hmmmmm, i cant see Lemmings if emulated in 26bit mode bringing an XScale or ARM10 to its knees

The point is that the 26bit and 32 bit ARMs are (in hardware terms) identical except in terms of how the program counter (R15) and the flags are handled. A full emulation would (in effect) attempt to create a simulation of a 26 bit ARM and then run ALL ARM code under a software emulation - the amount of loss in performance would be stunning. The irritating part is that (perhaps) 70-80% of the time it would not be needed (it after all is only code accessing R15, the flags and context switches where the problem arises and all of these are rare enough).

Have some code translate all obviously offending opcodes and map them into "Undefined instruction" territory and then have a handler that makes the 32 bit ARM look like a like its a 26 bit one to whatever caused the exception. You could then run the ARM (unemulated) for 70-80% of the time (at full speed) and take a hit on speed when the undefined instruction (the converted 26 bit one) is trapped (most of the time you run full speed only now and again do you take a hit). The problem with this is it won't work on ROM code (it can't be modified, so RISC OS can't be fixed this way) and won't be guaranteed to trap all possible means by which R15 might get updated (say you have code that is self modifying - your code converter would not know this - and when it hits the unmodified code well it dies). Still who writes self modifying code anyway (tres naughty )

[Edited by ams at 19:05, 19/11/2000]

well easy... what is the most problems arround the new tecnologies?

1) Up grading Bios, well this can be fixed useing FLASH ROMS for BIOS so Castle or other companies can let people download bios when they discover bugs or make new drivers

2) Up grading the Operating system, also this can be fixed useing FLASH ROMS, so RISC OS Ltd, can distribuite RISC OS on CD ROM it will be more faster and more cheap than the actual system

3) 32 or 26 bits, about a regitry codification it is possible make a library that re-adress all the old registry from 26 bits to 32 (so re-calc string of 26 bits in 32 bits), but anyway an emulation can be used.

4) PCI and USB... well this is the most difficult part, but the new CHIPSET for the new SA support PCI, we will see on the new boards

5) For DMAManager, Memory manager, and all about cache and memory, RISC OS must be REWRITED with two versions one for ARM10 and one for New SA (different chipset mean also different memoy management and specialy DMA management)

6) What service really need the new OS? well:

a) NETPROTOCOLS like SPX and ATM other than TCP, maybe also UDP and NETBEUI could be usefull.
b) Protected memory will allow the system to be more stable and to know wich application crashed etc...
c) MULTITREADING and MULTITASKING PREMPTIVE will let the applications be more fluid and not stop too much all the machine during the HD or CD ROM accesses.
d) Virtual Device Management will let all people make programs able to run every where and with all devices without buy all devices for try their programs.
e) Graphics libraries and C and C++ with the system, other than BASIC for beginners, also some MIDI Library and like this will let it be more interesting.

I think so, but anyway this is only my opinion that came from experiences with a lot of systems and computers like Windows yes, but also like Linux and BeOS and the OLD systems lik MS DOS etc...

Is there any reason Riscos 5 couldn't support multiple threads on multiple processors, and allow 26 bit apps to run if 2 6 bit 2nd processor is fitted.

Yeah, RISC OS 5 doesn't exist yet

Actually this could prove an interesting problem although in RiscStations case I believe that buying a RiscStation 7500 automatically entitles you to pay the difference for an upgrade to Evolution.

Try to compile a code under a pentium and after let it run on a 286 or 386 (also in turbo basic that is very old) the code will never run on old processors.

Idem for PowerPC and old Motorola 68000 series
Idem for Spark and it old versions

sometime i hear here things very strange, (let me say so please) how someone can thing to save old machines? anyway... i hope that all you all dream will be true also if i really don't believe that, we just saw a lot of bugs with a new version of the OS on old machines, we will see.

