OSDev.org

The Place to Start for Operating System Developers
It is currently Fri May 26, 2017 1:05 pm

All times are UTC - 6 hours




Post new topic Reply to topic  [ 131 posts ]  Go to page Previous  1 ... 3, 4, 5, 6, 7, 8, 9  Next
Author Message
 Post subject: Re: How to make an operating system to x86 within a month fr
PostPosted: Mon Apr 17, 2017 11:22 am 
Offline
Member
Member
User avatar

Joined: Tue Aug 02, 2016 1:52 pm
Posts: 272
Location: East Riding of Yorkshire, UK
Geri wrote:
ah dont misunderstand me, i know its history until the roman invasion

Don't worry, it can't be any worse than history education in the UK.

_________________
com.sun.java.swing.plaf.nimbus.InternalFrameInternalFrameTitlePaneInternalFrameTitlePaneMaximizeButtonWindowNotFocusedState
Compiler Development Forum


Top
 Profile  
 
 Post subject: Re: How to make an operating system to x86 within a month fr
PostPosted: Mon Apr 17, 2017 11:24 am 
Offline
Member
Member
User avatar

Joined: Sun Jul 14, 2013 6:01 pm
Posts: 364
yeah i have read from that long time ago. it was interesting. isnt now negotiations about reunifying the island as a free state?

_________________
Operating system for SUBLEQ cpu architecture:
http://DawnOS.tk


Top
 Profile  
 
 Post subject: Re: How to make an operating system to x86 within a month fr
PostPosted: Mon Apr 17, 2017 11:39 am 
Offline
Member
Member
User avatar

Joined: Fri Oct 27, 2006 9:42 am
Posts: 839
Location: Athens, GA, USA
Geri wrote:
from your comment i guess you like europe overally.


I really wouldn't know, as I have never been there. I do, however, know the US, and most of it is... well, the stereotypes of the Ugly American are if anything understating things to be honest.

_________________
Rev. First Speaker Schol-R-LEA;2 LCF ELF JAM POEE KoR KCO PPWMTF
μή εἶναι βασιλικήν ἀτραπόν ἐπί γεωμετρίαν
Lisp programmers tend to seem very odd to outsiders, just like anyone else who has had a religious experience they can't quite explain to others.


Top
 Profile  
 
 Post subject: Re: How to make an operating system to x86 within a month fr
PostPosted: Mon Apr 17, 2017 11:39 am 
Offline
Member
Member
User avatar

Joined: Wed Jun 17, 2015 9:40 am
Posts: 393
Location: Athens, Greece
Hi,


There were some discussions a couple of months ago about reuniting Cyprus but, quite frankly, I'm not very optimistic about it.

Anyway, we went too far in the offtopicness. :P


Regards,
glauxosdever

_________________
Before implementing any piece of software, make sure you have designed it to be extendable. Backporting new features into old code is not always easy.


Top
 Profile  
 
 Post subject: Re: How to make an operating system to x86 within a month fr
PostPosted: Mon Apr 17, 2017 12:11 pm 
Offline
Member
Member
User avatar

Joined: Fri Oct 27, 2006 9:42 am
Posts: 839
Location: Athens, GA, USA
Geri wrote:
i am did a qick google search and i seen that mips hardware also used the PCI bus system very extensively in the 90s, for example ,,The 133 MHz TMPR3927 high performance, single-chip MIPS-based RISC microprocessor features integrated PCI and '' this is pretty much putted it into same hardware IO league with x86.


First off, you do realize that that is a CPU, not a full computer or even a motherboard, correct? Most of the systems that used the TMPR3927 (or any other MIPS implementation of the time) didn't actually use PCI, because they didn't have any extension buses - they were game consoles with fixed hardware and proprietary connectors for a limited selection of proprietary user controller devices. I don't know if the Sony consoles used the PCI bus model internally, but even if they did, that was only because that was what that particular implementation was designed for. That was Toshiba's decision, and had nothing to do with the MIPS in general.

