English Amiga Board Amiga Lore


Go Back   English Amiga Board > Main > Nostalgia & memories

 
 
Thread Tools
Old 22 June 2017, 20:35   #101
Gorf
Registered User

 
Join Date: May 2017
Location: Munich/Bavaria
Posts: 79
Quote:
Originally Posted by matthey View Post
Let's ignore the 500 pound gorilla in the room though and hope he doesn't want Boing Ball shoes .
Gorilla? We are no pet-shop either!
Gorf is online now  
AdSense AdSense  
Old 22 June 2017, 21:04   #102
matthey
Registered User
 
Join Date: Jan 2010
Location: Kansas
Posts: 1,200
Quote:
Originally Posted by Gorf View Post
Gorilla? We are no pet-shop either!


Pet shop? I thought the Amiga was a circus act. Oh wait, did we avoid that with your time line?
matthey is offline  
Old 22 June 2017, 21:31   #103
Gorf
Registered User

 
Join Date: May 2017
Location: Munich/Bavaria
Posts: 79
By the way: Gunnar has made a first small step towards multiprocessing on Apollo and enabled Hyperthreading.
Once there is software that takes advantage of 2 virtual cores, the demand for a real MP setup will rise.
Gorf is online now  
Old 22 June 2017, 22:54   #104
matthey
Registered User
 
Join Date: Jan 2010
Location: Kansas
Posts: 1,200
Quote:
Originally Posted by Gorf View Post
By the way: Gunnar has made a first small step towards multiprocessing on Apollo and enabled Hyperthreading.
Once there is software that takes advantage of 2 virtual cores, the demand for a real MP setup will rise.
It will be interesting to see the performance evaluated. I'm not as big of a fan of hyper-threading. You take a strong core and allow multiple tasks to share that core giving multiple weaker performing tasks. There is usually an overall benefit in the amount of work done as tasks often spend time waiting (for memory, cache fills, branch misses, etc.) but sometimes multi-threading gives worse performance and it is difficult to determine automatically when it is best to reduce the number of threads or turn it off. It makes sense for applications which consistently have many parallel tasks like a server. For a PC or gaming, I want more and many powerful cores with SMP. Multi-threading should be much cheaper to implement than another core or FPU though.
matthey is offline  
Old 22 June 2017, 23:27   #105
babsimov
Registered User

 
Join Date: Jun 2017
Location: France
Posts: 20
Quote:
Originally Posted by nogginthenog View Post
Note that it's Pentium 3 based, not Pentium 4 :-) P4 was a dead end.
Yes, Pentium 3 based, i make a mistake in my original comment.


Quote:
Originally Posted by matthey View Post
Quote:
:
Originally Posted by babsimov View Post
I have read your post about the Alpha, so you think it's not a really good move for intel to have buy Alpha ? Not as efficient as pentium and even less that Motorola 68060. I understand correctly ?
I believe Intel received some useful technology from the Alpha, especially in regard to fab technology. The cost/benefit of removing a major competitor can't be underestimated either.

There are different measures of efficiency. The Alpha was very efficient by some metrics and very inefficient by others. It takes a certain kind of efficiency (simplicity and streamlining) to be able to clock a CPU that high but then it is not performance or energy efficient to do so.
Intel and Microsoft are famous to buy compagnies for removing it from the market.

About high clock, at the time, the market always ask for more mhz, not carring about energy efficiency, if i remember correctly.

Quote:
Quote:
Originally Posted by babsimov
As you say i remember at the time the Alpha is the faster processor on the market. HP-PA is the second (if i remember correctly).

About the HP-PA, what do you think of it ? Latter HP switch to Intel technology. Do you think the HP-PA can have compete with intel if HP have choose to continue this processor ?

I believe the PA-RISC was more practical than Alpha (ISA and CPU designs). HP was earlier to use an SIMD unit which gives energy efficient parallel processing capabilities and improved code density. Their processors were lower clocked and more energy efficient in general. They did suffer from the large cache needs from poor code density (not as bad as Alpha though) and early cache setups left them with lackluster performance as can be seen by my 68060 vs PA-RISC RiVa performance comparison post. The SIMD unit put the PA-RISC ahead finally but it also had certain limitations which were bad ideas like sharing the integer unit and SIMD unit register files which would not have allowed the SIMD unit width to grow easily beyond 64 bits or floating point to be introduced (as I have been arguing in another thread as a mistake of the Apollo ISA). The last PA-RISC CPU was produced in 2005 but these ISA limitations were never addressed.
So, after reading your answer, i think Commodore ingeneers have make good choice with the HP-PA for Hombre.

