Click Here
home features news forums classifieds faqs links search
6090 members 
Amiga Q&A /  Free for All /  Emulation /  Gaming / (Latest Posts)
Login

Nickname

Password

Lost Password?

Don't have an account yet?
Register now!

Support Amigaworld.net
Your support is needed and is appreciated as Amigaworld.net is primarily dependent upon the support of its users.
Donate

Menu
Main sections
» Home
» Features
» News
» Forums
» Classifieds
» Links
» Downloads
Extras
» OS4 Zone
» IRC Network
» AmigaWorld Radio
» Newsfeed
» Top Members
» Amiga Dealers
Information
» About Us
» FAQs
» Advertise
» Polls
» Terms of Service
» Search

IRC Channel
Server: irc.amigaworld.net
Ports: 1024,5555, 6665-6669
SSL port: 6697
Channel: #Amigaworld
Channel Policy and Guidelines

Who's Online
22 crawler(s) on-line.
 95 guest(s) on-line.
 0 member(s) on-line.



You are an anonymous user.
Register Now!
 DiscreetFX:  6 mins ago
 matthey:  35 mins ago
 agami:  1 hr 48 mins ago
 Hammer:  2 hrs 12 mins ago
 a3000d/t:  2 hrs 53 mins ago
 emilianojay:  4 hrs ago
 CharlotteMurphy:  4 hrs 49 mins ago
 lewac:  4 hrs 57 mins ago
 tonyadams:  5 hrs 1 min ago
 rosebl22:  5 hrs 6 mins ago

/  Forum Index
   /  Amiga General Chat
      /  what is wrong with 68k
Register To Post

Goto page ( Previous Page 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 Next Page )
PosterThread
Hammer 
Re: what is wrong with 68k
Posted on 25-Dec-2024 23:45:41
#341 ]
Elite Member
Joined: 9-Mar-2003
Posts: 6171
From: Australia

@matthey

Quote:

68060 some integer instruction castration too including less than rare 64-bit MUL/DIV

68060 has castrated 4 byes per cycle L1 fetch which SysInfo's benchmark trips over. This aspect limits 68060's potential. 68040 bus also limits 68060 which should have been 88000's 64-bit wide 60x bus.

AC68080's L1 fetch is wider at 16 bytes (128 bits) per cycle.

Modernization would need to be applied for 68060 to improve scalar streaming compute performance.

68060 is effectively a half-completed CPU. R&D resources shifted beyond the IBM-developed PowerPC 601 core.

Using the CPU for software 3D renderer needs stream compute bias designs.

As Intel's flagship product, the Pentium (P5) was quickly replaced by the Pentium Pro (P6) in 1995. Other Pentium clones have Pentium Rating (PR) as a friendly indicator of IPC gains with mostly integer performance. Quake's FPU bias was a good tactic to cull competing CPUs with weak FPUs.

Streaming compute performance bias applies to raytracing and deep learning.

Modern CPUs have FMA3 scalar and vector instructions.
Modern CPUs have dot product instructions.


Quote:

@matthey

The 68k had poor planning after the 68000 ISA.

68010 includes a corrective instruction set for Popek and Goldberg virtualization requirements. Motorola should price out 68000 with a lower-priced 68010.

Motorola should have exercised market manipulation tactics towards an improved 68K instruction set foundation.

When Jay's Amiga or Jack's Atari selects a 68K, it should have been a 68010 baseline.

_________________
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

 Status: Offline
Profile     Report this post  
Hammer 
Re: what is wrong with 68k
Posted on 26-Dec-2024 2:59:05
#342 ]
Elite Member
Joined: 9-Mar-2003
Posts: 6171
From: Australia

@kolla

Quote:

kolla wrote:
@Hammer

Quote:
Commodore licensed REXX from IBM for automation with different 3rd party applications.


This nonsense popped up in the late 90s and doesn’t make any sense - IBM had nothing to do with ARexx, so why would CBM license REXX from IBM? Btw, the “full” myth is that IBM got “the looks” of OS 2.x in return from CBM to use with OS/2, which also doesn’t make any sense as CBM got that look from (as you point out) NeXT as well as Motif. There used to be three articles around claiming this CBM/IBM nonsense, all pointing at each other as source.


