Login
Username:

Password:

Remember me



Lost Password?

Register now!
Main Menu
Who's Online
1 user(s) are online (1 user(s) are browsing Forum)

Members: 0
Guests: 1

more...

Browsing this Thread:   1 Anonymous Users



« 1 (2) 3 4 5 ... 14 »


Re: Vampire hyper threading and aros smp

Joined:
2007/3/13 15:52
From West Midlands, UK
Group:
Member
Posts: 907
Offline
Quote:

rollmeter wrote:
From my POV Apollo team is doing a great job. Let´s start be positive about it.

- Old 68K Amiga computers are compatible with old code, but slow and obsolete now.
- PowerPC Amiga computers are not compatible with old code (only with emulation), relative fast and now arquitecture is obsolete too, outside from server´s market. No future for it.
- X86 is great and fast but no compatible with old code (only through emulation) and difficult to get stable drivers due to the infinite hardware´s providers. Impossible to get a fully compatible driver, so AROS development could become eternal and slow.
- Apollo CPU highly compatible with old code, relative fast and has important enhancements to run nowadays code on it. I think is the perfect gate between old hardware, old software, AROS and new software too.

Lets try to run old code on an i7/i9/Ryzen CPU without emulator, drivers or patches and let´s see what happens... Because there is no fully compatibility too with an old 8086, 80286, 80386 PC....

I understand there is people who have invested thousand of hours developing and working in their own business/solution for the Amiga market and probably they do not see with good eyes Apollo core. But are the users the ones they choose what to use and what not...

And regarding Vampire boards sold... Market says Apollo Core is the way :)

Cheers,
Rollmeter


I'd agree with many of the points above, and certainly as an open and known system AROS can be optimised to be the perfect fit for Vampire/Apollo, but that shouldn't be at the expense of the other branches.

It's difficult to see the 68k solutions ever matching the raw power, low cost point and immediate availability of x86 hardware. Sure, finding compatable systems can be an utter a pain (see my 'New hardware growing pains' thread...) but the advantages outweigh that. Also, this is the hardware with laptops...

Equally it would be nice to see arm branch maturing more - devices like the Pi are so cheap thay can be bought on a whim, but does your average casual useer want to get bogged down in the intricacies of Linux or the bloat of Windows on what's supposed to be a cheap and simple device? That's where AROS could absolutely shine. And if AROS was ever to run on an arm tablet, that would be pretty great too (and despite the thread on another site the UI works fine on a touchscreen, I know because I've done it).

Don't get me wrong, I'm excited too about the potential for AROS on 68k, just so long as it's doesn't become the only focus.

Cheers,
Nigel.

Posted on: 6/25 2:51
Transfer the post to other applications Transfer


Re: Vampire hyper threading and aros smp

Joined:
6/25 1:07
From Madrid
Group:
Member
Posts: 3
Offline
To ntromans:

I agree with the ARM branch development, I just forgot to mention it on my post. ;)

Regarding X86 AROS branch, I think effort has been very high till now ... And cover PC hardware at 100% is not possible and would take tens of years. That's the reason why I think it might be a really good idea and opportunity to focus on the ARM and 68K.

To cdimauro:

I never told Power PC or POWER or customized POWER (OpenPOWER) are outside of server market, there was a comma between "too" and "outside" on my post ;)

I'm not against emulation and I'm agree with you on that, so there is no problem with any abstraction layer for me too. That's why I think there is no problem with new enhacements too... 64 bits, MMU, AMMX, different FPU... All it's ok, an abstraction layer/driver do the work...

Regarding X86 CPUs and hardware compatibility, there are device drivers and software layers managing that compatibility you are talking about too. So, there is not a full compatibility between nowadays X86 CPUs and old ones.

I think competition is good and helps growth. Fair play is good and positive too. I would like to see more of this in the Amiga world hardware / software developers, sellers and users too :)

I do not think there is any wrong affirmation when Apollo team people says Apollo CPU is fully compatible 68K, at least they said they were working on it and still working. That's the reason why I do not have one Vampire board yet, but I will have it.

68K is easy to code, and an enhaced version is good for Amiga world, it's good for other old computers, servers, and industrial equipment using old 68K too. CPUs that can not be replaced by an ARM or X86 board. Maybe it needs more time and work, but resurrected it is.

Anyway, to keep me in the thread title. Congratulations for the hyper-threading effort, and hope to see the 68K AROS version making use of it soon.