And, about the SIMD unit (Max 1 instruction ?), they were designed partialy by Dr Helpler for Hombre and integrated in the HP-PA Hombre core, if have i correctly understand some of his interview. HP use this for himself after (i guess).


Quote:
Originally Posted by Gorf View Post
@babsimov
I did some research:
In 1996 between 70.9 - 72.2 million personal computers where shipped.
So what marketshare would we have?
Apple got around 6%. Strong in USA but almost nowhere else.

We are traditionally strong in Europe, had some comeback in the US and took the opportunities of new markets in eastern Europe, South America and China.

If we reach more than 11% we are the best selling company, before Compaq with 10% and IBM with 8.6%.
Does not forget that IBM, Compaq and all the others PC compatible manufacturers add up their market shares in the end.

If Amiga is 11 %, MAC 6 %, PC is 83 %, so Amiga is not a leader.

This is why in one of my timeline (a world without PC) i make the PC compatible disapears from the market in the end 80. But they are 68000 based, not X86 in that timeline. So MAC, ST and Amiga are compatible and far better for less money (expect for the MAC that is not better and more expensive). In this timeline, Apple collaspe in the mid 90, just a little time after going PPC.

Commodore choose the HP-PA way, and Atari go PPC (and NeXT also).

Commodore buy HP and continue the HP-PA (multicore) to today. Commodore (Amiga) have more or less 50 % of the market at this time, and since some years ago. The Amiga have many first for technology, so they are leader.

Atari is second with Falcon PPC (multicore) like computers (with 3DFX in house for the 3D chipset).

NeXT is in third place, with PPC (multicore) and Nvidia in house for 3D chipset.

AMD make PPC compatible processor (after making 68k compatible in the 80/90)
babsimov is offline  
Old 23 June 2017, 01:19   #106
kovacm
Registered User
 
Join Date: Jan 2012
Location: Serbia
Posts: 202
Quote:
Originally Posted by babsimov View Post
Intel and Microsoft are famous to buy compagnies for removing it from the market.
This practice is so notorious that it had a name: Embrace, extend and extinguish https://en.wikipedia.org/wiki/Embrac...and_extinguish

Microsoft made more damage to planet, society and prosperity than all wars, corruptions, pollution in past half century combined!
kovacm is offline  
Old 23 June 2017, 02:06   #107
Gorf
Registered User

 
Join Date: May 2017
Location: Munich/Bavaria
Posts: 79
Quote:
Does not forget that IBM, Compaq and all the others PC compatible manufacturers add up their market shares in the end.

If Amiga is 11 %, MAC 6 %, PC is 83 %, so Amiga is not a leader.
I am aware of this.

In my calculations 11% is the minimum we must have reached by now - the maximum at 17%.
But this is only the PC market: we had the best selling console - the AmigaPS - for 4 years now. and it is basically the same hardware and the same OS.

So if we look at a combined console+PC market, we would have probably a share of 22-26%.
Gorf is online now  
Old 23 June 2017, 04:39   #108
matthey
Registered User
 
Join Date: Jan 2010
Location: Kansas
Posts: 1,200
Quote:
Originally Posted by babsimov View Post
About high clock, at the time, the market always ask for more mhz, not carring about energy efficiency, if i remember correctly.
The high end CPU markets want performance. Turning up the clock is easy but generates heat faster than performance. This is why parallelism and power efficiency (performance/Watt) became so important.

Quote:
Originally Posted by babsimov View Post
So, after reading your answer, i think Commodore engineers have make good choice with the HP-PA for Hombre.

