error from motorola ? switching 68k to ppc

[meynaf]

Originally Posted by turrican3

if you had 1,000,000,000 $ could it be for you a good goal to make a new 68k compatible amiga but with the power of a new intel i7 ??
Do you have an idea of what could be better between a powerfull 68k and corei7 ???
Could have been the present better in your pov ??
All of this is theorycall.

I sure would prefer to code on a powerful 68k rather than a core i7. But i don't think even a billion $$$ would change something in todays market and it would either be a niche or doomed to failure.
Anyway, even if i had the means, i wouldn't do a 68k compatible amiga. I would design my own.

[tero]

Originally Posted by Hewitson

The only way that's happening is if AmigaOS is ported to x86/x64 with a 68k emulation layer. Something that should have been done years ago, really. But no, a bunch of complete morons decided PPC was the way to go.

That's what morphos is doing. They showed an early version in Neuss/Amiga34.

[Hedeon]

Originally Posted by tero

That's what morphos is doing. They showed an early version in Neuss/Amiga34.

He said years ago. So these maybe qualify being in the same bunch

[meynaf]

Originally Posted by Hewitson

The only way that's happening is if AmigaOS is ported to x86/x64 with a 68k emulation layer. Something that should have been done years ago, really. But no, a bunch of complete morons decided PPC was the way to go.

x86 in 1994 weren't powerful enough to emulate 68k properly. It would have been a massive regression.

[turrican3]

Originally Posted by meynaf

I sure would prefer to code on a powerful 68k rather than a core i7. But i don't think even a billion $$$ would change something in todays market and it would either be a niche or doomed to failure.
Anyway, even if i had the means, i wouldn't do a 68k compatible amiga. I would design my own.

In theory meynaf, do you think a 68k processor faster than today cpu could be done ?? if yes, why ?? And please forget just for 2 sec the market and the cost.
Do you think only on paper that it could be done ??
2. what do you think of the progress of intel cpu in the last 10 years vs during 95 -> 2010 periode ?? I don't talk about gfx card, their enhancement are fine, but i feel that the cpu of the pc has slowed those last 10 years ??
What's your thought about it ??

[meynaf]

Originally Posted by turrican3

In theory meynaf, do you think a 68k processor faster than today cpu could be done ?? if yes, why ?? And please forget just for 2 sec the market and the cost.

Yes, a 68k processor faster than today cpu could be done, and it would be faster for the simple reason it has better instruction set. So everything else being equal (strong out-of-order implementation at same quality of todays i7, same factories used, etc), the simple fact of having to execute less instructions to perform the same task, while in the same time having as many instructions per cycle, would make it faster.

Originally Posted by turrican3

Do you think only on paper that it could be done ??

Only on paper because it's not going to happen. It is technically possible, but this alone does not decide.

Originally Posted by turrican3

2. what do you think of the progress of intel cpu in the last 10 years vs during 95 -> 2010 periode ?? I don't talk about gfx card, their enhancement are fine, but i feel that the cpu of the pc has slowed those last 10 years ??
What's your thought about it ??

The cpu itself, no.
But it is true that software is more and more bloat as time passes, and this does not appear to be slowing down.
While at the same time they reach more and more limits and current cpu power does not rise as fast as it used to.

[turrican3]

Thank you Meynaf,
I appreciate your direct answer.
We think the same, but you, you can explain why.
I always felt that we should be more advanced now,
I can keep a pc now at least 5 yaers, perhaps changing the gfx card
one time. By the pas i kept it 2 years because the cpu changed so fast,
it'sn't the case anymore. And for the moment Multi Processor is not the answer hoped.
Sadly like you said it certainly won't change, we have to hope that they will use MP better.
Or that a new technology will happen, do you see a new cpu which could change things or a possible evolution which could shake the world ??
Like the atari st and the amiga did or the pentium doubling frequency really fast or the 1st 3d accelerated cards 3dfx, etc...
The cpu evolution (for me) is boring now.

Originally Posted by meynaf

Yes, a 68k processor faster than today cpu could be done, and it would be faster for the simple reason it has better instruction set. So everything else being equal (strong out-of-order implementation at same quality of todays i7, same factories used, etc), the simple fact of having to execute less instructions to perform the same task, while in the same time having as many instructions per cycle, would make it faster.



Only on paper because it's not going to happen. It is technically possible, but this alone does not decide.



