@OneTimer1

Quote:

OneTimer1 wrote: So just a retro console with no possibility for anything new.

Historically, there's nothing new with A500's off-the-shelf 68000 like other 68000-equipped platforms. Commodore's purchasing strength resulting an affordable asking price range and Amiga graphics/audio hardware added value for the customer.

Quote:

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

Is there 10,000 programmers programming for AROS? The bulk of Vampire's "more than 10,000 unit sales" goes to end consumers and a few of them get involved with programming for Vampire.

Last edited by Hammer on 09-Feb-2025 at 12:41 PM.
Last edited by Hammer on 09-Feb-2025 at 12:34 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

@OneTimer1

Quote:

OneTimer1 wrote: 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.

Are you arguing the bulk of 4.8 million Amiga owners are programmers instead of end consumers?

_________________
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:

Hammer wrote: Are you arguing the bulk of 4.8 million Amiga owners are programmers instead of end consumers?

4.8 million Amiga owners

@Hammer

Quote:

Hammer wrote: Historically, there's nothing new with A500's off-the-shelf 68000 like other 68000-equipped platforms.

nothing new except:
flexible bitplane grafik, HAM, D/A 8-Bit DMA Audio, preemptive multitasking OS, GUI with mouse, ...

Quote:

Is there 10,000 programmers programming for AROS?

No, I never claimed that, but there are more programmers working on this OS than on any other 68k OS.

Quote:

The bulk of Vampire's "more than 10,000 unit sales"

There are no 10,000 Vampires but you can see it's OS as a result of AROS and not of AOS3.1

Oh BTW.: The A600GS are also driven by AROS, and I don't know what RTG system the PiStorm will get, maybe something from AROS.

Last edited by OneTimer1 on 09-Feb-2025 at 01:25 PM.
Last edited by OneTimer1 on 09-Feb-2025 at 01:21 PM.

OneTimer1 Quote:

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.

Commodore used to dominate Acorn which struggled to compete and now it is the Commodore replacement A-EonKit which is far more noncompetitive than Acorn was while the Acorn replacement RPi Foundation has become the competitive modern low end hardware replacement of Commodore. It is not just the RPi and Amiga communities that have swapped places, but RPi and A-EonKit hardware.

OneTimer1 Quote:

nothing new except: flexible bitplane grafik, HAM, D/A 8-Bit DMA Audio, preemptive multitasking OS, GUI with mouse, ...

The 68k Amiga was an elegant DMA/interrupt/message driven masterpiece that offloaded the CPU allowing Commodore to use low power low cost CPUs. ARM emulation of the 68k Amiga is a 100% CPU usage polled emulation inefficient kludge. RPi hardware is better, even their MCUs.

Feature | RP2040 | RP2350
Package QFN-56EP QFN-60EP/QFN-80EP
CPU Cores 2xARM_Cortex-M0+ 2xARM_Cortex-M33+FPU_or_2xRISC-V_Hazard3
CPU Clock 133 MHz 150 MHz
SRAM 264kiB 520kiB
Flash None RP2350=none/RP2354=2MiB
OTP None 8kiB
DMA 12chan+2IRQ 16chan+4IRQ
PIO 2(8_state_machines) 3(12_state_machines)
PWM 16 24
ADC 4-chan12-bit 4-chan12-bit(QFN-60EP)/8-chan12-bit(QFN-80EP)
DAC None None
HSTX None One
Engines ? RNG|SHA-256

https://en.wikipedia.org/wiki/RP2350

The RPi MCU SoC specs resemble more modern 68k Amiga SoC specs, minus the fact that Commodore botched creating the single chip SoC that would have dramatically improved their competitiveness and profit margins. With emulation, efficiencies drop off the map and practically nobody uses the A-EonKit hardware for embedded use. It is ironic that Acorn Archimedes hardware avoided DMA to reduce transistors and simplify the chipset using a similar philosophy as simplifying ARM RISC, both the ARM hardware design and ISA, resulting in more and more expensive memory used. ARM Holdings borrowed code density from the 68k and SuperH licensed from Hitachi. ARM/RPi SoC/MCU chipsets bare more resemble to the 68k Amiga than ARM Acorn Archimedes as well. The 68k Amiga is the enviable hardware to draw on but ARM has decades of incremental improvements borrowed from hardware like the 68k Amiga and a huge modern silicon advantage. The largest Amiga disadvantage though is lack of leadership with A-EonKit thinking that it is possible to compete with outdated silicon and emulation.

OneTimer1 Quote:

No, I never claimed that, but there are more programmers working on this OS than on any other 68k OS.

No new 68k Amiga hardware using semi-modern silicon is a development killer. I like the 68k Amiga but it is dead without new hardware. The retro 68k Amiga market is where the opportunity is so without new 68k Amiga silicon, the Amiga is dead.