Cheers,
Rollmeter.

Posted on: 6/25 7:23
Transfer the post to other applications Transfer


Re: Vampire hyper threading and aros smp

Joined:
2010/8/30 7:20
Group:
Member
Posts: 899
Offline
Quote:

cdimauro wrote:

Apollo is claimed to be 100% 68K compatible, but that's simply not true, since it lacks an FPU and an MMU. I mean, one of the 68K FPUs / MMUs which were used by the Amiga software.


then 68000 is itself not 68k compatible since it lacks fpu and mmu. hell 68020 is not compatible! here already go two cpu models built in most popular amiga hardware, not to talk of few more. we arriva at a conclusion amiga was basically not 68k compatible, according to your reasoning.

but lets not split hairs.
there is at least two factors i as see advantageous in 68k platform for aros as a whole:
1. it allows or rather motivates people to test low end hardware, which eventulally may lead to improvements in cross platform effectivity of the system. we have already seen the results.
2. it may actually improve overall aros popularity, gaining some user base, no matter what platform, and perhaps even attract developers. also this is something we have already seen the results of, with most prominent example being jason and toni.

Posted on: 6/25 9:50
Transfer the post to other applications Transfer


Re: Vampire hyper threading and aros smp

Joined:
2007/8/21 11:21
Group:
Member
Posts: 148
Offline
Quote:

wawa wrote:
Quote:

cdimauro wrote:

Apollo is claimed to be 100% 68K compatible, but that's simply not true, since it lacks an FPU and an MMU. I mean, one of the 68K FPUs / MMUs which were used by the Amiga software.


then 68000 is itself not 68k compatible since it lacks fpu and mmu. hell 68020 is not compatible!


The Apollo-Core homepage lists under features of the Apollo 68080 CPU:
"Fully pipelined, double/extended FPU"
http://www.apollo-core.com/index.htm?page=features

Under products it says both available cards do have the "Apollo 68080 CPU"

Ergo, a customer has to assume, this product has a "100% code compatible" fully pipelined FPU.

This was criticized more than once in the Apollo-forum, but for unknown reasons, the Apollo-team chooses to mislead new customers.

Don´t get me wrong: I will buy one nevertheless, as soon as the standalone is available, but I do not understand such behavior.

Posted on: 6/25 11:09
Transfer the post to other applications Transfer


Re: Vampire hyper threading and aros smp

Joined:
2013/4/21 1:35
From Germany
Group:
Member
Posts: 272
Offline
Quote:

rollmeter wrote:
To ntromans:

I agree with the ARM branch development, I just forgot to mention it on my post. ;)

Regarding X86 AROS branch, I think effort has been very high till now ... And cover PC hardware at 100% is not possible and would take tens of years. That's the reason why I think it might be a really good idea and opportunity to focus on the ARM and 68K.

Covering 100% hardware is not possible, but AROS has a big park of x86/x64 hardware where it can run. Without spending tons of money.
Quote:
To cdimauro:

I never told Power PC or POWER or customized POWER (OpenPOWER) are outside of server market, there was a comma between "too" and "outside" on my post ;)

Which should mean "and", so you've added another "quality" to the PowerPC family.

At least, that's what I've understood.
Quote:
I'm not against emulation and I'm agree with you on that, so there is no problem with any abstraction layer for me too.

Good.
Quote:
That's why I think there is no problem with new enhacements too... 64 bits, MMU, AMMX, different FPU... All it's ok, an abstraction layer/driver do the work...

The 4 things that you've mentioned aren't abstraction layers, but features.

From this PoV they require support, which is quite difficult to achieve (especially asking the developers to go back programming in assembly).

Whereas a regular FPU and MMU have already it (there's existing software which is available).
Quote:
Regarding X86 CPUs and hardware compatibility, there are device drivers and software layers managing that compatibility you are talking about too. So, there is not a full compatibility between nowadays X86 CPUs and old ones.

There's no compatibility layer: the x86 CPUs are still able to run 8086 software without requiring a compatibility layer.

From:

Intel® 64 and IA-32 Architectures
Software Developer’s Manual
Combined Volumes:
1, 2A, 2B, 2C, 2D, 3A, 3B, 3C, 3D and 4,

at pag. 65 (in the PDF):

"The IA-32 architecture supports three basic operating modes: protected mode, real-address mode, and system
management mode. The operating mode determines which instructions and architectural features are accessible:
[...]
Real-address mode — This mode implements the programming environment of the Intel 8086 processor with
extensions (such as the ability to switch to protected or system management mode). The processor is placed in
real-address mode following power-up or a reset."