The cpu itself, no.
But it is true that software is more and more bloat as time passes, and this does not appear to be slowing down.
While at the same time they reach more and more limits and current cpu power does not rise as fast as it used to.

[meynaf]

Originally Posted by turrican3

Or that a new technology will happen, do you see a new cpu which could change things or a possible evolution which could shake the world ??

I don't see anything good coming in the foreseeable future.

Originally Posted by turrican3

The cpu evolution (for me) is boring now.

It has been boring for me since 1994.

[modrobert]

Originally Posted by meynaf

I don't see anything good coming in the foreseeable future.

QPU?

CPU = Central Processing Unit
GPU = Graphics Processing Unit
QPU = Quantum Processing Unit

Because we all want to cheat like nature by predicting angles of possible paths instead of crunching numbers old school, and O(sqrt(N)) is so much sweeter than O(N) according to Grover's algorithm.

The reason this future seems plausible is because some clever scientist realized that funding is easy if you claim the new tech can break encryption, regardless if the statement is true or not.

https://en.wikipedia.org/wiki/Timeli...ntum_computing

[meynaf]

Originally Posted by modrobert

QPU?

Does it provide anything that will make programming easier ?
(Same reason ppc didn't qualify as 68k replacement, as its asm is actually much worse.)

Can it be made reliable enough so it can provide exactness in the same way as classical computers ?
(Actually even for simplest of tasks they need heavy error correction.)

Can it work at normal temperature like todays computers ?
(Near absolute zero cooler at home would be very funny to have...)

Can it be adapted to any algorithm so it can replace classical computers ?
(I think it can't, even on the most optimistic paper.)

If answer to any of the above is 'no' then for me it does not qualify as "anything good"...

Originally Posted by modrobert

The reason this future seems plausible is because some clever scientist realized that funding is easy if you claim the new tech can break encryption, regardless if the statement is true or not.

And the statement is more or less already wrong. Encryption is ready to face quantum computing, or at least, it is preparing itself for it and is likely to be ready before.
https://csrc.nist.gov/Projects/Post-...m-Cryptography

In addition, the reason you mentioned for a plausible future does not look right to me because if they really reach the goal, they will lose their funding... so currently the best approach is lots and lots of claims and theoretical work, with some apparently successful experiment here and there, to make people believe in it - and this is exactly what we have today.

Anyway, if asked i wouldn't bet on quantum computer at all - actually i would bet against it.
I think one day they will find a limit, some basic law of nature they didn't know or anticipate, that they simply cannot bypass, and it will remain a lab's curiosity forever. This is just my view on it, it may be wrong of course. But remember nobody can claim to have fully understood quantum physics so who knows what is lurking...

[Glen M]

Originally Posted by meynaf

and current cpu power does not rise as fast as it used to.

This graph would indicate to me that Moore's law still hold true

https://en.wikipedia.org/wiki/Moore%..._1971-2018.png

As for the original question I'm not so sure moving to PPC would have changed anything. Keep your timelines in check and remember that circa 1992/93 when commodore released their last machines the motorola flag ship was the 68040 at 40mhz. The best intel chip of the time was the 486 dx2 at 50mhz. The motorola at 25mhz can grind out 28MIPS . The 486dx2 @ 66mhz can only achieve 30MIPS so it seems pretty obvious to me that the speeds of the processors was not the problem. In fact you can almost understand why apple and commodore stayed with motorola, they were faster after all.

Figures quoted above obtained from - https://gamicus.gamepedia.com/Instructions_per_second

It would not be until the Pentium that Intel would pull away but that's not for a year or 2 yet on the above time frame and in the 90s a year or 2 was the lifespan of anything computer related.

[meynaf]

Originally Posted by Glen M

This graph would indicate to me that Moore's law still hold true

https://en.wikipedia.org/wiki/Moore%..._1971-2018.png

Transistor count and cpu speed are two different things.

[Glen M]

Originally Posted by meynaf

Transistor count and cpu speed are two different things.

Good point. In terms of physical clock speed that seems to have levelled of a fair bit but with increase in transistor count is there not also an equal increase in speed as operations per clock cycle increase. It's been 20 years since I had a brief look at CPU design in tech so please excuse if that's a stupid question.

