English Amiga Board


Go Back   English Amiga Board > Main > Amiga scene

 
 
Thread Tools
Old 30 March 2020, 21:50   #1081
chb
Registered User

 
Join Date: Dec 2014
Location: germany
Posts: 288
Quote:
Originally Posted by Kyle_Human View Post
At which time you close it and move forward to the newer obsolete fab. Commodore were doing this, many later day chips were made for them in HP fabs. They were even transitioning away from owning their own computer factories and instead contracting out assembly work. Like every other computer company the path is to get out of making and concentrate on designing.
I think you're overlooking one important point: Changing to the newer obsolete fab would have meant a change to CMOS instead of the NMOS/HMOS process used. That means a complete redesign of the chip, not an easy feat if it originally was drawn by hand on big rolls of paper, without layout software. C= simply did not have the resources do it, that's why they didn't change Alice and Paula in the AGA chipset. LISA was CMOS (1.5 um I think) and was indeed also sourced out to HP, but the older chips were a different beast...
chb is offline  
Old 30 March 2020, 22:12   #1082
Gorf
Registered User

 
Join Date: May 2017
Location: Munich/Bavaria
Posts: 1,157
Quote:
Originally Posted by chb View Post
I think you're overlooking one important point: Changing to the newer obsolete fab would have meant a change to CMOS instead of the NMOS/HMOS process used. That means a complete redesign of the chip, not an easy feat if it originally was drawn by hand on big rolls of paper, without layout software. C= simply did not have the resources do it,...
That is true.
Despite of having its own chip factory, C= never really invested enough in R&D to actually design new chips .. at least after they let Chuck Peddle and later Bob Yannes (SID) go...

At Amiga (as Startup) obviously a handful of people were capable of not only designing and outlining the custom chips, but also draw the maps for the masks ... but C= did of course not hold on to them ....
Gorf is offline  
Old 30 March 2020, 22:59   #1083
StefaanC
Registered User

 
Join Date: Mar 2020
Location: Leest /Belgium
Posts: 26
Amstrad was a more successful company , founded in 1968 and took over by BskyB in 2007 for 125 million pound. Amstrad was already making satellite receivers so from then on Amstrad was removed from exchange in 09/10/2008.
BskyB took only the set top box division over.

In 1984 was Apple was not far from bankrupt after a disturbing advertising and lost many clients.

Sinclair was only active in the computer business and any business that is betting on one horse is asking for trouble. See commodore also. Apple survived due to I phone, Itunes etc.
StefaanC is offline  
Old 31 March 2020, 02:41   #1084
Gorf
Registered User

 
Join Date: May 2017
Location: Munich/Bavaria
Posts: 1,157
Quote:
Originally Posted by StefaanC View Post
Amstrad was a more successful company , founded in 1968 and took over by BskyB in 2007 for 125 million pound. Amstrad was already making satellite receivers so from then on Amstrad was removed from exchange in 09/10/2008.
BskyB took only the set top box division over.

In 1984 was Apple was not far from bankrupt after a disturbing advertising and lost many clients.
That was not 84 but 85.
The 84 (Super Bowl) ad is the famous and good one.
The ad from 85 is the creepy one.

But despite of the praised campaign, the Mac did not sell that good in 84.
The price was still too high and it was still sold alongside the Apple II and the even more expensive Lisa from 83.
But was incompatible to both.

The first months of 85, after the failed campaign, were also hard for apple, but already in August they sold their 500.000th Mac.
At a point were neither Commodore nor Atari had sold any relevant numbers of their new products at all!

And despite the rumors and the closing of one factory: Apple was never close to bankruptcy in the 80s - just look up their annual reports.
Gorf is offline  
Old 31 March 2020, 08:52   #1085
Kyle_Human
Banned

 
Join Date: Mar 2020
Location: Nijmegen, Netherlands
Posts: 46
Quote:
Originally Posted by chb View Post
I think you're overlooking one important point: Changing to the newer obsolete fab would have meant a change to CMOS instead of the NMOS/HMOS process used. That means a complete redesign of the chip, not an easy feat if it originally was drawn by hand on big rolls of paper, without layout software. C= simply did not have the resources do it, that's why they didn't change Alice and Paula in the AGA chipset. LISA was CMOS (1.5 um I think) and was indeed also sourced out to HP, but the older chips were a different beast...
You're overstating this. Redrawing the chips into layout software is nowhere near as hard work as designing them from scratch was.