OneTimer1 Quote:

There are no 10,000 Vampires but you can see it's OS as a result of AROS and not of AOS3.1

There may be 10,000 Vampires but current Vamp/AC hardware is not the same as the original Vamp hardware which was low priced. The original Amiga 600 accelerator was 90 Euros, quickly creating a 6 month backlog of orders even with 2 people producing the hardware. Majsta was aggressively targeting Amiga hardware prices, trying to drive down hardware prices kind of like Eben Upton did with the RPi. Production techniques were all wrong as quickly learned, the FPGA was too small and the accelerator was just for the limited supply of Amiga 600s reducing economies of scale but the low price did wonders for boosting sales. The current Vamp/AC borrows Majsta's branding only and abandons Majsta's goals.

Last edited by matthey on 09-Feb-2025 at 08:50 PM.
Last edited by matthey on 09-Feb-2025 at 08:46 PM.
Last edited by matthey on 09-Feb-2025 at 05:51 PM.
Last edited by matthey on 09-Feb-2025 at 05:46 PM.

10,000 Vampire have been sold!

Wow I didnt know it had been that successful.

_________________
AmigaNG, YouTube, LeaveReality Studio

@amigang

Quote:

10,000 Vampire have been sold!

Take it with a grain of salt, even if it was true it doesn't represent 10,000 users.

The V2 cards sold well as they were affordable and available, many of us got several. It would surprise me if the V4 series has sold equally well, and no doubt many (perhaps even most) V4 users also used V2 cards. And no doubt many of us have since replaced V2 cards with PiStorms, so...

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@OneTimer1

Quote:

nothing new except: flexible bitplane grafik, HAM, D/A 8-Bit DMA Audio, preemptive multitasking OS, GUI with mouse, ...

Again, Commodore's purchasing strength resulting an affordable asking price range and Amiga graphics/audio hardware added value for the customer.

Atari ST has a GUI (DRi GEM), a mouse and bit-plane graphics system. Jack Tramiel decided to apply his product segment 1986 MegaST's blitter from ST models which wreaked baseline ST platform evolution since STE's 1989 release was too late.

For the majority of Amiga owners, it's 1989 A500 Rev6 is the majority that covered 1989, 1990 and 1991. For my family in 1989 market place, it a decision between C= A500 Rev6A vs Atari STE. A500's superior delivered game quality is the major factor e.g. R-Type, Hybris, Silk Worm, TV Sports: Basketball and 'etc'. My Dad's workmates (owns A2000s as a gaming machines) and my older cousin (owns 286-16 with fast VGA clone PC) recommended the Amiga.

Atari 1024 STE's 1989 game demonstration is weaker than Amiga's 1989 games. My family's Xmas 1989 is the A500. My Dad and my older cousin heavily played TV Sports: Basketball game.

Atari STE would be sufficient for my school work's DTP and word processing needs, but it was Amiga's superior games experience that pushed it over the line for the Amiga.

"Pre-emptive multitasking OS" wasn't a major factor for "kick-the-OS" Amiga game majority use cases. The critics against AmigaOS's pre-emptive multitasking virtue is Commodore didn't enforce the OS environment like on Apple's MacOS and 3DO.

On 3DO, the OS environment is enforced and it has the game APIs for its hardware.

AmigaOS only platforms like AmigaOS PPC is a failure in comparison to WHDLoad games capable Vampire's "more than 10,000 units" sold. AmigaOS only platforms had their chance since the 1st AmigaOne and 68K DraCo.

Commodore's marketing for pre-emptive multitasking OS was weak and A500's standard 512K memory is not enough for proper pre-emptive multitasking with major business apps.

For major business apps, A500's pre-emptive multitasking comes into play with larger usable memory.

In the end of the day, AmigaOS pre-emptive multitasking was a half-bake solution when competition gained memory protection with 32-bit pre-emptive multitasking OS evolution.

The original C= goal for cheap C= PMMU with 68000 was distracted by keeping the MMU feature for higher end Amiga offerings. CSG could have made C= PMMU cheap and standard for the Amiga. Two C= MMUs were developed for 68000 and 68020 respectively before 68551's release.

Quote:

There are no 10,000 Vampires but you can see it's OS as a result of AROS and not of AOS3.1

Wrong. Read https://eab.abime.net/showpost.php?p=1563593&postcount=801

When I was in the queue for Vampire V2 during COVID19 lockdown, it was advertised "more than 10,000 Vampires" have been shipped.

Last edited by Hammer on 10-Feb-2025 at 03:16 AM.
Last edited by Hammer on 10-Feb-2025 at 03:03 AM.
Last edited by Hammer on 10-Feb-2025 at 02:11 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

@matthey

Quote:

Commodore used to dominate Acorn which struggled to compete and now it is the Commodore replacement A-EonKit which is far more noncompetitive than Acorn was while the Acorn replacement RPi Foundation has become the competitive modern low end hardware replacement of Commodore. It is not just the RPi and Amiga communities that have swapped places, but RPi and A-EonKit hardware.