As you can see, an x86 (and x64 too) processor still wakes-up in the old real-mode, running 8086 instructions.
Quote:
I think competition is good and helps growth. Fair play is good and positive too. I would like to see more of this in the Amiga world hardware / software developers, sellers and users too :)

That's difficult, because the post-Amiga community is split.
Quote:
I do not think there is any wrong affirmation when Apollo team people says Apollo CPU is fully compatible 68K, at least they said they were working on it and still working. That's the reason why I do not have one Vampire board yet, but I will have it.

cybergorf just reported how is the situation: they made/make false statements.
Quote:
68K is easy to code, and an enhaced version is good for Amiga world, it's good for other old computers, servers, and industrial equipment using old 68K too. CPUs that can not be replaced by an ARM or X86 board.

Nobody outside of the retro fans micro-niches are interested in having another 68K processor.

And only insane people can think to start again writing 68K code in assembly, except for small, strategical pieces.
Quote:
Maybe it needs more time and work, but resurrected it is.

You remind me Matt (matthey): another dreamer.

Sorry, but there's no chance for another resurrection.

The 68K ISA was a pleasure to program in assembly (which is NOT the case nowadays), but it had a lot of issues too, which crippled its future. Now it's too late for thinking about a revival, unless for the joy of some retro-fans.
Quote:
Anyway, to keep me in the thread title. Congratulations for the hyper-threading effort, and hope to see the 68K AROS version making use of it soon.

Cheers,
Rollmeter.

IMO a bounty is needed, otherwise I don't see any chance to have it implemented, because current AROS SMP code is bound to x64.

Cheers

Posted on: 6/25 12:11
Transfer the post to other applications Transfer


Re: Vampire hyper threading and aros smp

Joined:
2013/4/21 1:35
From Germany
Group:
Member
Posts: 272
Offline
Quote:

wawa wrote:
Quote:

cdimauro wrote:

Apollo is claimed to be 100% 68K compatible, but that's simply not true, since it lacks an FPU and an MMU. I mean, one of the 68K FPUs / MMUs which were used by the Amiga software.


then 68000 is itself not 68k compatible since it lacks fpu and mmu. hell 68020 is not compatible! here already go two cpu models built in most popular amiga hardware, not to talk of few more. we arriva at a conclusion amiga was basically not 68k compatible, according to your reasoning.

Take a look at what cybergorf reported.
Quote:
but lets not split hairs.

The point which I (and not only me) raised is about useless features (in the Amiga/68K land): 64 bits, incompatible MMU, AMMX, and now HyperThreading.

As I said, there's no software and not even o.s. support for them.

A regular FPU, MMU, and out-of-order execution are much more useful, and immediately usable by already existing software.
Quote:
there is at least two factors i as see advantageous in 68k platform for aros as a whole:
1. it allows or rather motivates people to test low end hardware, which eventulally may lead to improvements in cross platform effectivity of the system. we have already seen the results.

I agree.
Quote:
2. it may actually improve overall aros popularity, gaining some user base, no matter what platform, and perhaps even attract developers. also this is something we have already seen the results of, with most prominent example being jason and toni.

But here I don't agree.

Jason substantially disappeared. About Toni, I've reported a couple of links, and you should know the situation. deadwood (which might have helped on the 68K side) completely disappeared. kalamatee might be interested, but I think that bounties are needed in this case (he makes A LOT of work, and deserves some compensation IMO).

Not even talking about a compiler support for such new features, which is missing (and the plain 68K support just sucks).

I don't have to tell you that the situation is certaily NOT happy, because you follow all post-Amiga forums, and you already know it.

Quote:

cybergorf wrote:
Quote:

wawa wrote:
Quote:

cdimauro wrote:

Apollo is claimed to be 100% 68K compatible, but that's simply not true, since it lacks an FPU and an MMU. I mean, one of the 68K FPUs / MMUs which were used by the Amiga software.


then 68000 is itself not 68k compatible since it lacks fpu and mmu. hell 68020 is not compatible!


The Apollo-Core homepage lists under features of the Apollo 68080 CPU:
"Fully pipelined, double/extended FPU"
http://www.apollo-core.com/index.htm?page=features

Under products it says both available cards do have the "Apollo 68080 CPU"

Ergo, a customer has to assume, this product has a "100% code compatible" fully pipelined FPU.