And, about the SIMD unit (Max 1 instruction ?), they were designed partialy by Dr Helpler for Hombre and integrated in the HP-PA Hombre core, if have i correctly understand some of his interview. HP use this for himself after (i guess).
Judging by the year of Hombre, it looks like the SIMD unit would have been MAX-1 which only supported 2x16 bit integer operations and is primitive compared to today's SIMD units. The 64 bit PA-RISC 2.0 ISA expanded the integer/SIMD registers to 64 bits for MAX-2 (still primitive but the same size as MMX/AMMX) which now only did 4x16 bit integer SIMD operations. Perhaps C= had early PA-RISC processors with MAX-2. I haven't read the recently released Hombre documentation. I expect the instructions added by C= would have been custom SIMD instructions as they would have made the biggest difference for graphics processing.

Quote:
Originally Posted by kovacm View Post
This practice is so notorious that it had a name: Embrace, extend and extinguish https://en.wikipedia.org/wiki/Embrac...and_extinguish

Microsoft made more damage to planet, society and prosperity than all wars, corruptions, pollution in past half century combined!
No doubt Microsoft has received anti-trust punishments which make white collar crimes pay. Intel has not been as arrogant but they do leverage their size and position well. Let's not be biased about their products though. They do make high quality chips and the x86 and even Atom have ugly ISAs but they still perform reasonably well in the performance/watt category (an indicator that the CPU can be scaled up in performance with more cores). I found another article with an 8 core Atom used in servers which gave competitive performance in performance/watt as predicted in the "The final ISA showdown" article I linked previously.

http://www.anandtech.com/show/8357/e...r-platforms/17

I made a picture of where the 68k could be in raw average energy (like performance/watt) if a 68k CPU was the same performance but 42% more power efficient than the in order Atom (the 68060 was about the same performance and used 42% less max power than the most comparable Pentium). I have attached the image which shows the 68k CPU could be the overall best in performance/watt compared to several ARM and x86 processors (lower is better). The OoO ARM Cortex-A9 gives energy efficient performance but requires more cores and caches to scale up to match the more powerful in order CISC cores which are better for performance tasks like gaming.
Attached Thumbnails
Click image for larger version

Name:	CPUAverageEnergy.png
Views:	19
Size:	24.8 KB
ID:	53512  

Last edited by matthey; 23 June 2017 at 23:36.
matthey is offline  
Old 23 June 2017, 12:13   #109
idrougge
Registered User
 
Join Date: Sep 2007
Location: Stockholm
Posts: 2,587
Quote:
Originally Posted by matthey View Post
Diversifying into faster growing and more defensive markets like embedded is an excellent way to improve survivability. The PC market is inconsistent with popularity hit and miss, general economic fluctuations and dips between product launches giving cash flow a roller coaster ride. If you want to weather the PC market, I recommend transitioning from a narrow mote business to a wide mote business. Perhaps you would rather be a high stakes gambler and roll the dice until you lose everything though?
Yeah, look at how the Acorn Archimedes/RiscPC survivedů
idrougge is offline  
Old 24 June 2017, 01:07   #110
matthey
Registered User
 
Join Date: Jan 2010
Location: Kansas
Posts: 1,200
Quote:
Originally Posted by idrougge View Post
Yeah, look at how the Acorn Archimedes/RiscPC survived…
Acorn was a business success. The owners (which changed) and management chose to diversify into more profitable business segments and drop less profitable ones (the PC market is cyclical and they were caught in a downturn in the economy). This is a pattern for a growth company where you grow as fast as possible and sell pieces when there is a good opportunity. They could have kept their divisions, diversified and focused on cash flow thus becoming a conglomerate rewarding their investors with dividends instead (it is still necessary to cull dinosaur divisions). Acorn chose the growth business method which is often less kind to customers and employees but they did a super job of innovating, finding new markets and increasing shareholder value. Creating the ARM processor took great vision and was not easy to do at that time. The resulting ARM Holdings rewarded Acorn investors immensely as they went into the embedded market. I don't know if ARM Holdings had access to RISC OS but it is not as suitable for RTOS as the AmigaOS which has preemptive multitasking and a smaller footprint. ARM eventually obtained their own RTOS called RTX.

https://www.arm.com/products/tools/s...ing-system.php

Check out the features of RTX and then compare them to AmigaOS and RISC OS. The AmigaOS resembles a modern RTOS more than it does a modern desktop OS. RISC OS has some advantages which are great for embedded use also like open sources, modularity, fast boot and fast, affordable and energy efficient new hardware (with ARM CPU still). I expect RISC OS is being used more for embedded applications than the AmigaOS is right now with the current owners hiding it away.
matthey is offline  
Old 24 June 2017, 15:33   #111
babsimov
Registered User

 
Join Date: Jun 2017
Location: France
Posts: 20
Quote:
Originally Posted by kovacm View Post
This practice is so notorious that it had a name: Embrace, extend and extinguish https://en.wikipedia.org/wiki/Embrac...and_extinguish

Microsoft made more damage to planet, society and prosperity than all wars, corruptions, pollution in past half century combined!
Thanx for the link, very interesting.