REXX was developed at IBM by Mike Cowlishaw.

_________________
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

 Status: Offline
Profile     Report this post  
kolla 
Re: what is wrong with 68k
Posted on 26-Dec-2024 8:49:25
#343 ]
Elite Member
Joined: 20-Aug-2003
Posts: 3357
From: Trondheim, Norway

@Hammer

Quote:

Hammer wrote:

REXX was developed at IBM by Mike Cowlishaw.


Yes, and?!

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

 Status: Offline
Profile     Report this post  
Karlos 
Re: what is wrong with 68k
Posted on 26-Dec-2024 10:30:21
#344 ]
Elite Member
Joined: 24-Aug-2003
Posts: 4843
From: As-sassin-aaate! As-sassin-aaate! Ooh! We forgot the ammunition!

@kolla

And... Isn't it obvious?

_________________
Doing stupid things for fun...

 Status: Offline
Profile     Report this post  
kolla 
Re: what is wrong with 68k
Posted on 27-Dec-2024 0:44:06
#345 ]
Elite Member
Joined: 20-Aug-2003
Posts: 3357
From: Trondheim, Norway

@Karlos

Not at all, since when did using rexx or the even writing a rexx interpretor need licensing from IBM? I asked the original author of Regina about this a few decades ago and he thought it sounded quite unlikely. Rather the contrary, they worked towards rexx being ansi standard, which it became in 1992 iirc.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

 Status: Offline
Profile     Report this post  
Karlos 
Re: what is wrong with 68k
Posted on 27-Dec-2024 0:55:53
#346 ]
Elite Member
Joined: 24-Aug-2003
Posts: 4843
From: As-sassin-aaate! As-sassin-aaate! Ooh! We forgot the ammunition!

@kolla

Sorry, it was rhetorical sarcasm. I had no idea what point he was making either.

_________________
Doing stupid things for fun...

 Status: Offline
Profile     Report this post  
bhabbott 
Re: what is wrong with 68k
Posted on 27-Dec-2024 4:36:43
#347 ]
Cult Member
Joined: 6-Jun-2018
Posts: 509
From: Aotearoa

@Hammer

Quote:

Hammer wrote:

Motorola should have exercised market manipulation tactics towards an improved 68K instruction set foundation.

When Jay's Amiga or Jack's Atari selects a 68K, it should have been a 68010 baseline.

Shoulda coulda woulda.

Typical Amiga fan attitude. No appreciation of it what it did have, just complaints about a few things it didn't.

So what did the 68010 have that was worth it for us? Support for virtualization - no. Faster loop code - yes. Vector base register - perhaps. But none of these small improvements justified paying a much higher price.

But what about other stuff? With those extra 5,000 transistors they could have given it a barrel shifter. This would make graphics operations and dealing with byte quantities much faster. They might also have been able to speed up the multiply and divide instructions. That's stuff which could have significantly improved code efficiency in machines that couldn't justify the cost of a faster CPU or hardware blitter etc.

But this was the real world, so we didn't get that or a dozen other things you could think of - and it didn't matter. The 68000 was still much better than the competition. The only other CPU that mattered was the 8086, and it sucked in comparison. Those horrid segment registers, miserable 1MB addressing range, lack of 32-bit registers, 8 bit opcodes that unaligned the code etc. Furthermore the only x86 implementation that mattered was the IBM PC, which used the even weaker 8088 running at less than 5MHz.

The other thing the 68000 had that the competition didn't was - a nice orthogonal instruction set with asm code that was a joy to write. I would choose that even over a faster CPU if it was more difficult to code for.

One of the reasons the Amiga shone was the ability for 'normal' people to produce efficient asm code. On other 16/32 bit CPUs most developers relied on compilers to do the job, with the result being slower more bloated code. Compilers were also expensive and required powerful machines to run at an acceptable speed, putting them out of reach of typical home computer users. Furthermore many home computer 'veterans' had learned to code in asm on 8-bit machines, so a CPU with a nicer instruction set was more attractive.