Second, PCI was used on a lot of systems at the time, and while it arose out of the PC world, it wasn't tied to the PC specifically. PowerPC Macs used it. Pegasus II PPC motherboards used it. DEC Alpha workstations used it. Even some zSystem minis used it. It was a common interface which had been specifically designed not to be dependent on the x86. The cards that worked on it? Different story, since most were designed to only work on PCs.

That having been said, there weren't a lot of MIPS desktops and workstations which used PCI, because there were never a lot of MIPS desktops and workstations. I don't know of a single MIPS desktop-class system at all prior to the Loongson, more's the pity. The Advanced Computing Environment project, which was intended to come up with a replacement for the PC (which was already long in the tooth by 1991), was supposed to use MIPS, but the whole thing ended in acrimony and not a single system implementing the idea was ever sold, though a bunch of systems around that time which no one ever heard of took some of the ideas and made a hash of them to everyone's detriment.

The only MIPS workstations of any real note were the Silicon Graphics Indigo, Iris, Octane, and Indy systems, of which only the Octane (and the Origin rack-mounted server) used PCI to the best of my knowledge; the earlier ones had proprietary buses which they never licensed (meaning that for the most part, only SGI hardware worked in SGI machines). Oh, and there were later models of the Sony NEWS worsktations which used MIPS, too, and those didn't use PCI either; they died out before PCI had really come into the market in significant amounts at all, really. Sony would prefer people forgot their disastrous foray into that market, anyway, and frankly, no one misses them.

_________________
Rev. First Speaker Schol-R-LEA;2 LCF ELF JAM POEE KoR KCO PPWMTF
μή εἶναι βασιλικήν ἀτραπόν ἐπί γεωμετρίαν
Lisp programmers tend to seem very odd to outsiders, just like anyone else who has had a religious experience they can't quite explain to others.


Top
 Profile  
 
 Post subject: Re: How to make an operating system to x86 within a month fr
PostPosted: Mon Apr 17, 2017 12:26 pm 
Offline
Member
Member
User avatar

Joined: Fri Oct 27, 2006 9:42 am
Posts: 839
Location: Athens, GA, USA
And just to make everything clear on this point: no one thinks that x86 is a good design. No one ever has. Not even Intel. It is what it is through happenstance, and survives despite both its shortcoming and multiple attempts to kill it off by Intel themselves. The only reason they don't just drop the product line outright is because they can't afford to - not just because it would cost them their cash cow, but because it would lead to a line of lawsuits for breach of contract that wouldn't be resolved until some time around the heat death of the universe. We are stuck with this piece of garbage, and so are Intel, but they have to make do, and so do we.

_________________
Rev. First Speaker Schol-R-LEA;2 LCF ELF JAM POEE KoR KCO PPWMTF
μή εἶναι βασιλικήν ἀτραπόν ἐπί γεωμετρίαν
Lisp programmers tend to seem very odd to outsiders, just like anyone else who has had a religious experience they can't quite explain to others.


Last edited by Schol-R-LEA on Mon Apr 17, 2017 1:01 pm, edited 2 times in total.

Top
 Profile  
 
 Post subject: Re: How to make an operating system to x86 within a month fr
PostPosted: Mon Apr 17, 2017 12:50 pm 
Offline
Member
Member

Joined: Thu May 17, 2007 1:27 pm
Posts: 234
While nobody thinks x86 is good design I wouldn't trust to hardware engineers to come up with a better design if they started from scratch. The same way existing "legacy" software differs from miraculous rewrites, the x86 architecture differs from other architectures by actually working and existing.


Top
 Profile  
 
 Post subject: Re: How to make an operating system to x86 within a month fr
PostPosted: Mon Apr 17, 2017 1:05 pm 
Offline
Member
Member
User avatar

Joined: Fri Oct 27, 2006 9:42 am
Posts: 839
Location: Athens, GA, USA
Think of it this way: imagine that you own a company which makes making shoe horns. You hear that some people are trying to make knives out of your shoehorns, which just seems weird to you, and so you stop taking orders from the weirdos. You are just getting things started for a new business line in selling televisions, and figure that the new model of shoehorns will last just long enough to pay for that re-tooling, when some guys walk into the factory with a 50-star general following them. One of these guys is one you know who has been making knives out of your shoehorns, while the other one is rolling his eyes and is frantically trying to convince the general that he is making a mistake. The general goes up to you and says, "I need a million of these bayonets you are making by next week".

