GPR contents to be swapped out, hence incurring additional context switch overheads.

The main purpose of hardware simultaneous multithreading (SMT) is to reduce context switch overheads and saturate the pipelines.

68060 FPUs (8 80-bit FP data registers + 3 special purpose FP registers) have their own FP registers that are separated from GPRs.

_________________
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)

@BigD

Quote:

BigD wrote: @V8 plus A-EON, AAA ACube or anyone else who cares! Quote: It is not really a FPU and it is not compatible with anything else. I suppose FPUs are over-rated and SPEs created no issues for the PS3 whatsoever with everything working, simple to code for as expected and on par with the XBox 360! - We need some real world tests because obviously SPEs HAVE created issues in the past for the PS3, FPUs ARE an important requirement on the Amiga ever since the 030 in my view and the A1222+ IS now too expensive to justify an impulse buy on something that may be substandard in real world tests! SO SHOW US ALL THIS TRAPPING AND OPTIMISING ACTUALLY WORKS OR WHY WOULD ANYONE RISK IT BEING A BIG DISAPPOINTMENT?

There is a vast swath of tasks the CELL SPE
1) cannot do (software permissions, various interrupts, running the privileged software that handles memory mapping)
2) cannot do very well (anything needing a context switch, tons of branches, a lot of synchronization)

One of the SPEs is reserved for the use of the OS to provide certain services but an SPE is not capable of running a modern OS on its own. SPEs do not support different privilege modes for code (on a modern OS the kernel runs at a higher privilege level than user code), they do not support any kind of virtual memory or memory protection, they have very limited support for interrupts and they do not have full access to hardware for I/O.

The SPU has the following restrictions:
• No direct access to main storage (access to main storage using MFC facilities only)
• No distinction between user mode and privileged state
• No access to critical system control such as page-table entries (this restriction should be enforced by PPE-privileged software).
• No synchronization facilities for shared local storage access


-------------------
Meanwhile, Xbox 360's Xenos GpGPU can 64b pointer swap with PPE CPUs. Xenos GpGPU has 8 contexts async compute front end and in addition to sync graphics command front-end. It's effectively proto-GCN. Radeon HD7970 doubled 8 contexts async compute front end count.

Xbox 360's Xenos ROPS is capable of performing like-DirectX12_1 Rasterizer Ordered Views (ROVs), hence feature mapped with DirectX12_1 ROV when used with Xenia emulator.

Xbox One's GCN GPU and Xbox Series X's RDNA 2 GPU have no problems running XBox 360 GPU FP for backward compatibility with higher performance.

Last edited by Hammer on 19-Jul-2023 at 01:57 AM.

_________________
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)

@toRus

Quote:

toRus wrote: Who gives a shit about those lame Intel fanboys anyway.

FYI, my user name was named after AMD's K8 codename i.e. Sledgehammer.

Sledgehammering PowerPC bots is my favorite pastime.

_________________
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)

@ppcamiga1

Quote:

ppcamiga1 wrote: @Hammer stop trolling

Stop lying.

Against https://www.cucug.org/amiga/aminews/1995/at951111.html

Under Petro Tyschtschenko's administration, official AmigaOS 3.1 wasn't ported to PowerPC unless you count Phase 5's 1997 PowerUP or Haage & Partner's 1997 WarpOS.
Escom was bankrupt in July 1996.

Phase 5's 1997 PowerUP is not AmigaOS PowerPC.
Haage & Partner's WarpOS is not AmigaOS PowerPC.
1997 WarpOS was launched as a controversial alternative to Phase5's PowerUP kernel.

https://www.osnews.com/story/11034/how-qnx-failed-the-amiga/

Back in 1997, the QNX (pronounced either Q-N-X or Q-nix) operating system was positioned to play a vital role in the ongoing development of the Amiga platform. Georgios Panayio provides historical background about the proposed relationship between the two projects and shows how QNX’s alleged breach of contract resulted in causing significant damage to the Amiga platform.


https://www.cucug.org/sr/sr0108.html
Amiga Inc`s CEO Bill McEwen announced in a keynote speech held prior to the AmiWest 2001 show that new consumer PPC motherboards, AmigaOS 4.0 PPC and also AmigaOS for x86 will ship by November 1.

Legal issues between Haage & Partner and Amiga Inc wreaked AmigaOS X86 projects (Amithlon and QNX-based AmigaOS XL).

bPlan/Thendic-France's MorphOS project began in 1999. Thendic-France had financial problems and folded, but the collaboration continued under the new banner of "Genesi".

Hyperion Entertainment was founded in April 1999.

In 2001, Hyperion announced the licensing of the rights from Amiga, Inc, it would be working on the long-awaited successor to AmigaOS 3.9, and to this end concentrated most of its effort on the development of AmigaOS.

AmigaOS 4.0 was first released to end-users and second-level betatesters in April 2004

--------------
There's an unofficial split between Hyperion Entertainment's AmigaOS 4.1 FE and A-EON System54 (A-EON "Enhancer Software") e.g. https://www.amigans.net/modules/newbb/viewtopic.php?post_id=128489#forumpost128489

Hyperion Entertainment's recent software releases are related to the larger market 68K AmigaOS 3.2.x.

Last edited by Hammer on 19-Jul-2023 at 03:08 AM.
Last edited by Hammer on 19-Jul-2023 at 02:46 AM.
Last edited by Hammer on 19-Jul-2023 at 02:37 AM.
Last edited by Hammer on 19-Jul-2023 at 02:30 AM.

_________________
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)