Some of us still code in 68k asm today for the same reasons. The 68000 became irrelevant in the computing industry many years ago, but like a classic board game it never gets tired for us. I don't even code for 68020 because I am happy with the original 'rules of the game'. For me, asking what's wrong with 68k is like asking what's wrong with chess. Imagine changing the rules because you didn't think the moves were 'powerful' enough. You wouldn't be making the game better, you would be destroying it.

 Status: Offline
Profile     Report this post  
agami 
Re: what is wrong with 68k
Posted on 4-Jan-2025 0:59:41
#348 ]
Super Member
Joined: 30-Jun-2008
Posts: 1898
From: Melbourne, Australia

@Karlos

Quote:
Karlos wrote:
@kolla

Sorry, it was rhetorical sarcasm. I had no idea what point he was making either.

@Hammer tries to make points?

_________________
All the way, with 68k

 Status: Offline
Profile     Report this post  
Hammer 
Re: what is wrong with 68k
Posted on 4-Jan-2025 4:11:48
#349 ]
Elite Member
Joined: 9-Mar-2003
Posts: 6171
From: Australia

@bhabbott

Quote:

Shoulda coulda woulda.

Typical Amiga fan attitude. No appreciation of it what it did have, just complaints about a few things it didn't.


Wrong. 68000 vs 68010 is a Motorola problem. Motorola can push 68010 by pricing out 68000.

68010's corrective 68K instruction set should been a priority for Motorola.

Quote:

@bhabbott

So what did the 68010 have that was worth it for us? Support for virtualization - no. Faster loop code - yes. Vector base register - perhaps. But none of these small improvements justified paying a much higher price.

https://www.youtube.com/watch?v=ZVstab917TA
68010 vs 68000 running Xenon II CDTV. 68010's cache loop is useful for moving pixels.

68020/68030 family has a hardware barrel shifter.

68K CPU's price is a Motorola problem. System integrators like Commodore-Amiga Inc. follow Motorola's price guidance. e.g. during AA development in early 1991, Amiga's shift from the proposed baseline 68000 14Mhz to 68EC020 14Mhz (68EC020-16 rated) was based on Motorola's offering lower cost 68EC020 in 1991.

If AA Amiga was released in 1990, the baseline CPU for AA would have 68000 @ 14Mhz with only Lisa being 32-bit.

Baseline Amiga's CPU evolution is based on Motorola's price guidance.

The price difference between 68EC020-16 and 68EC020-25 is relatively minor, hence the fault is on Commodore.

Motorola's 68030-25 largely (#metoo) follows Intel's 386DX-25 price guidance and they were caught off guard by Am386-40.



Quote:

@bhabbott

But this was the real world, so we didn't get that or a dozen other things you could think of - and it didn't matter. The 68000 was still much better than the competition. The only other CPU that mattered was the 8086, and it sucked in comparison. Those horrid segment registers, miserable 1MB addressing range, lack of 32-bit registers, 8 bit opcodes that unaligned the code etc.

68000 didn't have an FPU until 68020's 1984 release along with 68881 FPU.

8088 and 8086 has 8087 FPU which supported FP32, FP64, FP80, INT32 and INT64 datatypes.
X87 FPU supports both integer and floating-point formats.

For longer datatypes beyond native INT16, the original IBM PC would need an optional 8087 FPU. IBM provided an 8087 FPU socket in the original IBM PC 5150.

8087 was superseded by IEEE-754 80287 FPU in 1982. Lotus 2.0's 1985 release supported x87 FPU for the existing PC install base.

MOS/CSG 65xx CPU family doesn't have FPU. CSG wasn't serious about FPU and the financial market's IT needs.

Quote:

@bhabbott,

Furthermore the only x86 implementation that mattered was the IBM PC, which used the even weaker 8088 running at less than 5MHz.

Wrong,

1. Intel's second-best platform was PC-98 with a larger install base when compared to Amiga and Atari TOS (68K DRi GEM/GEMDOS variant).