When they were designed from scratch it was creating something totally new, having an idea and developing it into the real world, gaining an understanding of it as you create it. You are finding new problems you never saw before and then having to find solutions. The basic concepts "what is a blitter" "what is a bitplane" is all new and nobody has experience of it.

To take an old prexisting design and put it in the layout software is 90% copying. It's transcription work. The most difficult part is not falling asleep from boredom, you don't even need to know how it works. I've done this work myself.
Kyle_Human is offline  
Old 31 March 2020, 09:09   #1086
grond
Registered User

 
Join Date: Jun 2015
Location: Germany
Posts: 873
Quote:
Originally Posted by Kyle_Human View Post
To take an old prexisting design and put it in the layout software is 90% copying. It's transcription work.

The point was that NMOS logic circuits are different from CMOS circuits because of all the complementary logic in the pull-up paths that CMOS logic has where NMOS logic only uses resistors or single transistors for load.



However, I have mixed feeling about whether this would have been difficult to change. For sure you can't just shrink the finished layout when transferring it from NMOS to CMOS like you can do when transferring from one CMOS node to the other. If you have actual schematics, converting from NMOS to CMOS should still be mostly monkey work (but a lot of it). You would have to convert the circuitry, dimension the pull-up paths and draw a completely new layout. Not difficult, just a lot of work. I guess it was like it always was at Commodore: they thought they could get away with doing it cheaply. They needed the millions for the boss's airplane.
grond is offline  
Old 31 March 2020, 16:28   #1087
d4rk3lf
Registered User

d4rk3lf's Avatar
 
Join Date: Jul 2015
Location: Novi Sad, Serbia
Posts: 897
Anyone knows if Commodore had plans on upgrading OCS and ECS machines to Aga?
I mean, not a cheap solution, cheaper then Aga itself, but Amiga 500 + Aga upgrade to be more expensive then Amiga 1200. Something like that.

Would that be a good idea?
d4rk3lf is offline  
Old 31 March 2020, 19:10   #1088
StefaanC
Registered User

 
Join Date: Mar 2020
Location: Leest /Belgium
Posts: 26
I doubt that, they where struggling the last 10 years a lot and trying to squeeze as much money down for their products.
StefaanC is offline  
Old 31 March 2020, 23:58   #1089
redblade
Zone Friend

redblade's Avatar
 
Join Date: Mar 2004
Location: Middle Earth
Age: 36
Posts: 1,592
Quote:
Originally Posted by d4rk3lf View Post
Anyone knows if Commodore had plans on upgrading OCS and ECS machines to Aga?
I mean, not a cheap solution, cheaper then Aga itself, but Amiga 500 + Aga upgrade to be more expensive then Amiga 1200. Something like that.

Would that be a good idea?
How would that work on a A500? Would you piggy back on the chip or use the ZorroII port?

I guess they would have wanted people to upgrade to the Amiga1200
redblade is offline  
Old 01 April 2020, 00:25   #1090
Gorf
Registered User

 
Join Date: May 2017
Location: Munich/Bavaria
Posts: 1,157
Quote:
Originally Posted by redblade View Post
Quote:
Anyone knows if Commodore had plans on upgrading OCS and ECS machines to Aga?
I mean, not a cheap solution, cheaper then Aga itself, but Amiga 500 + Aga upgrade to be more expensive then Amiga 1200. Something like that.
Would that be a good idea?How would that work on a A500? Would you piggy back on the chip or use the ZorroII port?

I guess they would have wanted people to upgrade to the Amiga1200
actually the idea is not wrong.

It would have made sense in 1989/1990 to update the 16bit chipset to take advantage of the faster RAM available at this time.

especially the "double-cas" modes - meaning the ability to read two consecutive memory addresses in one DMA time slot.
It its done in AGA with 32bit allowing Lisa to read 64 pixels at once ... it could have been done with a 16bit Denise as well - allowing 32 pixels accordingly.

That would have provided 256 colors in low res and a "VGA" mode with 640x400 at 60+Hz with 16 colors...

all that without really changing the motherboard design of the 500 ...

together with a 14 MHz 68000 it would have been a real "A600" ...
Gorf is offline  
Old 01 April 2020, 13:42   #1091
StefaanC
Registered User

 
Join Date: Mar 2020
Location: Leest /Belgium
Posts: 26
It would have made sense to use a full 32 bit cpu so 32 adres lines, 32 data lines in the first place.
Excuse me, aga max non interlaced resolution was 640x480 pixels 8Bpp @72 Hz 60Hz is too low for a rock stable screen, that's minimal 72 Hz.