About 32/26 bit code, RISC OS Ltd and Peace are just distribuiting new libraries and support for the future os, go to RISC OS Ltd site and see, and i think that this is a very good idea.
Well Lee, i can not believe in your wrds, before you talked about the new ARM10 with natively java code and after i hear here to save old machines????? Please explain to me what idea are you following... you know better than me that old machine will never run java (kokkino apart if it will be ready a day)

For the system bus you are right Annaroi, we can also make a system bus on a only one cable (with serial) the problem is about the system buss access but two different processors and about memory allocating adressing... one of the two processors will lose some of itslef features.

People want good 3D games?? really? hmmm well so the SIMD istruction will become a standard in ARM world eheheheh this istruction help a lot in that way for graphics and 3D gameing .... and floatting point will be another very important things (3D poligons need it a lot)... and the clock? more frequency = mode evaluable datas in the processor... life for games

we will see...

You can't win, forget these stupid people. If you have a package that could easily work on older machines then do it but at VOTI we code games for the computer systems we use, we then see if it can run on lesser machines, if it won't or requires too much work then we won't bother.

Quite right too.

Queue Standing ovation.

(clapclapclapclapclapclapclapclapclapclapclap)

Finally developers are making some decent choices!
I woudl personally even go to forgetting about SA Risc PCs, but to build software primarily for the new breed of ARM9/XScale machines.
Then with new technology we'll get new ideas.

Well Lee, i can not believe in your wrds, before you talked about the new ARM10 with natively java code and after i hear here to save old machines????? Please explain to me what idea are you following... you know better than me that old machine will never run java (kokkino apart if it will be ready a day)

No, I said that if you can maintain backwards compatibility without any real effort then you should do so. However developers shouldn't fret over this.

As far as the Java extensions go. Well any Java app won't run on machines without a VM or the hardware. This doesn't bother me. Anyone who is thinking they'll be able too run Java on anything less than a complete minimum of an ARM610 is deluding themselves anyway.

People want good 3D games?? really? hmmm well so the SIMD istruction will become a standard in ARM world eheheheh this istruction help a lot in that way for graphics and 3D gameing .... and floatting point will be another very important things (3D poligons need it a lot)... and the clock? more frequency = mode evaluable datas in the processor... life for games

The SIMD instructions may prove useful, although as I understand it they operate on 16bit values, so are possibly less useful than current instructions unless they can have instructions for large multiplies. Personally I think that widespread availability of the MULL and SMULL instructions (SA and above, or ARM7T which we don't have) would prove more useful and adequate if developers don't mind sticking to fixed point math.

Of couse a mega fast FPU would be better

A Fair point.

But there's no need to write off the RiscPC (after all, it's still a good brandname) - a new motherboard could see SA2 in a RiscPC sometime.

Anyway, maybe we're all needlessly overspeculating here...

This is going back slightly, but I agree.

And any bets that there will be a Kinetic II card out with XScale in it?

I think it's good that Castle are producing these upgrades. Much better than having to keep buying a new machine - it's proving just how long Acorns can last and how upgradeable they are.

Obviously, a new machine with a new motherboard is essential and it is fairly obvious that this is what they are working on. As long as they retain the podule facility - what about those of us who have specialist cards such as digitisers and digital audio production cards.

Addressing the issue of backwards compatibility with 26bit apps - without this, we would no longer have Sibelius, or the Oregan software and all the freeware. How they do it doesn't really matter, does it, as long as it works!

Speaking of Sibelius, John Kortink stated (rather bluntly) that he knew what had to be done to get ViewFinder to work with Sibelius, but wouldn't be implenting it for one "crippled" program. That counts me out as a customer, then

Here's looking forward to RISC OS 4.5

I think it's good that Castle are producing these upgrades. Much better than having to keep buying a new machine - it's proving just how long Acorns can last and how upgradeable they are.

The problem is that the current RPC isn't really upgradable anymore. A new motherboard would have to fit in the same case which would probably make it expensive as you couldn't use a standard one. Also the podules really are at the end of their life.

Obviously, a new machine with a new motherboard is essential and it is fairly obvious that this is what they are working on. As long as they retain the podule facility - what about those of us who have specialist cards such as digitisers and digital audio production cards.