Intel licensed NEC's ÎĽPD7220 PC-98 display standard vs IBM display standards that were widely cloned.

PC-98's first model, the PC-9801 used 8086 @ 5 Mhz.

The 8-bit external bus was a 1980-era cost reduction method and 68008 wasn't ready for IBM 5150's 1981 release window.

80286's built-in MMU allowed for commodity Xenix 286 workstation PCs.


2. IBM provided an 8087 FPU socket in the original IBM PC 5150. 8087 FPU supported FP32, FP64, FP80, INT32 and INT64 datatypes


3. For 32-bit ALU CPUs, 80386 was 1 year later than 68020's 1984 release.

80386 one-up the 68020 with 386's integrated MMU as standard which allowed for commodity Xenix 386 workstation PCs. 386's flat memory address-protected mode was used for Doom.

Before 68551 MMU's release, most 68K Unix workstations are vendor lock-in with custom MMUs.

For Compaq's 386 AT clone, Xenix 386 and Windows 2.0x 386 were released in 1987.
Intel, Compaq, Microsoft, and SCO co-operated like a tight 386 team.

Quote:

The other thing the 68000 had that the competition didn't was - a nice orthogonal instruction set with asm code that was a joy to write. I would choose that even over a faster CPU if it was more difficult to code for.

One of the reasons the Amiga shone was the ability for 'normal' people to produce efficient asm code. On other 16/32 bit CPUs most developers relied on compilers to do the job, with the result being slower more bloated code. Compilers were also expensive and required powerful machines to run at an acceptable speed, putting them out of reach of typical home computer users. Furthermore many home computer 'veterans' had learned to code in asm on 8-bit machines, so a CPU with a nicer instruction set was more attractive.

Amiga didn't exist before July 1985's A1000 release. From 1983 to 1985, A1000 had about 2 years of R&D.

Sinclair QL (68008) was released in 1984. Sales flopped with just 150,000 units sold.
Apple Lisa (68000) was released in 1983 with a higher price than IBM PC and PC clones.
Apple Mac 128K (68000) was released in 1984.
Apple Mac 512K (68000) was released in 1985. Mac GUI editions of MS Word, MS Excel, and Aldus Page Maker were released in 1985. Floating point emulated on 68000.

Apple Mac II color and 32bit 68020/68881 system integration R&D started around 1985. Mac II was released in 1987.

In 1987, NEC announced one million PC-98s were shipped while the Apple Mac passed a 1 million install base in 1987.


From https://jeremyreimer.com/rockets-item.lsp?p=137
Download the data in Excel format
https://jeremyreimer.com/uploads/Com..._1975-2012.xls

Amiga's annual sales
1985: 100K
1986: 200K
1987: 300K
Total: Amiga's install base for 1987 is about 600K units.

Without standardized MMU, 68000's early 32-bit programming model leadership was largely wasted and half-baked.

For Compaq's 386 AT clone standard (which the Gang of Nine supported), Xenix 386 and Windows 2.0x 386 were released in 1987. Intel, Compaq, Microsoft, and SCO co-operated like a 386 team.

32-bit desktop computing includes the PMMU standard. Half-baked PMMU-less 32-bit desktop platforms died.

Apple Mac was lucky with Steve Jobs' NextSTEP with PMMU foundation that was in R&D from 1986. NeXTSTEP 1.0 (1989)'s foundation PMMU-equipped CPU was NeXT Computer's October 1988 release with 68030 and NeXTSTEP 0.8. Meanwhile, Microsoft's PMMU OS foundation of Windows NT (OS/2 3.0) R&D started in 1988.

Apple's 1st 68030-based Mac was released in Sep 1988 via the Macintosh IIx model's 68030 @ 16 Mhz.

With the NeXTSTEP foundation, Steve Jobs's return to Apple allowed the MacOS X release which competed against Microsoft's full Windows NT shift via Windows XP's release.

Commodore wasted R&D on somebody else's PMMU Unix platform.