You might ask him what he is talking about, but he points to a shoehorn, and the knife guy hands you a suitcase full of money with an enthusiastic gleam in his eyes. When you tell him you don't make bayonets, and that those are shoehorns, the general insists that these are perfect for his needs.

OK, you think, we can sell the shoehorns to them, its obvious that something crazy is going on but since I can't seem to talk this general out of it, I might as well get the money.

A few months pass, and you keep getting more orders for 'bayonets'. Meanwhile, the television project isn't going well, and you're thinking you might need to close the factory entirely, or at least focus on cheaper shoehorns again. Then, out of nowhere, government officials from a dozen countries start showing up with duffel bags full of cash, demanding that you sell them bayonets. Would you turn down that sort of money, just at the point when your company is about to tank?

This is basically what happened to Intel. They never wanted to be in the home computer market; they were making embedded systems chips with an eye on the emerging workstation market. The 8086 was something they threw together to squeeze another year or two out of the 8080 line until their 'real' products, the iAPX 432 workstation CPU and the 8051 SBC controller, were ready. They were blindsided by the PC market, but it was roaring just at a time when Intel's big gamble in the 432 was falling apart. That left them with a product line they didn't want, but couldn't afford to leave.

The engineering is not even a factor in the story. What mattered was that IBM laid down the law about home computers, just as companies like Osborne and Proc Tech were dying of terrible management, and everyone still in the game figured that you couldn't lose money by copying what The Empire was doing. Intel could have been selling abaci, but if IBM were using them, then everyone bought their abacus from Intel.

_________________
Rev. First Speaker Schol-R-LEA;2 LCF ELF JAM POEE KoR KCO PPWMTF
μή εἶναι βασιλικήν ἀτραπόν ἐπί γεωμετρίαν
Lisp programmers tend to seem very odd to outsiders, just like anyone else who has had a religious experience they can't quite explain to others.


Last edited by Schol-R-LEA on Mon Apr 17, 2017 1:16 pm, edited 3 times in total.

Top
 Profile  
 
 Post subject: Re: How to make an operating system to x86 within a month fr
PostPosted: Mon Apr 17, 2017 1:08 pm 
Offline
Member
Member
User avatar

Joined: Fri Oct 27, 2006 9:42 am
Posts: 839
Location: Athens, GA, USA
Korona wrote:
While nobody thinks x86 is good design I wouldn't trust to hardware engineers to come up with a better design if they started from scratch. The same way existing "legacy" software differs from miraculous rewrites, the x86 architecture differs from other architectures by actually working and existing.


I won't disagree, but I was trying to explain to Geri that the engineering, and the general low opinion of the x86 held by nearly everyone, wasn't really a relevant in its success. He's going off on this conspiracy-nut approach and not looking at the actual history, or the reasons why it has persisted so long. My point was that incompetence and bad luck can look an awful lot like a conspiracy without actually being one.

_________________
Rev. First Speaker Schol-R-LEA;2 LCF ELF JAM POEE KoR KCO PPWMTF
μή εἶναι βασιλικήν ἀτραπόν ἐπί γεωμετρίαν
Lisp programmers tend to seem very odd to outsiders, just like anyone else who has had a religious experience they can't quite explain to others.


Last edited by Schol-R-LEA on Mon Apr 17, 2017 1:19 pm, edited 1 time in total.

