@Hammer, @matthey

Can you guys please start a completely new thread where you will discuss that? You both seem to fall into the very same topic in every thread ;)

michalsc Quote:

Can you guys please start a completely new thread where you will discuss that? You both seem to fall into the very same topic in every thread ;)

I resisted responding to Hammer trying to bait me into endless arguments I have already refuted. The superscalar in-order 68060 was a powerful and successful CPU despite the embedded treatment and 50MHz maximum political limitation. On topic, I would like to see more performance and compatibility oriented 68060 cores on more modern silicon. I would like to see weak in-order ARM CPU cores replaced by powerful in-order 68k cores in Amiga hardware. I would like to see interrupt and DMA driven 68k Amiga hardware rather than constant 100% CPU usage emulation. I would like to see a hardware target that SMP could be enabled which is not emulation or PPC. I know it is too much to ask to have modernized 68k Amiga hardware that could be produced per chip for less that a meal. Only Amiga and Amiga mental cases make it possible.

@matthey

That's BS. I have 68060 rev1 and 68LC060 rev4 for my clock speed programmable TF1260, and they have the same integer pipeline stages with different high-clock speed attainment outcomes.

Different stepping refinements can influence clock speed attainment, NOT just pipeline stage count. The same 68060 refinements debunked your simplistic pipeline stage count argument. Clock speed attainment is also influenced by the slowest section in the CPU core.

Again, clock speed attainment is also influenced by the slowest section in the CPU core i.e. you keep ignoring the FPU pipeline stage count. There are good reasons for Alpha 21064's high clock speed and a longer FPU pipeline.

-----

Alpha 21064's FPU is superscalar i.e. dual port FPU pipelines with separate FADD and FMUL (hint: similar higher level layout for AMD K7 Athlon's to Jaguar's dual ports for FADD and FMUL pipelines).

In modern times, SiFive U74 has a single port FPU vs ARM Cortex 53/55's dual port symmetric FPU pipelines. I already debunk your StarFive U74 propaganda with fatter Linux-based benchmarks.

Intel's semi-modern dual-issue in-order CPU design is Atom's Bonnell with SMT and SSE3 dual port FPU/SIMD units i.e. separate FADD and FMUL pipeline basic layout design. Intel's latest e-Core is Skymont with four SIMD pipelines.

-----

For the 600 nm process node, 1994 68060 rev1 already lost to 1994 Pentium P54C.

6 bytes instructions on 68060 are slower than 68040 while Pentium P5 is faster than 486DX. There's a design regression with 68060 when compared to 68040.

Both Pentium P5 66Mhz and 68040 40Mhz use the same 800 nm process node and Motorola lost the 800 nm process node generation. There's a good reason why Apple's Quadra 840AV included AT&T DSP3210 66Mhz (33 MFLOPS IEEE FP32) in 1993 when 68060 was missing in action i.e. late.

Motorola lost the clock speed race for 68040 @ 40Mhz (there's an unofficial 50Mhz OC 68040 from the Mac world) vs 80486DX4 @ 100Mhz. AMD released an even higher clocked enhanced 486DX as 5x86-P100 @ 160Mhz. This is for +1 million transistor level CPU budgets.

Motorola lost the single pipeline CPU release race for 68040's 1990 vs 80486DX's 1989. Motorola and Apple system integration wasn't as tight as Intel and Compaq i.e. the same 1989 year release vs the 1st 68040 system from Apple was in October 1991.

You can't directly compare RISC vs the CISC-RISC pipeline stage since the CISC ISA has an extra complex decoder stage e.g. Alpha 21064 doesn't have the CISC decoder stage.

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

The late 1990s Ghz race is with Alpha 21364 EV7 vs Pentium III vs K7 Athlon participants, not from Motorola/Freescale! AMD used a licensed Motorola copper interconnect process node with K7 Athlon's higher clock speed outcomes when compared to Motorola PowerPC G4+ counterpart.

High-clock speed design is developed by skilled engineers, not just a process node.

Unlike MIPS and RISC-V which had US government R&D funding, there are near zero US national security arguments for 68K. MIPS Inc. has moved to RISC-V.

ARM has national security arguments for the CPTPP members' UK and Japan.