For the 21st century, Acorn's nemesis Jack Tramiel's Commodore and Atari doesn't exist i.e. there is no Commodore's MOS low cost advantage to stop ARM.

Tramiel's Commodore's MOS 65xx CPU family is a major player in good "bang per buck" CPU in the early 1976 to 1984.

Acorn's RISC CPU project is successful while its RISCOS platform has failed. Low power consumption/affordability/high performance offered ARM a safe market space from X86's absolute performance/affordability and it developed ARM's critical R&D mass and financial
base and sales distribution channels. Smartphone OS platforms are important factors for ARM's application processor progress.

For smartphone OS platforms, ARM defeated Hitachi SuperH, Intel Atom, MIPS and Freescale 68000-based DragonBall VZ. ARM's win has rendered solutions like Bill McEwen's "Amiga Anywhere" redundant.

For ARM's application processor market, ARM's direct competitor is Qualcomm (US)'s Oryon ARM clones.

Due to Hitachi SuperH's lost, Japan Inc's Softbank purchased ARM Holdings. The UK effectively lost sovereign control over ARM and ARC (Argonaut RISC Core). Synopsys Inc (US) owns ARC.

Due to US national security issues, the US government funded US university sector' R&D learns from ARM and created RISC-V with a focus on code density. MIPS Inc. switched to RISC-V.

AMD64's patent has expired, hence it has reduced RISC-V's open-source ISA advantage.

Commodore's MOS 65xx CPU family's weak R&D progress spawns ARM and ARC CPU families. ARM1 is a solution for Acorn's post-65xx CPUs with the added original Berkeley RISC design concepts (US government-funded).

ARC was started from Nintendo's Argonaut SuperFX R&D when SNES's 65C816 evolution wasn't fast enough to address the 3D use cases. SNES was the last mainstream 65xxx CPU platform before switching to the MIPS CPU family.

My post is extra background information with national security aspects.

Unlike IBM OS/2 division, the IBM Semiconductor division addressed the 3D games market with PowerPC 602 for 3DO M2. Just before 3DO M2's release with a library of M2 games, Japan industrials tapped Panasonic to not release 3DO M2, hence protecting Sony's PS1, Nintendo's N64 and Sega Saturn.

3DO M2's dual PowerPC 602 with 66Mhz FMA units are potent for floating point 3D.

IBM Semiconductor division tried again with 3DO M3 (purchased by MS), MS Xbox 360 and Sony PS3. IBM PPE focused on 3D floating point math via VMX128.

Unlike SiFive U74, ARM Cortex A53's dual 64bit FADD/FMUL/FMA3/SIMD units have preserve floating point 3D bias.

68060 is not open source CPU core which needs a license from NXP.

_________________
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

@kolla

Quote:

kolla wrote: @amigang Take it with a grain of salt, even if it was true it doesn't represent 10,000 users. The V2 cards sold well as they were affordable and available, many of us got several. It would surprise me if the V4 series has sold equally well, and no doubt many (perhaps even most) V4 users also used V2 cards. And no doubt many of us have since replaced V2 cards with PiStorms, so...

Reddit's Amiga group has 24K members.
Facebook's Commodore Amiga group has 31.7K members.
Facebook's Amiga Game Selector has 14K members.
Facebook's The A500 Mini/Maxi User Group has 10.1K members.

_________________
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:

Hammer wrote: @kolla Quote: kolla wrote: @amigang Take it with a grain of salt, even if it was true it doesn't represent 10,000 users. The V2 cards sold well as they were affordable and available, many of us got several. It would surprise me if the V4 series has sold equally well, and no doubt many (perhaps even most) V4 users also used V2 cards. And no doubt many of us have since replaced V2 cards with PiStorms, so... Reddit's Amiga group has 24K members. Facebook's Commodore Amiga group has 31.7K members. Facebook's Amiga Game Selector has 14K members. Facebook's The A500 Mini/Maxi User Group has 10.1K members.

Yes, those numbers are much more representative.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

kolla Quote:

Take it with a grain of salt, even if it was true it doesn't represent 10,000 users. The V2 cards sold well as they were affordable and available, many of us got several. It would surprise me if the V4 series has sold equally well, and no doubt many (perhaps even most) V4 users also used V2 cards. And no doubt many of us have since replaced V2 cards with PiStorms, so...

I have similar expectations. Supply and demand works in Amiga Neverland too. One interesting consequence of early Vampire accelerators was the revival and demand for the Amiga 600 due to the low cost accelerators. As I recall, low Amiga 600 prices increased dramatically and even broken machines were in demand. There is a magical price point when hardware prices drop below ~100 USD/Euros. The low end SBC market is based on it with RPi, RPi clones and RISC-V VisionFive2 SBCs practically needing to reach this price level to have adequate economies of scale, and practically requires integrated GPUs for the higher end of this low end market.