Quote:
Originally Posted by Gorf View Post
I am aware of this.

In my calculations 11% is the minimum we must have reached by now - the maximum at 17%.
But this is only the PC market: we had the best selling console - the AmigaPS - for 4 years now. and it is basically the same hardware and the same OS.

So if we look at a combined console+PC market, we would have probably a share of 22-26%.
But the professionnal market don't like to buy computer that can be see "as a console first".

It is why i don't like the Commodore strategy. They let Amiga going with the "console" tag. Lot of "professionnal" think Amiga is only a console, not a computer that can do serious work.

So expect for the video and earlier multimedia, Amiga don't enter in office. And we know now that THIS IS the market that make a computer the main standard.

So 22 to 26 % with console, as i see it, can't make Amiga defeat the PC. If i understand correctly, it is the main platform in your timeline as it is now ?

I like what you have create with your timeline, expect that PC is the main platform. I have many times dreamed of a world were Amiga is the main platform and PC out.

Quote:
Originally Posted by matthey View Post
The high end CPU markets want performance. Turning up the clock is easy but generates heat faster than performance. This is why parallelism and power efficiency (performance/Watt) became so important.
Your right for the 2000 years, but for the 90 this is mhz that the peoples wants.

But i agree, performance/watt is the way to go.


Quote:
Judging by the year of Hombre, it looks like the SIMD unit would have been MAX-1 which only supported 2x16 bit integer operations and is primitive compared to today's SIMD units. The 64 bit PA-RISC 2.0 ISA expanded the integer/SIMD registers to 64 bits for MAX-2 (still primitive but the same size as MMX/AMMX) which now only did 4x16 bit integer SIMD operations. Perhaps C= had early PA-RISC processors with MAX-2. I haven't read the recently released Hombre documentation. I expect the instructions added by C= would have been custom SIMD instructions as they would have made the biggest difference for graphics processing.

As i understand it, MAX 1 was designed for the HP-PA included in Hombre, by Dr Helpler. After Commodore is out, HP salvage this for their HP-PA (the 7100LC is the first to include it.
babsimov is offline  
Old 24 June 2017, 16:11   #112
Gorf
Registered User

 
Join Date: May 2017
Location: Munich/Bavaria
Posts: 79
Quote:
Originally Posted by babsimov View Post
But the professionnal market don't like to buy computer that can be see "as a console first".
True to some point. We did a lot to improve our standing as a office-computer, e.g. by buying Applix and developing the first true all-plattform-office-suite.
Or by supporting the flicker free productivity-screenmode in Amber since 1989.

But to survive we needed numbers and 1992 was an excellent year to start with a 32-bit cd-equipped console. If you look back in history it is clear, that there was a big gap between 16-bit consoles and the arrival of the Sony PS.
We filled that gap.

Quote:
It is why i don't like the Commodore strategy. They let Amiga going with the "console" tag. Lot of "professionnal" think Amiga is only a console, not a computer that can do serious work.
The AmigaPS gave us the money we needed to buy Wang Laboratories. This acquisition made us finally a big player in the server and office market.

Quote:
So expect for the video and earlier multimedia, Amiga don't enter in office. And we know now that THIS IS the market that make a computer the main standard.
It is a long and hard battle, but we fight it.
We are strong on multimedia, digital video and photo, DTP, databases (Wang), servers, networking and internet. The classical office is "surrounded" by us: as soon as you need one of the above features in addition to word and excel, you are better off with an Amiga.
And through our OLE patents we forced MS to port Office to AmigaOS and use our IFF-format to save files

Quote:
So 22 to 26 % with console, as i see it, can't make Amiga defeat the PC. If i understand correctly, it is the main platform in your timeline as it is now ?
It is hard for a single company to get a bigger share. We fight against many.
But this can also be a good thing:
Look at Apple iOS vs. Android.
There are many companies selling Android phones - but actually only one of them in making some money (Samsung) - all others are so much fighting each other, that no one actually earns money!