RISC-V ISA is good for semiconductor companies with weak CPU instruction set ownership. Good RISC-V implementation is not free and needs to be licensed.

My NVIDIA RTX GPUs has a custom RISC-V microcontroller and it's meaningless for most desktop end users.

NVIDIA RTX GPUs can considered to be an APU when the custom RISC-V CPU core is exposed, but NVIDIA doesn't want the custom RISC-V CPU to be exposed since it can distract NVIDIA's focused support resources. NVIDIA CEO already made his statement against against RISC-V as an application CPU i.e. He dislikes RISC-V ecosystem fragmentation due to lack of ecosystem leadership.

Last edited by Hammer on 06-Feb-2025 at 01:29 AM.
Last edited by Hammer on 06-Feb-2025 at 01:28 AM.

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7950X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@thread

If we're discussing the kinds of Amiga products/projects we'd like to see, beyond the 5 options listed in the original post, then what I would like to see is for the active community to finally let go of the past and focus on the present. This means:
1) caring less about running archaic software designed and written decades ago for for sub-100MHz 68k silicon, and focus on porting/writing software for the 68k+ category.
2) caring less about proprietary Amiga OS 3.x and AmigaOS 4.x operating systems, and focusing more on AROS 68k, AROS x64, and AxRT.
3) spending less money on go-nowhere PPC software, and focusing more on supporting initiatives and bounties related to item 2, supporting item 1.
4) as I've mentioned earlier in this thread, an adapted port of Odyssey, OWB or even Wanderer to 68k+

The 2025 culmination of this focus on the present would take form of a Standalone emu68 System based on Raspberry Pi 5, running AROS 68k with 2D RTG drivers and 3D Waxp-compatible drivers, wired and wireless networking support, USB support for HID, Mass Storage, Audio/Video, and Hub classes, and pass-through access to the other ARM cores for acceleration of specific sub-systems.

Those who care about living in the past can work on enabling Amiga machine-level emulation on this new device, a la RunInUAE.

_________________
All the way, with 68k

@michalsc

Matthey is knee-deep in his reality distortion field.

No Raspberry Pi size company will fund the reborn 1Ghz 68060.

Last edited by Hammer on 06-Feb-2025 at 05:27 AM.
Last edited by Hammer on 06-Feb-2025 at 01:36 AM.

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7950X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@agami

Quote:

The 2025 culmination of this focus on the present would take form of a Standalone emu68 System based on Raspberry Pi 5, running AROS 68k with 2D RTG drivers and 3D Waxp-compatible drivers, wired and wireless networking support, USB support for HID, Mass Storage, Audio/Video, and Hub classes, and pass-through access to the other ARM cores for acceleration of specific sub-systems.

For backwards compatibility reasons, AROS 68K needs additional work to replace AmigaOS 3.9 based CoffeineOS.

https://eab.abime.net/showthread.php?t=118450
AROS for real Amigas and their issues.

Last edited by Hammer on 06-Feb-2025 at 02:09 AM.
Last edited by Hammer on 06-Feb-2025 at 02:07 AM.

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7950X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@Hammer

Quote:

to replace AmigaOS 3.9 based CoffeineOS

Since when did that become the benchmark?

Last edited by kolla on 06-Feb-2025 at 03:00 PM.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@agami
So leave the Amiga past and software behind to develop for the "true" NG AmigaOS?

AROS x86-64
AROS 68k
AxRT
AmigaOS 4
MorphOS
AmigaOS 3.9
AmigaOS 3.2
A-EonKit enhancer OS
AmiKit
Coffin OS
WarpOS
AFA OS

We are supposed to pick the winners and discard the losers from this list, dividing the Amiga market further? Is it so difficult to see that uniting is the answer instead of dividing? Is it possible to have a better NG Amiga while retaining compatibility so reuniting is possible?

Hammer Quote:

Matthey is knee-deep in his reality distortion field. No Raspberry Pi size company will fund the reborn 1Ghz 68060.