Hammer Quote:

Acorn's RISC CPU project is successful while its RISCOS platform has failed. Low power consumption/affordability/high performance offered ARM a safe market space from X86's absolute performance/affordability and it developed ARM's critical R&D mass and financial base and sales distribution channels. Smartphone OS platforms are important factors for ARM's application processor progress.

The original ARM ISA was only used in the Acorn Archimedes and 3DO outside of embedded use where it was the 4th place ISA by 32-bit volume in the late 1990s. 3DO was discontinued in 1996 and Archimedes in the mid 1990s so ARM was just a 4th place embedded ISA. What are the 3rd and 4th place 32-bit/64-bit embedded ISAs today? How many decades did it take for ARM to reach the top 2 embedded 32-bit ISAs and were they still using the original ARM ISA? Is that "successful"?

Hammer Quote:

For smartphone OS platforms, ARM defeated Hitachi SuperH, Intel Atom, MIPS and Freescale 68000-based DragonBall VZ. ARM's win has rendered solutions like Bill McEwen's "Amiga Anywhere" redundant.

Motorola/Freescale all but abandoned 68k development except for the heavily castrated ColdFire architecture designed to fill the market below where fat embedded PPC could scale. SuperH is a simple ISA for small cores and MCUs but the fixed length 16-bit encoding is a major performance handicap when scaling it up. MIPS is too fat and x86 too power hungry for embedded use. Thumb and Thumb-2 eventually won due to code density and low power but it was more of a lack of code density competition with the best competitor 68k AWOL due to Motorola/Freescale politics. Performance gained in importance for embedded and smart phone markets due to silicon improvements where Thumb(-2) ISAs have more limited performance than the 68k ISA so ARM switched ISAs again. Ironically, the original 68k technology ARM Thumb(-2) is derived from has both performance and code density.

Last edited by matthey on 10-Feb-2025 at 08:11 PM.

Fantasy Project idea 4
Up to date Amiga web browser that could run on 68K/PPC and be kept up todate.
[/quote]

#4 pretty fullfilled already.
Wayfarer just got updated. Works nice.
Only grain of salt: lack of cpu power.
OK, not for the entire Amiga family, but at least for MorphOS.

_________________
My programs: via.bckrs.de
MorphOS user since V0.4 (2001)

@matthey

Quote:

The original ARM ISA was only used in the Acorn Archimedes and 3DO outside of embedded use where it was the 4th place ISA by 32-bit volume in the late 1990s. 3DO was discontinued in 1996 and Archimedes in the mid 1990s so ARM was just a 4th place embedded ISA. What are the 3rd and 4th place 32-bit/64-bit embedded ISAs today? How many decades did it take for ARM to reach the top 2 embedded 32-bit ISAs and were they still using the original ARM ISA? Is that "successful"?

Widelogic Limited was founded on 5th of November 1990, but this was rapidly changed on 3rd of December 1990 to Advanced RISC Machines Limited and structured as a joint venture between Acorn Computers, Apple, and VLSI Technology.

Apple backed two RISC instruction sets, and Apple Newton used ARM, which serves as an example for the following smart handheld devices.

Due to Henri Rubin's inaction, the Commodore engineer who pushed for RISC CPU direction in 1987 joined Apple's Newton project in 1988. Bob Welland was an A500 co-designer, two custom C= MMUs for 68000 and 68020 respectively, and an original A2620 with C= MMU.

Bob Welland would work on RISC chips and help pioneer the Newton PDA.

Before Steve Jobs' NextStep OS entry, the Newton OS was also licensed to a number of third-party developers including Sharp and Motorola who developed additional PDA devices based on the Newton platform. Motorola added wireless connectivity, as well as made a unique two-part design, and shipped additional software with its Newton device, called the Marco. Sharp developed a line of Newton devices called the ExpertPad PI-7000/7100; those were the same as Apple's MessagePad and MessagePad 100, the only difference is the physical design (the ExpertPads feature a screen lid, which Apple added in 1994 with the release of the MessagePad 110) and the naming.

ARM CPU design was boosted by DEC's StrongARM. The StrongARM was a collaborative project between DEC and Advanced RISC Machines (ARM) to create a faster ARM microprocessor.

StrongARM SA-110 was the first microprocessor in the StrongARM family. The first versions, operating at 100, 160, and 200 MHz, were announced on 5 February 1996. When announced, samples of these versions were available, with volume production slated for mid-1996. Faster 166 and 233 MHz versions were announced on 12 September 1996. DEC's IP injection into ARM has boosted StrongARM's clock speed beyond 68060 rev 6.

The SA-110's first design win was the Apple MessagePad 2000. It was also used in a number of products including the Acorn Computers Risc PC and Eidos Optima video editing system.