[turrican3]

sp cpus brute force haven't evolutate so much those last 10 years... That's why they multiply the cores but 2 cores doesn't mean 2xspeed, and if the code isn't done for mp, it means 1xspeed.
Beleive me, i never kept a pc so much years than now !!!
I will have to change it surely in 2020, and this time completely.
How i would like to see more competions in the pc market !!!
My only hope, a big, big society like sony or else, coming shaking up everything... But actually even the biggest seem afraid to do that.
Someone could explain howmuch and why the market is so locked ??

[Glen M]

Originally Posted by turrican3

sp cpus brute force haven't evolutate so much those last 10 years... That's why they multiply the cores but 2 cores doesn't mean 2xspeed, and if the code isn't done for mp, it means 1xspeed.
Beleive me, i never kept a pc so much years than now !!!
I will have to change it surely in 2020, and this time completely.
How i would like to see more competions in the pc market !!!
My only hope, a big, big society like sony or else, coming shaking up everything... But actually even the biggest seem afraid to do that.
Someone could explain howmuch and why the market is so locked ??

I think you'd be hard pushed to find a new peice of software that doesn't utilise multiple cores but the problem is plenty of us are still running old stuff. In saying that single core performance is generally more than fast enough for any legacy software that needs it.

The reason your PC is still going strong is that hardware performance has been at a high enough level for all day to day tasks for years and it doesn't need to get any faster to browse the web, watch youtube, stream netflix, send e-mails etc. I have a core2 duo rig lying in front of the TV in the man cave and it'll happily stream 1080p all day long.

I would build systems for family and friends generally around Christmas time and my recommendation for a budget gaming rig is always to buy the low end of the current motherboard, CPU, ram etc and last generations GPU. There are so many system builders on the likes of facebook and gumtree that are still pushing the old i5 or i7 socket but with that there is no room for future upgrade. You can put together a £400 ryzen build that will run everything these days and give you scope to drop in 32gb of ram and a faster cpu in a few years to keep things fresh.

I don't see there being any fresh competition these days though. If Sony entered the desktop market it'd just be off the shelf parts probably an i5, GTX2060, 16gb DDR4 and an SSD and I only use intel and nvidia parts because that's what must lay people see and think performance. For a system I built for a friend of my wife last year I had try hard to convince her that the RX580 was faster than a GTX960 despite costing the same and having a lower number. Its the same argument with AMD vs Intel CPUs.

Despite there being really only 3 companies (Intel, AMD, Nvidia) I wouldn't call the market locked as there is plenty of scope for any budget throughout the range of products from them.

[robinsonb5]

Originally Posted by Glen M

Good point. In terms of physical clock speed that seems to have levelled of a fair bit but with increase in transistor count is there not also an equal increase in speed as operations per clock cycle increase.

Yes, but that increase is acheived by shovelling in more cores rather than improving the throughput of a single core.

Even if improving single-core performance were easily achievable, there's not a lot driving it right now; as you said, per-core performance has been more than adequate for most tasks for a very long time now. The emphasis is very much on multiple cores for high-throughput server loads, and best performance/power balance for the mobile space; the desktop user's experience isn't really improving much because (javascript-infested-swamps-masquerading-as-websites-yes-I'm-looking-at-you-facebook notwithstanding) it doesn't really need to.

[Megol]

Originally Posted by Hewitson

I'd have preferred to see the 68k series of CPU's achieve faster speeds and more capabilities, but this had absolutely NOTHING to do with Commodore going bankrupt.

Absolutely, Commodore were skilled in making bad choices for short term wins (if any).

Saved by Bill Gates? More like killed by Bill Gates.

Apple had a very bad economical situation leading to cooperation with Microsoft, don't you remember Steve Jobs presenting Bill Gates on stage to the surprise of everyone? Dislike MS and Gates but don't try to rewrite history.

[tero]

Originally Posted by Megol

don't you remember Steve Jobs presenting Bill Gates on stage to the surprise of everyone?

Didn't Gates only support/paid apple to stop the lawsuit against microsoft because gates stole ideas from the macintosh?
That lawsuit was going on quite some time and stopped after that.

[Hewitson]

Originally Posted by meynaf

x86 in 1994 weren't powerful enough to emulate 68k properly. It would have been a massive regression.

I'm sure a 486 DX4/100 could have emulated a 68k with ease.

[meynaf]

Originally Posted by Hewitson

I'm sure a 486 DX4/100 could have emulated a 68k with ease.

Old 68000 maybe (i'm not sure). But in 1994 the speed to achieve was that of 68040.