Last edited by Hammer on 04-Jan-2025 at 05:59 AM.
Last edited by Hammer on 04-Jan-2025 at 05:42 AM.
Last edited by Hammer on 04-Jan-2025 at 05:36 AM.
Last edited by Hammer on 04-Jan-2025 at 05:32 AM.
Last edited by Hammer on 04-Jan-2025 at 05:17 AM.
Last edited by Hammer on 04-Jan-2025 at 05:10 AM.
Last edited by Hammer on 04-Jan-2025 at 05:01 AM.
Last edited by Hammer on 04-Jan-2025 at 04:58 AM.
Last edited by Hammer on 04-Jan-2025 at 04:41 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

 Status: Offline
Profile     Report this post  
Hammer 
Re: what is wrong with 68k
Posted on 4-Jan-2025 4:20:24
#350 ]
Elite Member
Joined: 9-Mar-2003
Posts: 6171
From: Australia

@kolla

Quote:

kolla wrote:
@Hammer

Quote:

Hammer wrote:

REXX was developed at IBM by Mike Cowlishaw.


Yes, and?!


Disprove this claim https://www.landley.net/history/mirror/os2/OS2Warp.html

With Microsoft no longer doing development on the user interface, IBM was faced with creating this themselves. In this timeframe, a deal was made with Commodore. Commodore licensed IBM's REXX scripting language for inclusion in their AmigaOS, and IBM took many GUI design ideas from the AmigaOS for their new GUI.

_________________
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

 Status: Offline
Profile     Report this post  
Kronos 
Re: what is wrong with 68k
Posted on 4-Jan-2025 12:05:57
#351 ]
Elite Member
Joined: 8-Mar-2003
Posts: 2713
From: Unknown

@Hammer

A random quote on a random obscure website is just that.

At best it is a starting point for further investigations into such a topic.



Plenty of obvious nonsense on the internet making bold claims.

_________________
- We don't need good ideas, we haven't run out on bad ones yet
- blame Canada

 Status: Offline
Profile     Report this post  
Kronos 
Re: what is wrong with 68k
Posted on 4-Jan-2025 12:34:30
#352 ]
Elite Member
Joined: 8-Mar-2003
Posts: 2713
From: Unknown

@Hammer

Quote:

Hammer wrote:


The price difference between 68EC020-16 and 68EC020-25 is relatively minor, hence the fault is on Commodore.



No idea where that thread started but if you are implying that the A1200 should have had the 25MHz part, that would have been a total waste.

Sure they could have OCed it to run at the full 28MHz base clock, but with only 50% (at best) access to the 7MHz CHIP MEM it would have been just as slow while eating so much power it would have needed cooling (maybe even active).

-> would have made 0 sense in the A1200 we got
--> would have made limited sense in a better A1200 with at least some fast RAM (which would have cost more).

_________________
- We don't need good ideas, we haven't run out on bad ones yet
- blame Canada

 Status: Offline
Profile     Report this post  
kolla 
Re: what is wrong with 68k
Posted on 4-Jan-2025 14:11:03
#353 ]
Elite Member
Joined: 20-Aug-2003
Posts: 3357
From: Trondheim, Norway

@Hammer

It’s not to me to disprove anything, it’s up to you and your sources to prove the claim.

As I have mentioned a few times now, this was something that popped up in the late 90s, mostly claimed by amiga zealots who wished for Amiga to have some sort of role to play with the big names, then it was picked up by one or two OS/2 sites, and then they could point at each other (and for a while wikipedia) for reference.

But the claim makes zero sense; the timeline doesn’t fit, ARexx wasn’t IBM and there’s nothing in OS/2 that is specifically "Amigs like". So even if there ever was such a cross-licensing deal, neither IBM nor CBM made any use of it.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

 Status: Offline
Profile     Report this post  
Kronos 
Re: what is wrong with 68k
Posted on 4-Jan-2025 15:23:48
#354 ]
Elite Member
Joined: 8-Mar-2003
Posts: 2713
From: Unknown

@kolla

And there was also that story about how the ARexx author gor screwed over by C= and hence all copies included in AmigaOS2.1 to 3.x lacked a license.