Apple MessagePad 2000 is a precursor to the iPad and the missing component was Steve Jobs' NextStep OS which led to MacOS X and iOS with clone Unix strength with end user friendly GUI.

The 1st ARM GSM phone was the Nokia 6110 in December 1997.

DEC was fragged by Intel and Compaq.

StrongARM SA-110 beats Motorola's 68000-based Dragon Ball VZ in Plam PDA devices before ARM9 based Palm Tungsten.

StrongARM SA-110 includes MMU as standard.

ARM6 reached 33 Mhz via ARM600. Mid-life update 3DO's ARM60 reached 20Mhz.
ARM7 reached 40 Mhz via ARM700.
ARM7T reached 70Mhz via ARM7TDMI.
ARM8 reached 72 Mhz via ARM810.
ARM9T reached 180 Mhz via ARM920T.

DEC Alpha IP's influence over ARM (via Strong ARM), and X86 CPU families is significant. ARM's and X86's road map is DEC Alpha.


Quote:

Motorola/Freescale all but abandoned 68k development except for the heavily castrated ColdFire architecture designed to fill the market below where fat embedded PPC could scale. SuperH is a simple ISA for small cores and MCUs but the fixed length 16-bit encoding is a major performance handicap when scaling it up. MIPS is too fat and

MIPS R series has three major wins for desktop game consoles i.e. Sony PS1, Nintendo 64 and Sony PS2.

The MIPS R series based solutions are semi-custom for game 3D processing by platform vendors e.g. exploited MIPS's co-processor interface for custom 3D geometry co-processors. influenced by SGI's MIPS usage, these boxes have similar "power without the price" ideology.

Amiga Hombre's goal was to bring "big iron" RISC 3D workstation class for game console price range. The Amiga is about "power without the price".

Historically, A500's 68000 selection was contemporary with mainstream game consoles like Sega Mega Drive/Genesis. 68000 and 68020 were significant players in the Unix workstation market before big iron RISC's entry. SGI shifted to MIPS R2000 from 68K during 68030 dispointment.

Motorola attempted to preserve two chip business model for workstations i.e.
1. 68551 MMU + 68020,
2. 560xx strong integer DSP + 68030 (with built-in MMU),

SGI criticised Motorola's 68030's value vs performance proposition relative to Intel 386DX.

Baseline Amiga AGA departed from contemporary CPU performance and price during desktop game console's 32-bit generation shift.

Commodore went down with the sinking 68K ship.

Quote:

x86 too power hungry for embedded use.

High clock speed 486DX and Pentium chased after performance 1st from "big iron" RISC workstations.

Intel created a separate product line for mobile.
For P54C (600 nm)
Mobile Pentium 75 has 6/4.4 W TDP
Mobile Pentium 100 has 8/5.9 W TDP

For P54LM (350 nm)
Mobile Pentium 120 has 7.1 (Max.10.87) W TDP
Mobile Pentium 100 has 10 (Max.14.0) W TDP

For "P55LM" (350 nm)
Mobile Pentium MMX 120 has 4.2W W TDP
Mobile Pentium MMX 166 has 9 W TDP

For "Tillamook" (250 nm), purpose made for mobile that is separate from Pentium II
Mobile Pentium MMX 166 has 2.9 (Max. 4.1~5.4) W TDP
Mobile Pentium MMX 200 has 3.4 (Max. 5.0~6.1) W TDP

Celeron = budget desktop, budget gaming PC.
Pentium II = desktop, premium gaming PC.
Pentium II Xeon = Server chasing after "big iron" RISC CPU workstaions including ECC memory support.

Pentium MMX has improved 1 IPC FPU.

Compaq has influenced Intel's P5 and P6 CPU R&D for performance 1st.

Advanced Computing Environment (ACE)'s MIPS CPU cloneable desktop ACE PC reference designs are a threat to Intel X86's existence. Intel's focus is on this market.

During 1990s, AMD and Intel wouldn't be focusing on the game console market until the late 1990s original Xbox R&D phase i.e. AMD K7 Durion with 3DNow+ and Intel Coppermine 128K with SSE budget game CPUs. Bill Gates intervene and ordered K7 Duron to changed to Intel Coppermine 128K. The fallout caused AMD CPU division to not participate in Xbox 360 contract until AMD's hiring Lisa Su from IBM Semiconductor. Lisa Su was responsible for executive level interface between IBM Semiconductor with Sony's PS3 project.

Xbox 360 project was free from Bill Gates' pro-Intel interference and a new relationship was created with IBM Semiconductor.