Well assuming PCI was available then there will probably be PCI equivalents. However if Castle can build a machine with both PCI (or whatever) and podules then it'll ease the transition for RPC owners. I think Microdigital and RiscStation have made the right decision in not supporting them though - they're aiming for power users and people who probably need to make a huge upgrade leap (like myself) rather than RPC owners.

Addressing the issue of backwards compatibility with 26bit apps - without this, we would no longer have Sibelius, or the Oregan software and all the freeware. How they do it doesn't really matter, does it, as long as it works!

Agreed.

Speaking of Sibelius, John Kortink stated (rather bluntly) that he knew what had to be done to get ViewFinder to work with Sibelius, but wouldn't be implenting it for one "crippled" program. That counts me out as a customer, then

Look at it from Johns point of view. Siibelius probably bypasses the OS calls when accessing the screen for performance reasons, it may even make assumptions about the video hardware instead of asking the OS about it. This would make it difficult provide a solution for Viewfinder. Although it sounds bad it really is the fault of Sibelius and John has a perfectly reasonable stance on the situation. There is no excuse for any application to not be completely system legal. Games are slightly different (although again there isn't really any excuse for the most part). Should John be fixing what could be argued is a fault with Sibelius?

Maybe the ChiOS would become a serious option again, RiscPC emulation on a card?

BRING BACK CHIOS!

perhaps Castle Tech. would want to do a project like that???

hmmm possibly, but there woudl be more users to buy the software, & it would save me having to "find" some money to buy an Omega

No, the market would definitely die. Who's going to use RISC OS if they need to buy a PC which has it's own selection of operating systems already?

RISC OS on a card is a mad idea that sounds the end of RISC OS.

The problem is that the current RPC isn't really upgradable anymore. A new motherboard would have to fit in the same case which would probably make it expensive as you couldn't use a standard one. Also the podules really are at the end of their life.

Why are podules at the end of their life? If they work, why get rid of them? I think a combination of podules and maybe PCI capabilty would be a good thing.

The current RiscPC is upgradeable. I bet you never envisaged the Kinetic card.

Well assuming PCI was available then there will probably be PCI equivalents. However if Castle can build a machine with both PCI (or whatever) and podules then it'll ease the transition for RPC owners. I think Microdigital and RiscStation have made the right decision in not supporting them though - they're aiming for power users and people who probably need to make a huge upgrade leap (like myself) rather than RPC owners.

Just what card is going to replace, say, the Irlam i16 digital sound card. Sure, there's SoundBlaster etc, but that won't work with the Oregan software and who's going to sort all that out? Millipede have got the right idea

Look at it from Johns point of view. Siibelius probably bypasses the OS calls when accessing the screen for performance reasons, it may even make assumptions about the video hardware instead of asking the OS about it. This would make it difficult provide a solution for Viewfinder. Although it sounds bad it really is the fault of Sibelius and John has a perfectly reasonable stance on the situation. There is no excuse for any application to not be completely system legal. Games are slightly different (although again there isn't really any excuse for the most part). Should John be fixing what could be argued is a fault with Sibelius?

If he knows how to do it, and he does, why not just do it? It may be a Sibelius fault but as yet Sibelius are not supporting us anymore (nice of them) so unfortunately it's up to people like John to get round the little quirks of such software.

If Sibelius wasn't such an important piece of software I would not be upset it wasn't being supported.

Why are podules at the end of their life? If they work, why get rid of them? I think a combination of podules and maybe PCI capabilty would be a good thing.

What's wrong with podules? Well in the RPC they're slow (ok, so Imago and Mico go some way to addressing this) and they're expensive. While I don't expect a RISC OS machine with PCI slots to offer expansion cards as cheaply as a PC I also don't see how 120 quid for a 10Mb ethernet card can be justified or even commerically viable. With the exception of Viewfinder when was the last major podule release?

The current RiscPC is upgradeable. I bet you never envisaged the Kinetic card.

No I didn't. I bet people didn't envisage losing DMA either.