And Apple? they do not license iOS. They provide hard- AND software and while only having 20% marketshare Apple gets 80% of all profit.

Ergo: as long as we make profit, 25% are healthy and enough to provide a good ecosystem for programmers and users.

And unlike Apple, we will not just sit on that money: we reinvest almost everything in research and development.

Quote:
I like what you have create with your timeline, expect that PC is the main platform. I have many times dreamed of a world were Amiga is the main platform and PC out.
Our new mystery CPU will certainly turn the table a little bit in our favor.

Quote:
Your right for the 2000 years, but for the 90 this is mhz that the peoples wants.
But i agree, performance/watt is the way to go.
And our new CPU will provide both!
Gorf is online now  
Old 24 June 2017, 17:06   #113
babsimov
Registered User

 
Join Date: Jun 2017
Location: France
Posts: 20
Quote:
Originally Posted by Gorf View Post
True to some point. We did a lot to improve our standing as a office-computer, e.g. by buying Applix and developing the first true all-plattform-office-suite.
Or by supporting the flicker free productivity-screenmode in Amber since 1989.

But to survive we needed numbers and 1992 was an excellent year to start with a 32-bit cd-equipped console. If you look back in history it is clear, that there was a big gap between 16-bit consoles and the arrival of the Sony PS.
We filled that gap.
Yes, but it's console market, not a computer.

I have miss the amber since 1989. But remember an interview of Jay Miner from 1988. He was asked "do you think Amiga can compete to PC (as the main platform). And he respond "No, i think it's too late". At the time i don't understand why he have said that. The Amiga is ahead from PC in many fields. But, he was right, now i see it.

But, you right, Commodore do a far better job with Amiga in your timelime. At the time, if Amiga have 11 % of the market, it have made me very happy.

Quote:
The AmigaPS gave us the money we needed to buy Wang Laboratories. This acquisition made us finally a big player in the server and office market.
i miss that too, sorry.

Quote:
It is a long and hard battle, but we fight it.
We are strong on multimedia, digital video and photo, DTP, databases (Wang), servers, networking and internet. The classical office is "surrounded" by us: as soon as you need one of the above features in addition to word and excel, you are better off with an Amiga.
And through our OLE patents we forced MS to port Office to AmigaOS and use our IFF-format to save files
As i see it, it's better to try to buy Microsoft and scrap all his products, best way to make PC disapear from the market.

The only software i know that Miscrosft have do for Amiga is an unoptimised software (AmigaBasic). They care nothing about optimizing and using Amiga.

I remember that native word processing software by indepent compagnies are always faster and better than what PC compagnies do on Amiga (wordperfect is what i have in mind)

Quote:
It is hard for a single company to get a bigger share. We fight against many.
Microsoft had in fact 90 % of the computer market since years. So it is possible, but of course with a not so nice strategy.

Quote:
But this can also be a good thing:
Look at Apple iOS vs. Android.
There are many companies selling Android phones - but actually only one of them in making some money (Samsung) - all others are so much fighting each other, that no one actually earns money!

And Apple? they do not license iOS. They provide hard- AND software and while only having 20% marketshare Apple gets 80% of all profit.

Ergo: as long as we make profit, 25% are healthy and enough to provide a good ecosystem for programmers and users.

And unlike Apple, we will not just sit on that money: we reinvest almost everything in research and development.
I don't really follow the mobile market, but i see what you mean.

research and development, i like it. In the beginning Commodore do it a lot. It's later they reduce it each years.