Last edited by StefaanC; 01 April 2020 at 14:55.
StefaanC is offline  
Old 01 April 2020, 18:52   #1092
grond
Registered User

 
Join Date: Jun 2015
Location: Germany
Posts: 873
Nobody needed 32 address lines until some time this millenium. 24 bit address lines were enough for consumer computers until the mid-90s. A 32 bit data bus, on the other hand, should have been standard starting with ECS. This would have meant to turn the A500+ and the A600 into 020 machines and thus essentially to something very close to the A1200 as we know it. At that time this would have been a significant upgrade.
grond is offline  
Old 01 April 2020, 20:31   #1093
AmigaHope
Registered User
 
Join Date: Sep 2006
Location: New Sandusky
Posts: 753
Quote:
Originally Posted by grond View Post
The point was that NMOS logic circuits are different from CMOS circuits because of all the complementary logic in the pull-up paths that CMOS logic has where NMOS logic only uses resistors or single transistors for load.

However, I have mixed feeling about whether this would have been difficult to change. For sure you can't just shrink the finished layout when transferring it from NMOS to CMOS like you can do when transferring from one CMOS node to the other. If you have actual schematics, converting from NMOS to CMOS should still be mostly monkey work (but a lot of it). You would have to convert the circuitry, dimension the pull-up paths and draw a completely new layout. Not difficult, just a lot of work.
By 1990 the conversion could already be done automatically in Cadence probably some other software packages too. Not even monkey work, just clicking a button and then doing some hand optimizations to the layout. How fast you could clock the resultant chip varied depending on your optimizations as they didn't have good thermal simulations back then, but any good chip design shop could do it to one level of competence or another.

99% of the work would be copying the old hand-drawn schematics into Verilog. (I guess this would be the monkey work)

Quote:
I guess it was like it always was at Commodore: they thought they could get away with doing it cheaply. They needed the millions for the boss's airplane.
THIS was the real problem.
AmigaHope is offline  
Old 01 April 2020, 21:21   #1094
grond
Registered User

 
Join Date: Jun 2015
Location: Germany
Posts: 873
Quote:
Originally Posted by AmigaHope View Post
By 1990 the conversion could already be done automatically in Cadence probably some other software packages too.
As someone who did transistor-level microchip development in Cadence I'm curious to learn how that would have worked. Do you mean by changing an NMOS component library for a CMOS one?
grond is offline  
Old 01 April 2020, 21:38   #1095
duga
Registered User
 
Join Date: Nov 2010
Location: Sweden
Posts: 440
Jumper for 2 + 0 or 1 + 1 MB chip/fast memory. Machine is always delivered with 2 MB memory but can be jumpered differently depending on area of use.

SIMM slot for 1 to 8 MB fast memory (in combination with jumper set to 2 MB chip).
duga is offline  
Old 01 April 2020, 23:13   #1096
StefaanC
Registered User

 
Join Date: Mar 2020
Location: Leest /Belgium
Posts: 26
The amiga 3000 and 4000 are real 32 bit amiga's.
68020 : 32bit processor released in 1984
68030 : 32 bit processor released in 1987

Amiga 1200 : released in 1992 so we can carry on.
68040 : 32 bit processor released in 1990 so 5 years before their bankruptcy.

I must say, the A1200 cpu has data bus 32 bits all MC6800 16 bits
The A1200 has 24 address bus bits, same as 68000

32 bit address bus means or more ram but also expansion boards that have more possibilities.