Just what card is going to replace, say, the Irlam i16 digital sound card. Sure, there's SoundBlaster etc, but that won't work with the Oregan software and who's going to sort all that out? Millipede have got the right idea

You're not telling me that there isn't an equivalent card out there somewhere. As for it not working with Oregan software, well I don't know how their software works, but assuming they haven't hardwired the use of the Irlam card into the software (and they'd be very stupid if they have) then surely all that's required is to replace the driver.

If he knows how to do it, and he does, why not just do it? It may be a Sibelius fault but as yet Sibelius are not supporting us anymore (nice of them) so unfortunately it's up to people like John to get round the little quirks of such software.

Perhaps he has got better things to do than create a patch that will offer him no return whatsoever.

If Sibelius wasn't such an important piece of software I would not be upset it wasn't being supported.

Fair enough. It seems to me that the obvious answer is for all the Sibelius users with Viewfinder to mail John and ask for it. If he gets enough demand from people willing to pay for an upgrade he might do it.

What's wrong with podules? Well in the RPC they're slow (ok, so Imago and Mico go some way to addressing this) and they're expensive. While I don't expect a RISC OS machine with PCI slots to offer expansion cards as cheaply as a PC I also don't see how 120 quid for a 10Mb ethernet card can be justified or even commerically viable. With the exception of Viewfinder when was the last major podule release?

I am just saying if you get rid of podules you can't use your expensive add-ons anymore and I don't see piles of drivers for relacements.

No I didn't. I bet people didn't envisage losing DMA either.

You're not telling me that there isn't an equivalent card out there somewhere. As for it not working with Oregan software, well I don't know how their software works, but assuming they haven't hardwired the use of the Irlam card into the software (and they'd be very stupid if they have) then surely all that's required is to replace the driver.

Perhaps he has got better things to do than create a patch that will offer him no return whatsoever.

What rubbish! If he patches his firmware then it allows the thousands of Sibelius users to upgrade to ViewFinder. Sibelius works with the MicroDigital machine and they are new graphics chips. I will simply have to wait for a solution from Castle or Millipede that works!

Fair enough. It seems to me that the obvious answer is for all the Sibelius users with Viewfinder to mail John and ask for it. If he gets enough demand from people willing to pay for an upgrade he might do it.

I am just saying if you get rid of podules you can't use your expensive add-ons anymore and I don't see piles of drivers for relacements.

*sigh* I did say that it would be good if Castle produced a machine capable of taking both Podules and PCI. However I stand by my point that it shouldn't be expected of Microdigital and RiscStation who don't have a vested interest in maintaining the RPC.

As for not seeing piles of driver for replacements, well I'm not surprised. How many developers do you think currently have access to an Omega to write them? Give them a chance.

I'll repeat a point I've said on a few occasions, the advantage of going down the PCI and USB route is that, while drivers for hardware may be complex to write, you don't have to design your own hardware anymore. In theory it should make it easier (or less difficult) to bring new cards to the market.

It depends what you use the machine for. For a lot of people, the loss of DMA matters not.

Unless you've upgraded your hard disk interface. There was enough discussion on the newsgroups about it to suggest that a substantial minority were affected.

Who's going to write the drivers then? Oregan seem to have abandoned their production software for now

But you could argue that about any piece of RISC OS software that isn't being maintained by the original authors anymore. If developers perceive a large enough need for the software to be maintain (enough to make it worth their while as well as being in their interests) then it will probably happen. Look at Artworks, look at Starfighter.

What rubbish! If he patches his firmware then it allows the thousands of Sibelius users to upgrade to ViewFinder. Sibelius works with the MicroDigital machine and they are new graphics chips. I will simply have to wait for a solution from Castle or Millipede that works!

I assume these would be the thousands of Sibelius users not filling his email box requesting him to do it. This would suggest there is little demand for it therefore he would receive very little in return. Either that or they've all accepted it won't work with Viewfinder and resort back to the original hardware configuration to use it.

As for the Omega having new graphics chips, yes they are but they were designed to be compatible with the VIDC. Viewfinder uses an off the shelf AGP card which doesn't provide such functionality. Who says that even the next Castle machine will use a VIDC compatible chipset?