The Raspberry Pi Foundation was financed by 5 investors and one was Eben Upton's wife. The initial goal was to sell 1000 SBCs for educational use. The original 2012 RPi used a single 32-bit scalar in-order CPU core and the 2015 RPi 2 a 4 core 32-bit in-order superscalar CPU while the 1994 68060 CPU is a 32-bit superscalar in-order CPU. The Raspberry Pi Foundation has become a fabless semiconductor business and the Raspberry Pi Holdings IPO (ticker RPI) has a market cap of 1.48B GBP.

https://www.google.com/finance/quote/RPI:LON

Incidentally, RPI recently received a downgrade based on valuation, a concern I expressed in an earlier post.

https://www.investing.com/news/stock-market-news/jefferies-downgrades-raspberry-pi-to-hold-citing-full-valuation-concerns-3850794

Both the RPi and X1000 were released in 2012 and we can see the tale of two businesses since. RPi Foundation/Holdings is a very successful business worth 1.48B GBP and A-EonKit is an abject failure and endless money pit kept alive by endless cash injections in one form or another. The primary difference is as simple as RPi using competitive hardware and aggressively pushing down pricing and A-EonKit hardware being nowhere close to competitive while falling further and further behind the technology curve and protecting their hardware with lawsuits and software which is backfiring as potential customers realize their tactics and turn away.

Most businesses require more capital to get started and become competitive than the RPi Foundation/Holdings. It is possible they successfully entered the low end embedded, hobby and educational computer hardware markets with hundreds of thousands of GBP in financing. Entering the desktop and console markets are crazy expensive in comparison. Microsoft's XBox division lost $4 billion USD in 4 years to launch the original XBox and gain console market share. The Microsoft dominated desktop market would likely require higher spending than this to gain single digit market share but A-Eon launched a desktop computer with noncompetitive hardware and AmigaOS anyway. PPC hardware was noncompetitive when it launched around the time Steve Jobs was abandoning PPC for the desktop because it was not competitive but the silicon is 10 years further behind the tech curve today. There is still no SMP or process isolation after more than 10 years. Millions USD have likely been malinvested in an obvious PPC AmigaNOne "reality distortion field". The micro-business Retro Games Limited tried to get Jeri Ellsworth to create an Amiga ASIC which would have lowered the cost and price of THEA500 Mini distancing it from the ARM based emulation competition and likely resulting in hundreds of thousands of additional unit sales. I find this reasonable but can understand why uncertainty cause by A-EonKit and Hyperion challenging Amiga IP ownership made this less appealing. A few million USD cost for an ASIC is possible for small businesses and even investigated by micro-businesses. The cost of noncompetitive hardware can be much higher as A-EonKit demonstrates with their shenanigans used to protect their noncompetitive hardware. The RPi Foundation is as open as possible with their software and hardware even pushing Broadcom to release GPU documentation. Yep, the tale of two businesses, one successful and open with competitive hardware and the other unsuccessful and protective of noncompetitive hardware.

@kolla

Quote:

kolla wrote: @Hammer Quote: to replace AmigaOS 3.9 based CoffeineOS Since when did that become the benchmark?

C= AmigaOS is the benchmark for the Amiga.

Last edited by Hammer on 06-Feb-2025 at 10:11 PM.
Last edited by Hammer on 06-Feb-2025 at 10:10 PM.

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7950X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@matthey

Quote:

AROS x86-64 AROS 68k AxRT AmigaOS 4 MorphOS AmigaOS 3.9 AmigaOS 3.2 A-EonKit enhancer OS AmiKit Coffin OS WarpOS AFA OS

Add Cloanto's AmigaOS 3.X and Kickstart 3.X ROM distribution. Cloanto and Amiga Corp is owned by Michael Battilana.

Haage & Partner's AmigaOS 3.5/3.9 includes WarpOS. A wrapper was made for Hyperion Entertainment's AmigaOS 4.0 & 4.1 project. Hyperion Entertainment doesn't own the core parts of AmigaOS 4.x.

MorphOS also uses a wrapper to run WarpUP programs, they also have a wrapper for PowerUP, a WarpOS competitor.

Quote:

We are supposed to pick the winners and discard the losers from this list, dividing the Amiga market further? Is it so difficult to see that uniting is the answer instead of dividing? Is it possible to have a better NG Amiga while retaining compatibility so reuniting is possible?

The only common ground is a level of 68K AmigaOS 3.x compatibility.

Not including licensed 68K AmigaOS 3.1.3 and 3.2, each direction has a smaller native software library.