Back to the Amiga, Lorraine originally had 128k of memory (over the A1000s eventual 256k) and was primarily designed as a home flight simulator machine. In the AUI interview Jay Miner describes his experience of viewing of a military flight simulator developed by Singer-Link. Impressed by what he saw, Miner begins to consider the use of blitters to improve the graphics capabilities. http://www.bambi-amiga.co.uk/amigahistory/ahistory.html

https://www.youtube.com/watch?v=uy8sJ9AxvYI
Link Flight Simulation Demo (1984)

The original Amiga aimed high for the masses.

Sega's 3D texture map experience is via Lockheed Martin's Real3D division (purchased by Intel). https://segaretro.org/Sega_Model_2

Your "embedded" argument departed from Amiga's original mission.

Last edited by Hammer on 11-Feb-2025 at 01:43 AM.
Last edited by Hammer on 11-Feb-2025 at 01:40 AM.
Last edited by Hammer on 11-Feb-2025 at 12:14 AM.
Last edited by Hammer on 10-Feb-2025 at 11:15 PM.
Last edited by Hammer on 10-Feb-2025 at 11:13 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

@Hammer

Quote:

Reddit's Amiga group has 24K members. Facebook's Commodore Amiga group has 31.7K members. Facebook's Amiga Game Selector has 14K members. Facebook's The A500 Mini/Maxi User Group has 10.1K members.

That is why I was surprised by them numbers, I mean that just users/members on these social networks, many will be the same users, I suppose some dont like Facebook & these are English based, some German / Polish etc people might not hang out on them, so my rough guess for the Amiga community is about 50,000 strong, so 1 in 5 owning a Vampire is very high.

Quote:

The V2 cards sold well as they were affordable and available, many of us got several. It would surprise me if the V4 series has sold equally well, and no doubt many (perhaps even most) V4 users also used V2 cards. And no doubt many of us have since replaced V2 cards with PiStorms, so...

Very good point! Still I feel that good strong sales.

I do wonder what the sale figures for the A500 mini are.

Going back to the original topic, it is important to kinda know the market size of the Amiga community, actually that a good one for a new topic!

_________________
AmigaNG, YouTube, LeaveReality Studio

@matthey

Quote:

MIPS is too fat and x86 too power hungry for embedded use.

DM&P Electronics Vortex86 is still alive in the embedded market.

https://www.youtube.com/watch?v=BdSJgoP2a88
This Commercial Kitchen Appliance Uses an original Pentium clone CPU. These are used in restaurants like McDonald's.

Vortex86DX3 SoC reached 1.0 GHz dual-core i686-compatible CPU.

Vortex86EX2 reached Core 2's SSSE3 level which enhanced for HD audio/video decoding/encoding e.g. AAC.

Vortex86DX3 has a 6-stage pipeline i686 clone with 32KB L1 instruction, 32KB L1 data and 512 KB L2 caches.

https://theretroweb.com/chip/documentation/vortex86dx3-a9126-data-sheet-v13-bf-65e87f3549348113610580.pdf
i686 CPU has been semi-modernized into SoC with DDR3, integrated 2D GPU that includes two DSP to accelerate Video decode for MPEG2 MP @HL, VC-1 AP @L2 720P up to 20Mbps, and H.264 HP @L4.1 1080P up to 40Mbps video decode. Vortex86DX includes 40-pin GPIO for "industrial Pi" GPIO-like support.

Vortex86DX is an x86 compatible System-on-Chip (SoC) manufactured with a 90nm process and designed to consume less than 1 watt of electricity.

68060 doesn't have a multi-million dollar revenue company that is focused and competent like DM&P Electronics (from Taiwan). NXP's Freescale has abandoned 68060.

Unlike AC68060, Vortex86DX3 includes standard PMMU to run desktop class Linux, FreeBSD, QNX, and Windows 7 (NT 6.1). AC68060 V4 should be renamed into AC68EC080 V4 (with FP64 FPU).

Rochester Electronics has a 68040 license and frozen copy-and-paste engineering.

The company size for slow 68060 evolution is about the size of DM&P Electronics with a small competent engineering team. A significant amount of work would be needed to the 68060 into a modern embedded SoC that is on par with Vortex86DX3. Business development would be a major factor.

Last edited by Hammer on 12-Feb-2025 at 01:00 AM.
Last edited by Hammer on 12-Feb-2025 at 12:58 AM.
Last edited by Hammer on 12-Feb-2025 at 12:48 AM.
Last edited by Hammer on 12-Feb-2025 at 12:20 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:

DM&P Electronics Vortex86 is still alive in the embedded market. https://www.youtube.com/watch?v=BdSJgoP2a88 This Commercial Kitchen Appliance Uses an original Pentium clone CPU. These are used in restaurants like McDonald's. Vortex86DX3 SoC reached 1.0 GHz dual-core i686-compatible CPU. Vortex86EX2 reached Core 2's SSSE3 level which enhanced for HD audio/video decoding/encoding e.g. AAC. Vortex86DX3 has a 6-stage pipeline i686 clone with 32KB L1 instruction, 32KB L1 data and 512 KB L2 caches. https://theretroweb.com/chip/documentation/vortex86dx3-a9126-data-sheet-v13-bf-65e87f3549348113610580.pdf i686 CPU has been semi-modernized into SoC with DDR3, integrated 2D GPU that includes two DSP to accelerate Video decode for MPEG2 MP @HL, VC-1 AP @L2 720P up to 20Mbps, and H.264 HP @L4.1 1080P up to 40Mbps video decode. Vortex86DX includes 40-pin GPIO for "industrial Pi" GPIO-like support. Vortex86DX is an x86 compatible System-on-Chip (SoC) manufactured with a 90nm process and designed to consume less than 1 watt of electricity.