Indeed, but who's going to co-ordinate that? And how?

Well someone could post on the newsgroups that they wanted Sibelius to run on Viewfinder and was looking for people to support putting a petition to John. This co-ordinator would obviously be someone who wants to see this happen. I suspect asking people to mail John directly might be counter-productive though if his interest is already low.

I realise that I sound somewhat pessimistic but I'm just trying to be realistic. In an ideal world Sibelius would be patched but it would appear that the resources (or interest) are not there to do it right now.

Hey, maybe, just maybe, you could ask Tim really nicely if the next vote should be something like "Would you like to see Sibelius made compatible with the Viewfinder?"

[Edited by johnstlr at 21:49, 8/12/2000]

*sigh* I did say that it would be good if Castle produced a machine capable of taking both Podules and PCI. However I stand by my point that it shouldn't be expected of Microdigital and RiscStation who don't have a vested interest in maintaining the RPC.

As for not seeing piles of driver for replacements, well I'm not surprised. How many developers do you think currently have access to an Omega to write them? Give them a chance.

I'll repeat a point I've said on a few occasions, the advantage of going down the PCI and USB route is that, while drivers for hardware may be complex to write, you don't have to design your own hardware anymore. In theory it should make it easier (or less difficult) to bring new cards to the market.

Unless you've upgraded your hard disk interface. There was enough discussion on the newsgroups about it to suggest that a substantial minority were affected.

But you could argue that about any piece of RISC OS software that isn't being maintained by the original authors anymore. If developers perceive a large enough need for the software to be maintain (enough to make it worth their while as well as being in their interests) then it will probably happen. Look at Artworks, look at Starfighter.

I assume these would be the thousands of Sibelius users not filling his email box requesting him to do it. This would suggest there is little demand for it therefore he would receive very little in return. Either that or they've all accepted it won't work with Viewfinder and resort back to the original hardware configuration to use it.

As for the Omega having new graphics chips, yes they are but they were designed to be compatible with the VIDC. Viewfinder uses an off the shelf AGP card which doesn't provide such functionality. Who says that even the next Castle machine will use a VIDC compatible chipset?

Well someone could post on the newsgroups that they wanted Sibelius to run on Viewfinder and was looking for people to support putting a petition to John. This co-ordinator would obviously be someone who wants to see this happen. I suspect asking people to mail John directly might be counter-productive though if his interest is already low.

I realise that I sound somewhat pessimistic but I'm just trying to be realistic. In an ideal world Sibelius would be patched but it would appear that the resources (or interest) are not there to do it right now.

Hey, maybe, just maybe, you could ask Tim really nicely if the next vote should be something like "Would you like to see Sibelius made compatible with the Viewfinder?"

Well if a poll goes up on the IconBar or something like that, I'll co-ordinate this. I would also be happy to provide evidence of an interest for the new Sibelius to be released o the Acorn. I have managed to make contact with one of Sibelius's authors

Ah, Omega, Omega... has anyone seen proof that it works yet? Weren't they supposed to be shipping yesterday or something?

Certain people have but I can't reveal. Be assured it's very much working

Why is there so many people who have to think the worst, demanding proof and things like that? It's because of things like this Castle and MicroDigital keep silent about things.

They might not have a vested interest in maintaining the RiscPC, but they ought to have a vested interest in encouraging users to move over to them. Those with expansion cards simply can't until the equivalent PCI cards have drivers written for them. Besides which we can't all afford to buy a new machine AND replace all our add-ons!

I can understand this, and as you say Millipede will offer a solution because it is in their interests to do so. At some point though the transition has to be made. Perhaps it's ideal that right now machines are coming out that allow for both.

A fair point. Cheap PCI network cards would be nice

I believe Microdigital have announced that they're working on a 100Mb Ethernet card.

Indeed, I agree with this. However, maybe it would be nice to have PCI and/or USB in addition to podules. Millipede are going down that route for the very reason I am stating: their clients already have expansion cards and don't want to end up not being able to use them on new hardware.