Quote:
Our new mystery CPU will certainly turn the table a little bit in our favor.
And our new CPU will provide both!
so, i'm impatient to see what come next

Last edited by babsimov; Yesterday at 11:48.
babsimov is offline  
Old Yesterday, 02:48   #114
idrougge
Registered User
 
Join Date: Sep 2007
Location: Stockholm
Posts: 2,587
Quote:
Originally Posted by matthey View Post
Acorn was a business success. The owners (which changed) and management chose to diversify into more profitable business segments and drop less profitable ones (the PC market is cyclical and they were caught in a downturn in the economy). This is a pattern for a growth company where you grow as fast as possible and sell pieces when there is a good opportunity. They could have kept their divisions, diversified and focused on cash flow thus becoming a conglomerate rewarding their investors with dividends instead (it is still necessary to cull dinosaur divisions). Acorn chose the growth business method which is often less kind to customers and employees but they did a super job of innovating, finding new markets and increasing shareholder value.
Yes, Acorn chose to please their investors. The very same thing could happen to a Commodore which happened to survive to the year 1998, pleasing the stock market by dropping the risky and unprofitable computer market in favour of embedded technology which you said was the way to go.
idrougge is offline  
Old Yesterday, 20:24   #115
matthey
Registered User
 
Join Date: Jan 2010
Location: Kansas
Posts: 1,200
Quote:
Originally Posted by babsimov View Post
I have miss the amber since 1989. But remember an interview of Jay Miner from 1988. He was asked "do you think Amiga can compete to PC (as the main platform). And he respond "No, i think it's too late". At the time i don't understand why he have said that. The Amiga is ahead from PC in many fields. But, he was right, now i see it.
It is amazing that the IBM PC started with a weak 8 bit 8088 introduced the same year (1979) as the 68000. The biggest advantages the IBM PC had was that it was very open (because the U.S. DOJ was breathing down IBM's neck?) and had a company behind it with a good reputation for (business) computers. Somehow the turd improved and reached a critical mass which allowed it to dominate the market and even put superior technologies out of business with economies of scale (mainly Intel with the CPU). Most U.S. consumers chose a cheap open turd which turned into a black hole which sucked the life out of computing as the world entered the dark ages of computing.

Even if Wolfenstein and Doom had come out for the Amiga first, it was probably already too late. The IBM compatible PC may have already had too much market share by the time the Amiga came out (Jay Miner was insightful). Maybe IBM would have chose the 68k if Motorola had brought out the 68008 at the same time or soon after the 68000 but they may have been worried about supply still. Maybe this was the destiny of the 68k and Amiga only to be changed by the end of Moore's law, the invisible hand and the next great visionary, if only we could recognize him.

Quote:
Originally Posted by idrougge View Post
Yes, Acorn chose to please their investors. The very same thing could happen to a Commodore which happened to survive to the year 1998, pleasing the stock market by dropping the risky and unprofitable computer market in favour of embedded technology which you said was the way to go.
The PC market is cyclical (economic ups and downs) and the profit margin has generally shrunk over the years. This does not mean it is "unprofitable" but it has caused consolidation and bankruptcies as the remaining players have become bigger and more efficient. It is important to realize this if you plan to survive. Diversifying into more defensive (consistent) markets can reduce the business risks by providing consistent cash flow for R&D and product startup during economic downturns and under-performing product introductions. It is not necessary to exit the PC market and it can have synergies with other higher profit margin markets and provide advertising. Embedded is just one huge market I identified as having synergies with the Amiga. Gorf also identified several related service, media content, and niche embedded markets which could be entered to diversify. Technology is tricky and fast changing with many grow fast and burn out businesses. Mixing conservative business strategies with flexibility and open minded ideas may have been what Commodore lacked. They certainly should have understood their product better and cut the fat cat payouts quicker.
matthey is offline  
AdSense AdSense  
 


Currently Active Users Viewing This Thread: 2 (0 members and 2 guests)
 
Thread Tools

Similar Threads
Thread Thread Starter Forum Replies Last Post
Alternate monitor BanisterDK support.Hardware 4 12 January 2017 22:42
Amiga timeline TroyWilkins Nostalgia & memories 23 05 September 2016 15:30
Timeline Yesideez Amiga scene 1 13 September 2007 08:12
Magazine cover artwork = Timeline? alexh AMR suggestions and feedback 1 28 May 2007 02:04
CAPS Release Timeline fiath project.SPS (was CAPS) 10 29 June 2004 17:10

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 08:58.


Powered by vBulletin® Version 3.8.8 Beta 1
Copyright ©2000 - 2017, vBulletin Solutions, Inc.
Page generated in 0.17490 seconds with 14 queries