If that author had been an IBM employee this would surely have seen some legal action.

So yeah, story makes little to no sense.

_________________
- We don't need good ideas, we haven't run out on bad ones yet
- blame Canada

 Status: Offline
Profile     Report this post  
Hammer 
Re: what is wrong with 68k
Posted on 4-Jan-2025 22:56:53
#355 ]
Elite Member
Joined: 9-Mar-2003
Posts: 6171
From: Australia

@Kronos

Quote:

Kronos wrote:
@Hammer

No idea where that thread started but if you are implying that the A1200 should have had the 25MHz part, that would have been a total waste.

Sure they could have OCed it to run at the full 28MHz base clock, but with only 50% (at best) access to the 7MHz CHIP MEM it would have been just as slow while eating so much power it would have needed cooling (maybe even active).

-> would have made 0 sense in the A1200 we got
--> would have made limited sense in a better A1200 with at least some fast RAM (which would have cost more).

28 MHz 020 accelerators didn't need active cooling.

For CD32/A1200's FP DRAM with 140 ns read/write speeds, the price difference between 2 MB vs 8 MB is only $20.

A1200 was spec'ed by Jeff Frank and the blame is on him.

Commodore -The Final Year book made the point about Jeff Porter's CD32 8 MB RAM rejection was done by Commodore management's marketing team.

2MB CD32/A1200 missed 1993 MPC2's 4 MB target and 3 MB RISC based games consoles group (3DO, PS1 and Saturn).

A1200/CD32 2MB RAM was largely frozen in time 1991 MPC1's 2MB RAM minimum.

Jeff Frank didn't project into 1993 MPC2's 4 MB RAM target.

VCD focus is blamed on Commodore's chairman, wasting a RISC CPU on non-core business activity.

Commodore The Final Years book is about attaching specific blame on individuals and group.

MIPS CPU family deserves the design wins over Motorola's 68K.

When I purchased A1200 in 2020, I quickly purchased 8 MB Fast RAM board since my Q4 1992 era 386DX-33 (over clocked to 40 MHz) PC has 4 MB RAM. To me, A1200 with Fast RAM seems to be non-64KB cache equiped 386DX-16 with ET4000AX.

A1200 wasn't a major unit sellar in Xmas Q4 1992 due to low 44,000 production units. A1200 is MIA in Australian XMas 1992. Customers knows A1200 caused Osborne effect on non-graphics upgradable A600. Many A500 owners purchased gaming PCs or SNES for Xmas Q4 1992. The money for buying games for the Amiga shifted to the PC or SNES.

Bruce from tiny market NZ has argued business use case on Amiga platform, but he in a minority (hint, Phase 5 went bust). The Amiga is not the Mac. Only the PowerPC neo-Amiga camp continues to think Amiga is like Mac.







Last edited by Hammer on 04-Jan-2025 at 11:35 PM.
Last edited by Hammer on 04-Jan-2025 at 11:24 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

 Status: Offline
Profile     Report this post  
Hammer 
Re: what is wrong with 68k
Posted on 5-Jan-2025 13:18:04
#356 ]
Elite Member
Joined: 9-Mar-2003
Posts: 6171
From: Australia

@kolla

Quote:

kolla wrote:
@Hammer

It’s not to me to disprove anything, it’s up to you and your sources to prove the claim.

As I have mentioned a few times now, this was something that popped up in the late 90s, mostly claimed by amiga zealots who wished for Amiga to have some sort of role to play with the big names, then it was picked up by one or two OS/2 sites, and then they could point at each other (and for a while wikipedia) for reference.

But the claim makes zero sense; the timeline doesn’t fit, ARexx wasn’t IBM and there’s nothing in OS/2 that is specifically "Amigs like". So even if there ever was such a cross-licensing deal, neither IBM nor CBM made any use of it.


From Commodore's Amiga Manual: Using the System Software V2.05 (1991)
https://archive.org/details/Using_the_System_Software_V2.05_1991_Commodore/page/n755/mode/2up
AREXX is the Amiga version of the IBM Rexx programming language. - Commodore

Commodore credits REXX to IBM.