So the amiga 1200 could have a 6820@14Mhz the a2000: 68020@25Mhz ,the A3000: 68030@33Mhz and the a4000: 68040@50Mhz and the amiga 500+ and 600: the 68EC20@14Mhz
And alse a bit more ram for each higher system

But commodore had to mess up with the hardware that lead to consumers confusion.
And would have also mention less competition with Atari. Their advertising was also not so great
StefaanC is offline  
Old 02 April 2020, 00:01   #1097
grond
Registered User

 
Join Date: Jun 2015
Location: Germany
Posts: 873
There never was an 040 at 50MHz. Also, those CPUs were ridiculously expensive. I think the real problem wasn't the high end but rather sticking to the 7MHz 68000 and 16 bits for seven years.
grond is offline  
Old 02 April 2020, 01:35   #1098
AmigaHope
Registered User
 
Join Date: Sep 2006
Location: New Sandusky
Posts: 753
Quote:
Originally Posted by grond View Post
As someone who did transistor-level microchip development in Cadence I'm curious to learn how that would have worked. Do you mean by changing an NMOS component library for a CMOS one?
SKILL script to replace like-for-like with equivalent logic. You'd get some extra cruft and nonoptimal solutions but you could fix that up with some hand-editing.

EDIT: Should point out that I am not an expert in VLSI, I worked at a VLSI lab (training center attached to a fab that did a lot of second-source runs) but I did systems/network operations and only messed around in Cadence in my off hours. One of the things done was scripted process/die conversion so the fab could produce equivalent ASICs on the process the fab could do.

Last edited by AmigaHope; 02 April 2020 at 01:50.
AmigaHope is offline  
Old 02 April 2020, 15:17   #1099
grond
Registered User

 
Join Date: Jun 2015
Location: Germany
Posts: 873
Quote:
Originally Posted by AmigaHope View Post
SKILL script to replace like-for-like with equivalent logic. You'd get some extra cruft and nonoptimal solutions but you could fix that up with some hand-editing.

OK, that's pretty much what I thought but I would expect that the chipset being bleeding edge technology at the time probably got many hand touches to make it work that would not convert easily.



We should also consider some other historical aspects of the chipset and its design process. It was designed between something like 1983 and 1985. I have no idea what chip design was like in those days. We don't know in what form the technology was transferred from Amiga Inc. to Commodore. I would also not be surprised if Commodore already saved the money for converting whatever 1985-technology they got into some more useful format. They were treating the technology business pretty much like a lollipop business. They had a product they could market and thought they were done. If they had used the time well, transferring the design from NMOS to CMOS would probably have been feasible. If they just sat on layouts and schematics from 1985 for some more years, it probably became very difficult and expensive.
grond is offline  
Old 02 April 2020, 16:08   #1100
Gorf
Registered User

 
Join Date: May 2017
Location: Munich/Bavaria
Posts: 1,157
Quote:
Originally Posted by redblade View Post
How would that work on a A500? Would you piggy back on the chip or use the ZorroII port?

I guess they would have wanted people to upgrade to the Amiga1200
Quote:
Originally Posted by grond View Post
OK, that's pretty much what I thought but I would expect that the chipset being bleeding edge technology at the time probably got many hand touches to make it work that would not convert easily.
even 1984 the Amiga Chipset was far from "bleeding" edge technology-wise!
It was drawn for a 7mu NMOS process ... while Inmos produced it's first samples of their transputer CPU in 1.5mu CMOS in the same year.

Quote:
We should also consider some other historical aspects of the chipset and its design process. It was designed between something like 1983 and 1985. I have no idea what chip design was like in those days. We don't know in what form the technology was transferred from Amiga Inc. to Commodore.
Theoretically everything.
In practice: not so much.
Most of the know-how apparently never left the "Los Gatos Amiga" ... and when it was closed down by C= (1987) many important things were simply lost.

Quote:
....If they just sat on layouts and schematics from 1985 for some more years, it probably became very difficult and expensive.
there where no more layouts and schematics for e.g. Paula ... just the masks
Gorf is offline  
 


Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools

Similar Threads
Thread Thread Starter Forum Replies Last Post
A1200 RF module removal pics + A1200 chips overview eXeler0 Hardware pics 2 08 March 2017 01:09
Sale - 2 auctions: A1200 mobo + flickerfixer & A1200 tower case w/ kit blakespot MarketPlace 0 27 August 2015 19:50
For Sale - A1200/A1000/IndiAGA MkII/A1200 Trapdoor Ram & Other Goodies! fitzsteve MarketPlace 1 11 December 2012 11:32
Trading A1200 030 acc and A1200 indivision for Amiga stuff 8bitbubsy MarketPlace 17 14 December 2009 22:50
Trade Mac g3 300/400 or A1200 for an A1200 accellerator BiL0 MarketPlace 0 07 June 2006 18:41

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +2. The time now is 14:15.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, vBulletin Solutions Inc.
Page generated in 0.13235 seconds with 14 queries