Top
 Profile  
 
 Post subject: Re: How to make an operating system to x86 within a month fr
PostPosted: Mon Apr 17, 2017 1:16 pm 
Offline
Member
Member

Joined: Thu May 17, 2007 1:27 pm
Posts: 234
I got ninja'd by you but let me still say this: That is the history but today Intel's shoehorns are no longer made of wood or plastic but instead of carbon fiber reinforced titanium with lightsabers attached to both sides. You can still use them go get your shoes on (because some of your clients are actually doing this) but their primary function is providing the best bayonet available. Instead of trying to make new "proper" bayonets from scratch it's much cheaper and more reliable to just keep using those shoehorns. And even if somebody did actually try to manufacture proper bayonets there is a high chance that he will implement some design decision that renders the proper bayonets worse than the shoehorns.


Top
 Profile  
 
 Post subject: Re: How to make an operating system to x86 within a month fr
PostPosted: Mon Apr 17, 2017 1:21 pm 
Offline
Member
Member
User avatar

Joined: Fri Oct 27, 2006 9:42 am
Posts: 839
Location: Athens, GA, USA
OK, that's very true. As much as I hate the architecture, I can't disagree that for what people want out of them - that is to say, performance and lots of it - they are the best. I may be wistful about what might have been, but I still use an i5 laptop myself.

_________________
Rev. First Speaker Schol-R-LEA;2 LCF ELF JAM POEE KoR KCO PPWMTF
μή εἶναι βασιλικήν ἀτραπόν ἐπί γεωμετρίαν
Lisp programmers tend to seem very odd to outsiders, just like anyone else who has had a religious experience they can't quite explain to others.


Top
 Profile  
 
 Post subject: Re: How to make an operating system to x86 within a month fr
PostPosted: Mon Apr 17, 2017 1:38 pm 
Offline
Member
Member
User avatar

Joined: Sat Jan 15, 2005 12:00 am
Posts: 7897
Location: At his keyboard!
Hi,

Schol-R-LEA wrote:
And just to make everything clear on this point: no one thinks that x86 is a good design.


I actually do think that, with historical perspective and without ignoring market forces, 80x86 is good design (not perfect, but not bad). The problem is that without historical perspective things that were good at the time can seem bad now in hindsight; and if you ignore market forces (e.g. the importance of backward compatibility) you can end up focusing on things that are completely irrelevant for 99.9% of people (e.g. A20 gate, which is only relevant to people that write boot loaders and isn't something that normal consumers will ever care about).

Note that "market forces" includes things like deciding to create a whole new ISA (Itanium) in the hope of killing your largest competitor (AMD); and then failing because AMD were smart and created a "more pleasant for the market" alternative for 64-bit.

Korona wrote:
While nobody thinks x86 is good design I wouldn't trust to hardware engineers to come up with a better design if they started from scratch. The same way existing "legacy" software differs from miraculous rewrites, the x86 architecture differs from other architectures by actually working and existing.


If I were running Intel; I'd wait for UEFI to kill BIOS, then I'd create an alternative line of "80x86 like" CPUs that don't support deprecated things (real mode, FPU) but are fully compatible for 64-bit software (while still providing/supporting "100% compatible with 80x86" CPUs so that people have a choice). I'd also consider the addition of an "alternative opcode mode" designed to improve code density (a little like thumb mode for ARM) where all software (at any privilege level) can freely switch between "normal 64-bit opcodes" and "alternative 64-bit opcodes"; and add that to both the "100% compatible with 80x86" CPUs and the new "80x86 like" CPUs. After a very long time (after everything that matters has added support for the "less deprecated stuff" alternative CPU) I'd start increasing the prices for "100% compatible with 80x86" CPUs.

The important thing here is to create a slow transition path that doesn't suddenly break compatibility and doesn't cause the sudden massive loss of market share; but does lead to eventual removal of some of the ancient warts.


Cheers,

Brendan

_________________
For all things; perfection is, and will always remain, impossible to achieve in practice. However; by striving for perfection we create things that are as perfect as practically possible. Let the pursuit of perfection be our guide.