WarpOS was intended to be used as a basis for AmigaOS 4 but Haage & Partner dropped the project when their AmigaOS 4 PPC contract was canceled by Bill McEwen's Amiga, Inc. in 2000.

There was disagreement between Bill McEwen's Amiga, Inc and Haage & Partner's support for Amithlon i.e. Haage & Partner is not licensed for AmigaOS x86. Bill McEwen went "Gary Kildall" of Digital Research Inc. demanding a GEM port license fee, locking out PC clone vendors, and supporting only genuine IBM PCs, Compaq (and PC clones) responded by throwing their full support behind MS Windows 2.x 386 and MS/SCO Xenix 386 R&D. Amithlon was withdrawn from the market and AmigaOS x86 pathway died at the hands of Bill McEwen.

Cloanto's Amiga Forever has its own bootable Linux-UAE-AmigaOS 3.X distro.


Background: Angry Compaq effectively killed Gary Kildall's GEM X86. PC clones followed Compaq's 386AT standard over IBM's PS/2 standard. Compaq being the 1st 386 PC release is due to a tight relationship with Intel and Microsoft. With the Gang of Nine PC clones, Intel and Microsoft backed stabs IBM. Bill Gates is aware IBM's OS/2 project attempts to minimize Microsoft's role in the PC market. Intel and Bill Gates are also aware of IBM's attempts to slow down the PC's 32bit 386 uptake since IBM has a self-interest with the IBM RT PC (released in Jan 1986) which is the precursor to POWER1/PowerPC. IBM RT PC's ROMP CPU is 32-bit RISC. IBM RT PC's ROMP CPU has 16 32-bit GPRs and uses IBM's 2 micron NMOS process node.



Quote:

The Raspberry Pi Foundation was financed by 5 investors and one was Eben Upton's wife. The initial goal was to sell 1000 SBCs for educational use. The original 2012 RPi used a single 32-bit scalar in-order CPU core and the 2015 RPi 2 a 4 core 32-bit in-order superscalar CPU while the 1994 68060 CPU is a 32-bit superscalar in-order CPU.

That's a red herring for the current RPI with $0.5 billion revenue base that started to design its own ARM-licensed copy-and-paste SoC and commissioned its own TSMC wafer starts.

https://core-electronics.com.au/guides/raspberry-pi-where-it-began/
Upton continued his pursuit to revive interest in computer science and in early 2009 the Raspberry Pi Foundation was assembled. Consisting of Eben Upton, Pete Lomas (Norcott Technologies), David Braben (Elite, Frontier, BBC Micro), and Cambridge University lecturers Jack Lang, Dr Rob Mullins and Prof Alan Mycroft.

Over the next 4 years, an intensive development cycle began. With no official team of staff or large-scale infrastructure, work began with just a handful of volunteers working out of their home. With this team of individuals from the University of Cambridge designing and Broadcom manufacturing, the Raspberry Pi finally had a powerhouse behind it to get moving.

Progress was slow in the early years of development as sticking to delivering a board capable of being a useful educational tool while keeping the costing around $30 (USD) was a difficult goal.

Luckily in early 2011 a design for a low-cost processor developed by Broadcom, Upton’s employer at the time, had been produced. This ARM 11 chip was squeezed into the Broadcom BCM2835 system-on-chip, combined with onboard graphics and RAM. With this low-cost processing strength, the Raspberry Pi would be capable of running a desktop version of the opensource Linux operating systems that had been developed over the previous years.


Upton’s employer at the time was Broadcom. Upton’s day job was paid by Broadcom.


https://en.wikipedia.org/wiki/ARM11
ARM 11 CPU powered the following SoCs

Ambarella A5s, A7, A7L
ASPEED Technology Inc. AST25xx
Broadcom BCM2835 (Raspberry Pi 1 A/B, Pi Zero), BCM21553
Cavium ECONA CNS3000 series[9]
CSR Quatro 4230, 45xx, 53xx
Freescale Semiconductor i.MX3x series, such as i.MX31, i.MX35
Infotmic IMAPX2xx
Nintendo CTR-CPU (Nintendo 3DS, New Nintendo 3DS)
NTC Module 1879VYa1Ya, K1879KhB1Ya, 1879KhK1Ya, K1888VS018
Nvidia Tegra
MediaTek MT6276, MT6573
Mindspeed Comcerto 1000 (Freescale LS102MA)
Philips Semiconductor/NXP/ST-NXP Wireless Nomadik STn8820
PLX Technology NAS782x
Qualcomm MSM720x, MSM7x27
Qualcomm Atheros AR7400
Samsung S3C64xx, S5P64xx, S5L87xx, S5L89xx or Exynos Dual with Logic11
Telechips TCC8902
Texas Instruments OMAP2 series, with a TMS320 C55x or C64x DSP as a second core
iPhone 3G series, with a Samsung ARM 1176JZ chip
Xcometic KVM2800