This was criticized more than once in the Apollo-forum, but for unknown reasons, the Apollo-team chooses to mislead new customers.

Don´t get me wrong: I will buy one nevertheless, as soon as the standalone is available, but I do not understand such behavior.

+1

Posted on: 6/25 12:23
Transfer the post to other applications Transfer


Re: Vampire hyper threading and aros smp

Joined:
2010/8/30 7:20
Group:
Member
Posts: 899
Offline
Quote:

cdimauro wrote:


Jason substantially disappeared.

jason has now a job that doesnt let him contribute.

Quote:

About Toni, I've reported a couple of links, and you should know the situation.


i think i know his position very well. i didnt mention him because of his opinion on apollo, which i understand, but because he works on aros(68k) nonetheless, one simply needs to offload him as much as possible while pursuing some issue, and then he will eventually come up with a solution. we were debugging apollo ide problems lately, that paradoxically let him fix long standing a4000 boot bug.

Quote:

deadwood (which might have helped on the 68K side) completely disappeared.

i told krzysztof from the start these bounties didint have a chance. now i think its real life interferring.

Quote:

kalamatee might be interested, but I think that bounties are needed in this case (he makes A LOT of work, and deserves some compensation IMO).

we will see. he did some sunstantial improvements just last days before the problem surfaced, also for generic amiga-m68k target.

Quote:

Not even talking about a compiler support for such new features, which is missing (and the plain 68K support just sucks).

we have bebbo with his m68k gcc6.3 backend.

Quote:

I don't have to tell you that the situation is certaily NOT happy, because you follow all post-Amiga forums, and you already know it.

actually thanks apollo it is not that bad as it would be otherwise. nge amiga is burning out, unfortunatelly, on all fronts, only 68k sustains at a low level admittedly. thats where ng led us ignoring 68k for so long.

Posted on: 6/25 15:12
Transfer the post to other applications Transfer


Re: Vampire hyper threading and aros smp

Joined:
5/24 4:59
From Norway
Group:
Member
Posts: 33
Offline
cdimauro is right about this being a retro platform, and in mainstream terms its dead. Same can be said for C64 scene, but still its active and you can actually find a minor market as a vendor/software developer.
As Daytona, developer on AOS4.x commented; as long as the money I make from sales covers the coffe and sigarettes required to develop Im happy.

I think its a bit more promising for the Vampire branch. wawa mentioned the bounties didnt stand a chance. That might change with the 1000++ Vampire users. There are resources available in the community if its excited/awoken. Just look at how fast Kalamatee got helped once people realised the severity of the situation.

I have absolutely no illusions about what can be accomplished, but I do think bounties stand a better chance now compared to the past.

Posted on: 6/25 15:33
Transfer the post to other applications Transfer


Re: Vampire hyper threading and aros smp

Joined:
6/25 1:07
From Madrid
Group:
Member
Posts: 3
Offline
Quote:

cdimauro wrote:
Quote:
To cdimauro:

I never told Power PC or POWER or customized POWER (OpenPOWER) are outside of server market, there was a comma between "too" and "outside" on my post ;)

Which should mean "and", so you've added another "quality" to the PowerPC family.


Nop... You are wrong, and trying to modify what I said is not a good way to answer.

Quote:
At least, that's what I've understood.


Yeah, you understood that, but instead of changing your words, you decided to correct me despite not being errors on my statement...

Quote:

Quote:
That's why I think there is no problem with new enhacements too... 64 bits, MMU, AMMX, different FPU... All it's ok, an abstraction layer/driver do the work...

The 4 things that you've mentioned aren't abstraction layers, but features.


Second time... Again you are correcting what it is not wrong and you are telling something everybody knows... Why are you trying to modify what I said? Click to see original Image in a new window

Quote:

Quote:
Regarding X86 CPUs and hardware compatibility, there are device drivers and software layers managing that compatibility you are talking about too. So, there is not a full compatibility between nowadays X86 CPUs and old ones.

There's no compatibility layer: the x86 CPUs are still able to run 8086 software without requiring a compatibility layer.