Vortex86 SoCs have survived and become specialized for legacy embedded and retro markets. Supporting x86 compatibility with older I/O while adding newer I/O is sometimes important. The embedded market is huge and even embedded niches can survive. There is demand for embedded SoCs using 68k, ColdFire and SuperH CPU cores too. These were the big endian industrial embedded cores that dominated the world for decades and now there are few big endian core options and support is difficult to find with most ARM big endian support being eliminated in newer cores. ColdFire support can be added to 68k cores and 68k like SuperH assembly code is easier to convert to the 68k than any little endian architecture. There is the SuperH J-Core project with open cores but no plans to produce commodity SoCs that I am aware and I do not believe the architecture will scale up in performance without major changes. Commodity 68k SoC ASICs should attract embedded and retro customers looking for big endian 68k, ColdFire and to a lesser extent SuperH compatibility much like the Vortex86 does for embedded and retro x86 customers.

The Vortex86 CPU cores have suspect competitiveness for general purpose embedded use. I believe the original Vortex86 core used a Rise Technology mP6 8-stage superscalar in-order core.

https://en.wikipedia.org/wiki/MP6

The mP6 is comparable to a Pentium MMX but it used too much power and area/transistors to be competitive in the embedded market. A core used 3.6 million transistors with 8kiB I+D caches and max power @250MHz was 10.72W using a 250nm process.

https://web.archive.org/web/20060615180701/http://www.sandpile.org/impl/mp6.htm

The original 6-stage OoO PPC604 used 3.6 million transistors with 16kiB I+D caches and drawing 14-17W @133 MHz using a 500nm process, more than the 2.5 million transistor 68060 using the same process and 8kiB I+D caches. Die shrink the hot PPC604 to 250nm and increase the caches to 32kiB I+D with a few other improvements increasing the transistors to 5.1 million and the PPC604ev only draws 6W @250MHz. The mP6 core draws almost 11W @250MHz using the same process but with 1/4 of the caches. The 68060 is lower power and area than the PPC604 and would likely already be less than 5W @250MHz using a long outdated 250nm process. The mP6@250MHz using a 250nm process may need a fan while the 68060@250MHz using a 250nm process can likely be passively cooled which is a large advantage for embedded use.

The mP6 core was hot and power hungry while performance was behind AMD and Intel offerings. The mP6 core was undesirable for embedded use too. The power could be reduced by reducing the transistors but the caches were already small. Reducing the pipeline stages reduces the number of transistors as well as eliminating the superscalar capabilities, trading performance for lower power and area. Rather than strip the 8-stage superscalar mP6 core down, I believe DM&P replaced it with a 6-stage scalar Cyrix 5x86 core originally using ~2 million transistors which is more comparable to a 80486. They clocked up the lower performance 5x86 core to a similar performance as the mP6 core with a small reduction in power and likely a larger reduction in area. This is an undesirable strategy necessitated by the high power and area making the core less competitive. The shallower pipeline reduces the max clock speed but it was already limited for embedded use by the high power draw. Superscalar in-order cores offer superior performance and performance efficiency compared to scalar cores but, again, power reduction was a higher priority. The x86 cores have a power problem!

See the chart and read the comment at the bottom of the following page to see the mention of the Cyrix 5x86.

https://www.cpushack.com/2010/10/07/the-rise-of-the-vortex86-embedded-x86/

ChkCPU CPU info also reveals the newer Vortex86 cores as Cyrix 5x86 cores (see benchS.jpg).

https://www.vogons.org/viewtopic.php?t=81027

The chart.png from the same link reveals performance and power benchmarks with the Vortex86@166MHz having a much better performance efficiency than the Vortex86DX@800MHz and Vortex86DX2@933MHz but clocking the Vortex86 up to similar clock speeds would use several times the power of the Vortex86DX2 and Vortex86DX3 cores. The later Vortex86 cores based on the Cyrix 5x86 have higher clock speeds, larger caches, more memory bandwidth and likely newer silicon which offsets the loss of mP6 performance but these SoCs are more like a 80486 or 68040 with semi-modern technology improvements. An 8-stage superscalar 68060 is lower power than some 6-stage scalar CISC cores using the same process so it is not necessary to give up Pentium like performance for embedded use unless using x86 or worse, x86-64.