A significant number of ARM11 based SoC powered handheld smart devices e.g. mobile phones.

ARM11 supports
1. Java Decode,

2. ARMv6 SIMD instruction set that supportst 8 and 16-bit SIMD arithmetic, including four 8-bit and two 16-bit operations, parallel add and subtract, selection, packing and unpacking.

3. 64-bit (8 byte) path allows two instructions to be fetched from L1 cache in one cycle.

4. Out-of-order completion.

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

68060's 4 bytes fetch from L1 cache only allows for two 2 byte instructions fetch per cycle and performance degradation for larger instruction length.

ARM 11's single instruction issue includes 32-bit SIMD pack-math i.e. four 8bit or two 16bit data processing).

ARM9 and ARM11 pushed out 68000 DragonBall VZ from handheld smart device market.


Quote:

Most businesses require more capital to get started and become competitive than the RPi Foundation/Holdings. It is possible they successfully entered the low end embedded, hobby and educational computer hardware markets with hundreds of thousands of GBP in financing. Entering the desktop and console markets are crazy expensive in comparison. Microsoft's XBox division lost $4 billion USD in 4 years to launch the original XBox and gain console market share. The Microsoft dominated desktop market would likely require higher spending than this to gain single digit market share but A-Eon launched a desktop computer with noncompetitive hardware and AmigaOS anyway.

The original Xbox was code-named Project Midway which refers to WWII's Battle of Midway and that's the intention for the Xbox project.

Xbox is DirectXbox with DRM enforcement and platform access fees i.e. Nintendo business model.

Sony's original PlayStation R&D was subsidized by other Sony divisions.

MIPS-X CPU and ISA design was subsidized by the US taxpayer via US universities and DARPA. MIPS Inc. was created to commercialize the MIPS R&D. With the lesson from ARM, RISC-V ISA was designed by US taxpayer-paid US universities, and a similar progress as the MIP adventure was executed.


Quote:

PPC hardware was noncompetitive when it launched around the time Steve Jobs was abandoning PPC for the desktop because it was not competitive but the silicon is 10 years further behind the tech curve today.

The promise of 3 Ghz, "performance per watt" and absolute performance factors relative to Intel Core 2. PowerPC 970's CPC925 power-consuming bloat didn't help PPC's cause. CPC925 was mostly designed by Apple and it had itself to blame.

Apple's selection for Intel's Core 2 was a road map vs road map issue.

PowerPC wasn't the only victim of Intel Core 2 i.e. AMD K8 was pushed into the budget segment.

Intel was guilty of anti-competitive practices with PC clone vendors.

Quote:

There is still no SMP or process isolation after more than 10 years. Millions USD have likely been malinvested in an obvious PPC AmigaNOne "reality distortion field".

You have your own "reality distortion field".

Neo-Amiga PPC follows the Mac 68K-to-PPC's 1994 era transition approach and it was frozen in time when A-EON entered the scene. A problem, Amiga is not a Mac.

Quote:

The micro-business Retro Games Limited tried to get Jeri Ellsworth to create an Amiga ASIC which would have lowered the cost and price of THEA500 Mini distancing it from the ARM based emulation competition and likely resulting in hundreds of thousands of additional unit sales.

That's not the ASIC 68060 reborn project.

Rochester Electronics has a 68040 license, $77 million annual revenue range and 677 employees strength


From Jeri Ellsworth's story on A500 and Williams home arcade ASICs https://www.youtube.com/watch?v=5uaDzF99a80
A500 ASIC was cancelled and Williams home arcade ASIC was a sales flop.


https://en.wikipedia.org/wiki/C64_Direct-to-TV
Jeri Ellsworth's C64 Direct-to-TV custom AISC runs at 32 Mhz. The project is backed by Ironstone Partners, DC Studios, and Mammoth Toys. QVC purchased the entire first production run of 250,000 units and sold 70,000 of them on the first day that they were offered. There is a sales guarantor and the risk is on QVC.