_________________
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

 Status: Offline
Profile     Report this post  
Kronos 
Re: what is wrong with 68k
Posted on 5-Jan-2025 17:00:21
#357 ]
Elite Member
Joined: 8-Mar-2003
Posts: 2713
From: Unknown

@Hammer

Quote:

Hammer wrote:


28 MHz 020 accelerators didn't need active cooling.


So? You'd still be using more power and extra $ on a chip with no benefits in this use case.

Quote:

For CD32/A1200's FP DRAM with 140 ns read/write speeds, the price difference between 2 MB vs 8 MB is only $20.


Which would have been nice if the "free beer" jumper worked (aka 8MB Chip) but even then it would have made 0 sense for the A1200 which was supposed to cheap above everything else.
Want that to be Fast RAM? You would need 2 banks and the control/refresh logic so you have now added atleast 50$ to the BOM.

Could've would've should've, the A1200 was too expensive to build compared to what it could do hence it didn't repeat A500 sales number.
Increasing both specs and price would not have changed anything.

C= simple lost years of progress in the late 80s and just didn't have the funds to fasttrack delayed developments. So they did release a "good for 89" chipset in 92 and if they had lived a bit longer we might have gotten a "good for 92" chipset (AAA) in 95.

Just adding more RAM or faster CPU (at higher cost) wouldn't have changed any of that.

_________________
- We don't need good ideas, we haven't run out on bad ones yet
- blame Canada

 Status: Offline
Profile     Report this post  
kolla 
Re: what is wrong with 68k
Posted on 5-Jan-2025 17:22:48
#358 ]
Elite Member
Joined: 20-Aug-2003
Posts: 3357
From: Trondheim, Norway

@Hammer

Quote:

Commodore credits REXX to IBM.


Yes, and? Noone ever disputed that REXX has its origins at IBM where it was (and is!) used throughout their portfolio of operating systems. But REXX also existed (and exists) outside ot IBM, with implementations for just about any operating system, both proprietary and open source. It’s a language, not a product.

Last edited by kolla on 05-Jan-2025 at 06:10 PM.
Last edited by kolla on 05-Jan-2025 at 06:09 PM.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

 Status: Offline
Profile     Report this post  
bhabbott 
Re: what is wrong with 68k
Posted on 5-Jan-2025 22:58:45
#359 ]
Cult Member
Joined: 6-Jun-2018
Posts: 509
From: Aotearoa

@kolla

Quote:

kolla wrote:
@Hammer

Quote:

Commodore credits REXX to IBM.


Yes, and? Noone ever disputed that REXX has its origins at IBM where it was (and is!) used throughout their portfolio of operating systems. But REXX also existed (and exists) outside ot IBM...

Hey, don't deflate one of our most cherished myths!

The fact is that IBM was looking to cross-license Amiga stuff, and Commodore was too so they wouldn't be blocked from selling PCs. This wasn't unusual. However it didn't mean that IBM intended to use any Amiga technology in their machines, just that they didn't want to be on the wrong end of a patent or copyright dispute. But this didn't apply to AREXX because IBM had already released the REXX language spec. to the community on a permissive license.

 Status: Offline
Profile     Report this post  
Hammer 
Re: what is wrong with 68k
Posted on 7-Jan-2025 3:33:57
#360 ]
Elite Member
Joined: 9-Mar-2003
Posts: 6171
From: Australia

@Kronos

Quote:

So? You'd still be using more power and extra $ on a chip with no benefits in this use case.


For a smoother Wing Commander, at least 1 MB of Fast RAM is needed for stock 68EC020-14 or higher.

Chip RAM only A1200/CD32 caused the bundled 68EC020 to operate like 32-bit 7Mhz 68EC020 instead of 14 Mhz 68EC020. Commodore's "32-bit" is just token "32-bit" marketing.

The IPC difference between 68020 and 68030 is very minor.

A1200/CD32 targeted 1991 MPC level 1 with 386SX-16's 2MB system memory and VGA's 256K discrete video memory setup. A1200/CD32 was frozen in the 1991 era of MPC level 1's system memory storage.