Top
 Profile  
 
 Post subject: Re: How to make an operating system to x86 within a month fr
PostPosted: Mon Apr 17, 2017 1:39 pm 
Offline
Member
Member
User avatar

Joined: Fri Oct 27, 2006 9:42 am
Posts: 839
Location: Athens, GA, USA
@Geri: Getting back on topic a bit, the point has been made - and beaten to death - that regardless of what anyone thinks of the x86, we're stuck with it for the foreseeable future. The point has also been made that despite the apparent simplicity of the Subleq 'instruction set', there is zero chance of a hardware implementation of it with similar - or even adequate - performance coming to light. Similarly, the very 'simplicity' (which, as I also said, is misleading in the extreme) works against your in a software simulation, as the low code density - the ratio of pseudo-machine instructions to native instructions - is overwhelming. It isn't even a case where the 'simple' system is easier to write assembly code or target a compiler for - it is orders of magnitude harder on both of them, if anything.

You counter all of these arguments with talk about 'freedom', but I am still trying to figure out what you mean by that, as whole issue of re-implementation... isn't one. Seriously, it isn't a real world issue at all. It certainly isn't one to the buying public, to whom the whole topic would seem like a waste of breath. Most people don't want to use computers; they want to get things done, and in many cases the best way to get them done in with a computer. Talking about 'computer users' as something separate from what they use them would make no more sense than talking about 'microwave oven users' or 'automobile users', except that there are so many different things that computers can (and even must) be used for, and so many different and often conflicting ways to do them, that the topic is very important - to the developers. Not to the users. The users' main goal is to not have to learn about computers, to have things that they can use without a lot of extra things to learn, which means that they will stick to what they have already learned whenever they can.

You say a lot of things which simply aren't true, or even comprehensible. I think we've hit a case of Poe's Law here, where we can't even tell if you are a fanatic, or just a troll. Either way, no one should be taking you seriously.

_________________
Rev. First Speaker Schol-R-LEA;2 LCF ELF JAM POEE KoR KCO PPWMTF
μή εἶναι βασιλικήν ἀτραπόν ἐπί γεωμετρίαν
Lisp programmers tend to seem very odd to outsiders, just like anyone else who has had a religious experience they can't quite explain to others.


Top
 Profile  
 
 Post subject: Re: How to make an operating system to x86 within a month fr
PostPosted: Mon Apr 17, 2017 1:45 pm 
Offline
Member
Member
User avatar

Joined: Sun Jul 14, 2013 6:01 pm
Posts: 364
Schol-R-LEA - i dont have now time to read the newest comments and reply, but i will do it later

_________________
Operating system for SUBLEQ cpu architecture:
http://DawnOS.tk


Top
 Profile  
 
 Post subject: Re: How to make an operating system to x86 within a month fr
PostPosted: Mon Apr 17, 2017 2:02 pm 
Offline
Member
Member

Joined: Thu May 17, 2007 1:27 pm
Posts: 234
Brendan wrote:
If I were running Intel; I'd wait for UEFI to kill BIOS, then I'd create an alternative line of "80x86 like" CPUs that don't support deprecated things (real mode, FPU) but are fully compatible for 64-bit software (while still providing/supporting "100% compatible with 80x86" CPUs so that people have a choice). I'd also consider the addition of an "alternative opcode mode" designed to improve code density (a little like thumb mode for ARM) where all software (at any privilege level) can freely switch between "normal 64-bit opcodes" and "alternative 64-bit opcodes"; and add that to both the "100% compatible with 80x86" CPUs and the new "80x86 like" CPUs. After a very long time (after everything that matters has added support for the "less deprecated stuff" alternative CPU) I'd start increasing the prices for "100% compatible with 80x86" CPUs.

The important thing here is to create a slow transition path that doesn't suddenly break compatibility and doesn't cause the sudden massive loss of market share; but does lead to eventual removal of some of the ancient warts.

Agreed. I do expect that this is the path that Intel will take with x86.


Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 131 posts ]  Go to page Previous  1 ... 3, 4, 5, 6, 7, 8, 9  Next

All times are UTC - 6 hours


Who is online

Users browsing this forum: No registered users and 3 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group