From:

Intel® 64 and IA-32 Architectures
Software Developer’s Manual
Combined Volumes:
1, 2A, 2B, 2C, 2D, 3A, 3B, 3C, 3D and 4,

at pag. 65 (in the PDF):

"The IA-32 architecture supports three basic operating modes: protected mode, real-address mode, and system
management mode. The operating mode determines which instructions and architectural features are accessible:
[...]
Real-address mode — This mode implements the programming environment of the Intel 8086 processor with
extensions (such as the ability to switch to protected or system management mode). The processor is placed in
real-address mode following power-up or a reset."

As you can see, an x86 (and x64 too) processor still wakes-up in the old real-mode, running 8086 instructions.


Again... You did it again, 3 times now, Incredible... Click to see original Image in a new window

Click to see original Image in a new window

You can attach 1000 page manual if you want... But tell me please... Why are you trying to correct me with arguments that are not related to my words, man...?

Quote:
cybergorf just reported how is the situation: they made/make false statements.


Nop, they don´t... I have been reading Apollo forum posts and statements from the begining and they give what they say.

But now I see the way you read my comments, and how you change and try to modify my words... So I can understand now why you think that about them...

Quote:

Quote:
68K is easy to code, and an enhaced version is good for Amiga world, it's good for other old computers, servers, and industrial equipment using old 68K too. CPUs that can not be replaced by an ARM or X86 board.

Nobody outside of the retro fans micro-niches are interested in having another 68K processor.

And only insane people can think to start again writing 68K code in assembly, except for small, strategical pieces.


Better to say you don´t like it and you are not interested on it. Good... :)

Insane people? Why? Then are you sane despite you modify people´s words and statements? Click to see original Image in a new window

Quote:

You remind me Matt (matthey): another dreamer.

You remind me Pedro Sanchez (Spaniard Politician). He has the ability to change what people say to disqualify them, fighting with them despite nobody is fighting with him and he think he has the only truth, his truth... If he think snow is green, then everybody is wrong...

Fortunately everybody knows about his toxic way of managing conversations, so people prefer preventing problems with him...Click to see original Image in a new window
Click to see original Image in a new window

Cheers,
Rollmeter

Posted on: 6/25 16:29
Transfer the post to other applications Transfer


Re: Vampire hyper threading and aros smp

Joined:
2013/4/21 1:35
From Germany
Group:
Member
Posts: 272
Offline
Quote:

wawa wrote:
[... about contributors ...]

I know the situation that you reported. But, as I've already stated, it's not happy.
Quote:
Quote:

Not even talking about a compiler support for such new features, which is missing (and the plain 68K support just sucks).

we have bebbo with his m68k gcc6.3 backend.

I know this also, but it's still incomplete for the pure 68K family, and maintaining a mega-patch over the regular branch is already not simple (read: it needs to be upstreamed to become easier to handle).

I doubt that someone can add what's missing for supporting the new Apollo features: there's so much work to do.
Quote:
Quote:

I don't have to tell you that the situation is certaily NOT happy, because you follow all post-Amiga forums, and you already know it.

actually thanks apollo it is not that bad as it would be otherwise. nge amiga is burning out, unfortunatelly, on all fronts, only 68k sustains at a low level admittedly. thats where ng led us ignoring 68k for so long.

A niche fighting against another niche for supremacy / leadership / revenge doesn't change the situation: it remains a niche, and will not impact the real world.

I'm pretty sure that Apollo will give a good contribute / toy to the 68K aficionados, but it's still a retro-niche.

P.S. I wouldn't call a port as "NG": there's nothing "NG" on a port of an o.s. to another platform. NG for me means that the new platform solves at least some serious problems of the old one, which is not the case in the post-Amiga land.

@Overflow: bounties might be a solution, but someone should write them, and start putting some money.

Posted on: 6/25 21:03
Transfer the post to other applications Transfer



« 1 (2) 3 4 5 ... 14 »



You can view topic.
You cannot start a new topic.
You cannot reply to posts.
You cannot edit your posts.
You cannot delete your posts.
You cannot add new polls.
You cannot vote in polls.
You cannot attach files to posts.
You cannot post without approval.

[Advanced Search]


Search
Top Posters
1 paolone
paolone
4338
2 magorium
magorium
4095
3 phoenixkonsole
phoenixkonsole
3887
4 nikolaos
nikolaos
3677
5 deadwood
deadwood
2923
6 ncafferkey
ncafferkey
2611
7 mazze
mazze
2204
8 clusteruk
clusteruk
2066
9 Kalamatee
Kalamatee
2023
10 damocles
damocles
1789
© 2004-2017 AROS Exec