Gaming PC's minimum switched to 4MB system RAM via 1993 MPC Level 2 and there's 3MB RAM games console group i.e. 3DO, PS1, and Saturn.


Quote:

@Kronos

Which would have been nice if the "free beer" jumper worked (aka 8MB Chip) but even then it would have made 0 sense for the A1200 which was supposed to cheap above everything else.

Want that to be Fast RAM? You would need 2 banks and the control/refresh logic so you have now added atleast 50$ to the BOM.

A1200's Budgie can support Zorro II address range Fast RAM.

With AA Gayle, Budgie is a hybrid design with Buster's Zorro II 24-bit memory address space, 32-bit data bus Ramsey memory controller, and 32-bit Bridgette. The 4 MB Fast RAM card for A1200 is relatively simple.

A1200 could have tolerated higher BOM costs since Mehdi Ali threw A1200's healthy profit margin for CD32's lower profit margin.

1 MB to 2 MB Fast RAM is needed for stock 68EC020-14 and beyond.

8 MB Chip RAM wouldn't be possible due to Alice's address range limitation. The multi-media group can only improve the hardware around the edges of the AGA chipset.

Jeff Porter's multi-media group doesn't have the authority to modify the core AGA chipset.
Jeff Porter tried to push for an MIPS-X SoC-boosted powered CD-ROM console with A600's guts.

The Amiga group is administrated by Commodore PC's Jeff Frank.

CDTV-CR's FMV module had a 24bit display, 16bit PCM stereo sound, 40Mhz custom MIPS-X SoC with compute power exceeding PS1's MIPS R3000A @ 32Mhz (VCD MPEG1 at 10 fps)

MIPS R3000A @ 30Mhz can play MP3 at 256 kbit/s rate without using DSP.

East Asian PS1 has an extra chip for VCD playback i.e. MPEG1 at 24 fps.

For MP3 playback capability at lowish cost, the Motorola solution is found in Atari Falcon's 68EC030 @ 16 Mhz with 24-bit DPS56001 @ 32Mhz. Atari Falcon's sales flopped hard.

My point for this topic, the MIPS camp defeated 68K on performance vs cost factor.

Amiga's fortune is linked with the sinking 68K ship.

Quote:

@Kronos

Could've would've should've, the A1200 was too expensive to build compared to what it could do hence it didn't repeat A500 sales number.
Increasing both specs and price would not have changed anything.

A1200's production wouldn't match A500's production scale due to 1 million A600 units' debt spiral and unproven Philippines A1200 assembly capability.


Quote:

C= simple lost years of progress in the late 80s and just didn't have the funds to fasttrack delayed developments. So they did release a "good for 89" chipset in 92 and if they had lived a bit longer we might have gotten a "good for 92" chipset (AAA) in 95.

Just adding more RAM or faster CPU (at higher cost) wouldn't have changed any of that.

One problem, PS1's hardware is mostly LSI (for MIPS R3000A, MIPS-based GTE) and Toshiba (for GPU) licensed IP.

With careful 3rd party IP selection, Sony's Ken Kutaragi leadership beats Commodore, Atari, and 3DO with 3rd party license IP!

Jeff Porter has the right idea about using RISC-based MIPS IP, but couldn't fully execute it.

Many A500 owners had enough of waiting and purchased an SNES or a gaming PC in 1992.

Many of my school friends in Xmas Q4 1992 purchased SNES or a full 32-bit gaming PC.

Nintendo addressed the RISC threat with RISC-based SuperFX stop-gap tactics.

Last edited by Hammer on 07-Jan-2025 at 03:55 AM.
Last edited by Hammer on 07-Jan-2025 at 03:51 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

 Status: Offline
Profile     Report this post  
Goto page ( Previous Page 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 Next Page )

[ home ][ about us ][ privacy ] [ forums ][ classifieds ] [ links ][ news archive ] [ link to us ][ user account ]
Copyright (C) 2000 - 2019 Amigaworld.net.
Amigaworld.net was originally founded by David Doyle