C64 Direct-to-TV custom AISC is from Atmel's Application-Specific Integrated Circuit product. https://en.wikipedia.org/wiki/Atmel

https://www.c64-wiki.com/wiki/C64DTV

Atmel Application-Specific Integrated Circuit (ASIC).
Simulates MOS Technology's 6510, VIC-II, SID, CIA, and PLA.
System ASIC clock running at 32 MHz.
Emulated 6510 clock running at 1 MHz.



https://thec64community.online/thread/426/inside-thec64-maxi
The units were sold from 26 November 2004 on the QVC channel for USD 25. Including both NTSC version 1 and PAL versions 2 and 3, the total number produced was 600,000 units.

https://youtu.be/3fdAX42BHS0?t=468
C64DTV's game selector UI

vs

https://youtu.be/GXXCj5kqPcM?t=371
THEC64Mini's modern game selector GUI. Retro Games Limited has considered to use C64DTV chip but the main concern is HDMI output and it's target audience.

---------------
Retro Games Limited's THEA500 Mini has up to AGA and RTG compatibility and a Linux GUI front end for the modern audience.

THEA500 Mini has Allwinner H6 (sun50iw6p1) SoC with a Quad-Core Cortex-A53 ARM CPU, and a Mali-T720 MP2 GPU from ARM. Allwinner H6 selection is reducing investment risk.

THEA500 Mini can be repurposed for other ARM Linux-related projects, hence it's not stuck with just the Amiga. ARM Linux market is larger than 68K Amiga.

https://www.youtube.com/watch?v=Ib6AMVXCGtc
The A500 Mini running Workbench 3.9 with RTG and 68K SysInfo 237 MIPS.

https://x.com/Claude1079/status/1878319994129223970
Claude Schwarz's PiStorm with Till Harbaum's cheap Nanomig (Amiga ECS FPGA), a hybrid approach. For a given ARM CPU, baremetal Emu68 is faster than ARM--Linux UAE.

Last edited by Hammer on 09-Feb-2025 at 12:45 AM.
Last edited by Hammer on 09-Feb-2025 at 12:44 AM.
Last edited by Hammer on 07-Feb-2025 at 04:52 AM.
Last edited by Hammer on 07-Feb-2025 at 04:27 AM.
Last edited by Hammer on 07-Feb-2025 at 03:25 AM.
Last edited by Hammer on 07-Feb-2025 at 03:06 AM.
Last edited by Hammer on 07-Feb-2025 at 02:56 AM.
Last edited by Hammer on 07-Feb-2025 at 01:48 AM.
Last edited by Hammer on 07-Feb-2025 at 01:36 AM.
Last edited by Hammer on 07-Feb-2025 at 01:33 AM.
Last edited by Hammer on 07-Feb-2025 at 01:26 AM.
Last edited by Hammer on 07-Feb-2025 at 12:40 AM.
Last edited by Hammer on 07-Feb-2025 at 12:29 AM.
Last edited by Hammer on 06-Feb-2025 at 11:28 PM.
Last edited by Hammer on 06-Feb-2025 at 10:50 PM.

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7950X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@matthey

Quote:

matthey wrote: @agami So leave the Amiga past and software behind to develop for the "true" NG AmigaOS? AROS x86-64 AROS 68k AxRT AmigaOS 4 MorphOS AmigaOS 3.9 AmigaOS 3.2 A-EonKit enhancer OS AmiKit Coffin OS WarpOS AFA OS We are supposed to pick the winners and discard the losers from this list, dividing the Amiga market further? Is it so difficult to see that uniting is the answer instead of dividing? Is it possible to have a better NG Amiga while retaining compatibility so reuniting is possible?

Yes.

When Cortez reached the new world, he burned his ships.

And it wouldn't be dividing what is already a pretty divided ecosystem as much as filtering out the proverbial noise for a clearer signal.
Our problem is that we have too many options. Our second biggest problem is that many of those options are proprietary and are being dragged down by stupid decisions, and infighting.