Hammer Quote:

68060 doesn't have a multi-million dollar revenue company that is focused and competent like DM&P Electronics (from Taiwan). NXP's Freescale has abandoned 68060. Unlike AC68060, Vortex86DX3 includes standard PMMU to run desktop class Linux, FreeBSD, QNX, and Windows 7 (NT 6.1). AC68060 V4 should be renamed into AC68EC080 V4 (with FP64 FPU). Rochester Electronics has a 68040 license and frozen copy-and-paste engineering. The company size for slow 68060 evolution is about the size of DM&P Electronics with a small competent engineering team. A significant amount of work would be needed to the 68060 into a modern embedded SoC that is on par with Vortex86DX3. Business development would be a major factor.

DM&P development was able to reuse existing CPU core designs and perform basic enhancements like increased caches, memory performance upgrades, I/O upgrades and silicon improvements. Too bad they decided to reduce their CPU cores to 80486 levels to reduce power for embedded use but retro use volumes alone may not be enough to support the business and fanless designs are a cost advantage even for retro use. Further development of a Cyrix 6x86 core would have been more interesting and powerful than a Pentium.

Rochester buys the old stock of EOL printed wafers, cuts them up and finishes turning them into chips. It is not high production and does not modernize the silicon. They only create new chips from scratch when printed wafers run out and demand is high enough but they specialize in exact functionality recreations. They have a license for the EOL chips which likely includes all code and data necessary to create chips but it is a different market than licensing for development on more modern silicon like was done for most ColdFire cores and SoCs. The ColdFire does not have the retro appeal of the 68k though. The 68060 design is good for both retro and embedded markets but maybe potential licensees are concerned with the 68060 design age and the fact that it was not clocked up. Maybe this knowledge would just create more of a bargain when negotiating a license though. The real problem is that nobody in the retro markets wants to create competitive hardware, preferring to allow RPi hardware to eat their lunch. Retro IP is being bought up by the likes of Amiga Corporation and Atari but retro hardware licensing has not reached the same levels yet, preferring to rely on emulation where products like the THEA500 Mini and Atari VCS have lost major market share to lower cost emulation primarily due to RPi hardware. Maybe some day someone will realize the value and decide to compete with RPi hardware.

@Hammer

Thanks for responding with that bit of history.

_________________

@matthey

Quote:

Vortex86 SoCs have survived and become specialized for legacy embedded and retro markets. Supporting x86 compatibility with older I/O while adding newer I/O is sometimes important. The embedded market is huge and even embedded niches can survive.

I suspect Vortex86EX2 dual core is used for Space X's Falcon 9.

Falcon 9 has 3 dual core x86 processors running an instance of Linux on each core.

From https://www.vortex86.com/products/Vortex86EX2
Vortex86EX2 dual core can run an instance of Linux on each X86 core and supports ECC memory.

ECC memory support is a premium feature on Intel X86 CPU offerings. ECC memory support is semi-standard on AMD's AM4 and AM5 offerings since it's dependant on the motherboard's brand.

Space X Falcon 9 is not legacy embedded.

_________________
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

Vortex86DX doesn't support MMX.

Vortex86MX has MMX support.


Vortex86DX3 has MMX and SSE support. This Vortex86DX3 SKU has 999 Mhz clock speed
https://www.cpu-world.com/cgi-bin/CPUID.pl?CPUID=63347


Vortex86EX2's master CPU core has MMX, SSE, SSE2, SSE3, SSSE3 and NX support and it's produced using the 65 nm manufacturing process. This Vortex86EX2 SKU has 600 Mhz clock speed. https://www.cpu-world.com/cgi-bin/CPUID.pl?CPUID=72324


For Vortex86DX @ 933 Mhz
https://www.dr-lex.be/hardware/btplug_review.html

Speed
The next thing I tested is how well the Vortex86DX @933MHz performs, as well as the system as a whole. First of all, I ran a simple benchmark program of my own that counts how many times various calculations and data-processing operations can be executed during a fixed time. I have been running that program on many different machines so this allowed me to conclude that overall, the BTplug is somewhat comparable to a fast Pentium II or a slow Pentium III-based PC. It heavily depends on the task however, for convolving an image the Vortex86 is slower than anything I've benchmarked yet, but for creating a word histogram from text it is comparable to a Pentium 4. Emerging small packages in Gentoo is perfectly feasible, but you won't want to compile a kernel on the BTplug if you're in a hurry.

A note about the Vortex86DX: I have found rumours on some forums that the Vortex86DX would support MMX, but this is untrue. The cpu flags in /proc/cpuinfo only mention fpu tsc cx8, and a test program to check the availability of MMX fails. The newer Vortex86MX does support MMX.

_________________
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