So I say it's time for a new Amithlon/Umilator running the open source Amiga API compatible AROS. I would happily accept a x64 incarnation, but all the activity is happening over in the sub-$200 ARM camp with emu68, and later this year the new 3D driver.

There's no feeling quite like sticking it it to "The Man".

_________________
All the way, with 68k

The problems are multifaceted with so many options and the progression of time...

There is almost zero money or potential in Amiga for building a OS and applications or hardware.

Endeavors requires a talented team as full as possible that controls the os and application software components.

The Team would have to be able to stay focused where there is no money to be had. The Team must have leaders who can make things happen and bring in outsiders and most importantly provide a product roadmap and keep to it.

The target audience contains old farts like me that have chosen classic for just hobby and gave up asking much more of them.

The only Team with more than a few proven talented team members and track record is only MorphOS at this point, sorry not starting a red blue thing. It's just reality and I accepted it years ago. Not discounting any of the good stuff people have worked on and done, but comparatively speaking it isn't really close if we are honest.

So as I have done (although I still put cash into classics) is back the team that provides a "future", that is my new Amiga product I would like to see, keep working on what is already working instead of starting over yet again... 3.20 is already on it's way along with changes and updates to all the many applications used. I fully believe from what I have seen and experienced that 25 will be a good year for MorphOS as were the prior years.

I know everyone is worried about the hardware as we have discussed well enough to last a lifetime at this point, and I have complete confidence the Team will sort that out when the time is right. But for the moment, they have a inexpensive solution that delivers and you can ask for feature to make your life easier and is very actively developed by many people working together and supporting each other. Like is or not MorphOS is the complete opposite of what Amiga has been. They keep all their development and roadmaps private and just release it. Remember when Iris and Wayfarer were in pre version 1 5 years ago. There was some discussion and they became core parts of the OS with many changes to the OS even to make them a reality. Now with countless updates, they are really good stuff. I prefer to be surprised and see the progression rather than empty promises to waste time... A methodical team approach that is mature and productive...

It makes zero sense to create a new OS platform and expect everyone to show up and figure out who does what and hope for the best in 2025. Been there many times and after the novelty wear out with the A600+, PiAmiga, AROS, WinUAE or Vampire, it gets put away. Makes total sense to already have a great Team already working together and have a fully featured OS with all the Apps natively running to make a shift in platform easier as every items will need to be setup for the shift, similar in some regard to what Apple did when it shifted platforms back in the day...

@Agami

(Bet your a hunt for red october movie fan :) )

Love the example. I burned my ships and learned/supported MorphOS. Welcome to the new world.

Last edited by Matt3k on 08-Feb-2025 at 04:33 PM.
Last edited by Matt3k on 08-Feb-2025 at 04:23 PM.
Last edited by Matt3k on 08-Feb-2025 at 04:19 PM.

@agami

For myself, the future is ARM with PiStorm in my A1200.

AROS isn't going to figure in my future as I consider it to be a dead end. Productivity software that has always existed has benefitted from the PiStorm's speed and now WHDLoad slaves are starting to appear that address speed issues in games that rely too much on ChipRAM and are thus limited by it.

The future is bright for the Amiga right now, and I have no wish to leave 3.x or the enormous library of fantastic software behind :)

@ZXDunny

The AROS ecosystem is confusing to me because I can’t keep all of the variants straight in my head.

_________________

@Matt3k

Atm, the only path for making money from the Amiga is the retro market. This is like selling LP music discs and modern LP players in mainstream physical stores. Amiga Paula has its distinct characteristics for the retro market.

The PiStorm approach is to leverage Amiga retro hardware with modernized compute and graphics performance with low financial risk. The same for the Z3660 (using MYIR Z-turn ARM Cortex A9 + AMD/Xilinx FPGA or real 68060) project. The keyword is leverage from a market with demand. This demand is not strong enough to support RP2040-level SoC development.

https://www.electronicsweekly.com/news/business/manufacturing/rochester-supply-68040s-2015-07/

Rochester Electronics has been licensed by Freescale Semiconductor to manufacture the 68040 32bit microprocessor product family which Freescale will stop producing in November (2015).

Rochester has original Freescale packaged parts and silicon die from which it will manufacture a variety of device options. Freescale has also provided access to full design and test IP, making it possible for Rochester to continue manufacturing products even if all existing wafer stocks are exhausted.

In addition to the 68040 family, Rochester legacy support includes 68020, 68030, 68060 and the 68882 floating point coprocessor. Freescale will also be supporting Rochester on the supply of the MC68360 QUICC communications processor, which went EOL at the end of April.


Rochester specializes in retro legacy hardware supply chain and support.


Octavo Systems has Texas Instruments ARM-based AM335x on a SIP with built-in 68K bus support that is used for the Amiga-related Buffee project. Amiga is not the only 68K legacy system that needs ARM-based modernization. Octavo OSD335x System-in-Package (SIP) has ARM Cortex A8 which is road-killed by PiStorm16/PiStorm32's RPi CM4's ARM Cortex A72.

Z3660 may not survive when PiStorm32 is ported to A3000/A4000's local CPU slot due to PiStorm32's ARM Cortex A72 beating Z3660's ARM Cortex A9.


Atm, AROS 68K doesn't replace PiStorm-Emu68's AmigaOS 3.9 based CoffeineOS.

Last edited by Hammer on 08-Feb-2025 at 11:50 PM.
Last edited by Hammer on 08-Feb-2025 at 11:45 PM.
Last edited by Hammer on 08-Feb-2025 at 11:14 PM.

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7950X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@AmigaMac

I feel ya, I really do. Hosted (and where), not hosted, this ABI stuff, which hardware is supported and which is not... it gets complex.

And then you finally get it booted and... go back to 3.x because that's where all the games are.


But ironically for me I think the future will be ARM - possibly AROS (once it's usable) taking over from AOS4 on the ARM with emu68 providing seamless - and I mean 100% seamless right down to being able to boot Amiga floppies both physical and imaged, along with full hardware access - 68k integration so none of the software is lost.

This isn't a modern, going-concern platform like PC or Mac; we can't afford to just abandon all that history the way the Mac does when they upgrade to the next platform.

@ZXDunny

If you have the right visionary, the Amiga could bring back the fun in computing going beyond its hardcore user community. But that’s me being an optimist.

_________________

Quote:

ZXDunny wrote: For myself, the future is ARM with PiStorm in my A1200.

So just a retro console with no possibility for anything new.

Quote:

ZXDunny wrote: AROS isn't going to figure in my future as I consider it to be a dead end.

Funny, this is the only Amigoid version that is really available for programmers.

@AmigaMac

Based on history, C= A500's 68000 was contemporary with mainstream game consoles like the Sega Mega Drive/Genesis's 68000 and low price range.

Applying the same argument position for the current time, PiStorm-Emu68's RPi 4B's or CM4's ARM Cortex A72 is contemporary with Nintendo's mainstream Switch's ARM Cortex A57 and low price range. Cortex A53 is below Cortex A57. https://en.wikipedia.org/wiki/Comparison_of_ARM_processors#ARMv8-A

Neo-AmigaOne PPC G3 wasn't contemporary with Nintendo's Game Cube's low price range and excuses were made for a higher price PPC SE/XE motherboard + PPC G3/G4 combo.

Genesi's EfikaPPC has the right idea on the low price range, but the MPC5200B's PowerPC e300 is dated relative to a similar era Nintendo Wii's 729 MHz PPC 750CXe CPU (G3+) with custom 64bit SIMD. PowerPC e300 CPU is derived from PowerPC 603e (PowerPC G2).

Without UAE, pricey neo-Amiga PPC wasn't retro since they don't support the Amiga game legacy. It's most likely Vampire's "more than 10,000 unit sales" crushed neo-AmigaOne PPC i.e. Amiga is not a Mac since most 68K Amiga games behave like PC's hit-the-metal VGA games with 68K CPU and C= graphics/audio IP. Amiga didn't have a large business customer base to support a Mac-style 68K-to-PC transition and Amithlon's "we don't care about games" also failed i.e. Bernd Meyer didn't understand Amiga's usebase majority.

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7950X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

Quote:

AmigaMac wrote: ..., the Amiga could bring back the fun in computing going beyond its hardcore user community.

The Raspberry PI community is much more like the Amiga community was , I have even installed an SQL server and an AI image generator to an RPi4, there are also some version of similar hardware built into old Amigas.