The one problem with having both is that you either compromise future expansion capability of push the cost up. IIRC Phoebe was going to have 4 PCI and 3 podule slots but it was also going to cost 1500 quid + VAT.

So these people could wait for the new machines.

If they have the patience.

A lot of people won't actively seek to influence products. The will just take what comes along. So like you say, they will have accepted that it won't work with ViewFinder and proably haven't bought it because of that. Hopefully Castle will emulate the Acorn chips on their new machines, too

Yes but if people don't seek to influence products developers won't perceive a demand. As for Castle remaining compatible with the VIDC, again it'll be useful but at some point it'll have to go and then emulation will be the only option.

Well if a poll goes up on the IconBar or something like that, I'll co-ordinate this. I would also be happy to provide evidence of an interest for the new Sibelius to be released o the Acorn. I have managed to make contact with one of Sibelius's authors

I believe this is now Tim's call.

Ah, Omega, Omega... has anyone seen proof that it works yet? Weren't they supposed
to be shipping yesterday or something?

Certain people have but I can't reveal. Be assured it's very much working

Why is there so many people who have to think the worst, demanding proof and things like that? It's because of things like this Castle and MicroDigital keep silent about things.

I think the problem is that people have been burnt in the past. With Acorn bowing out a lot of people lost their confidence in the market. I was considering an Omega after RiscStation announced that Evolution was still some way off. Unfortunately my emails enquiring about it (and certain features) have not been answered. This doesn't inspire me to put a deposit down so now I'm waiting to see what happens. The CD writer offer would be nice but I can live without it to make sure I'm getting what I want....

...and all the time the RiscStation laptop is supposed to be getting closer. It's expensive, it's no where near as powerful but it'll be really convenient.

Well if a poll goes up on the IconBar or something like that, I'll co-ordinate this. I would also be happy to provide evidence of an interest for the new Sibelius to be released o the Acorn.

I believe this is now Tim's call.

There are several problems with a poll question like "Would you like to see Sibelius made compatible with the Viewfinder?" The main one being that it's not the sort of thing many people are going to answer "no" to, yet a lot of people would answer "yes" when they'd have no interest in buying an updated version.

BTW: Version 0.20 of this forum already has code in it that splits long threads over multiple pages (this page is getting ridiculously long), but I'm afraid you're not going to see that version for a couple of weeks at least as we need to upgrade the version of PHP on the server before I can install it. It works on my hard disc though if that's any comfort .

This is the "drool over new expansion bus technology" thread, right?

Somebody please start a PCI vs AGP vs USB thread! This one is too long already, and I hear there's a new strongARM chip out there!

Well, things are indeed looking up

Oddly early AGP (x1) cards were SLOWER than identical cards done using PCI (I recall a Matrox PCI card being slightly faster than its AGP cousin). AGP (x4) is faster than PCI but it all seems a bit contrived, a designed from the ground up approach would be better. Bear in mind Millipede's Imago is about as fast as AGPx6 (which don't even EXIST !).

USB is (asaik) to be supported by Microdigital and Imago (and probably RiscStation too). The presence of USB is NOT dependant on PCI.

As ever its the software drivers that are going to take the time coming (once compliant hardware exists).

And yes this thread is WAY off topic and maybe its time to start another, how about it guys

About use a PCI bus well yes annraoi is right when he say that PCI have problems, but it have also a lot of cards with low prices and we are in a not good time so it could help us a lot (make a driver is easier than make also the hardware, but ou are right when you said that we have to know PCI structure very well before come in an adventure like this)

DMA... well i want remember that i can not use it with my POWERTEC SCSI 3 UDMA 32 BIT card so i just forget it (sigh!) but anyway we need a good UDMA.

AGP... well like i said we don't need AGP for speed increase the ARM boards are too much faster for graphic than an AGP so is not possible have an expansion for go back eheheheeh

I want say that PUDULE and new RISC OS 32 bits will make a lot of problems most of the podules was projected for be plugged in ARCHIMEDES MACHINES AND FOR USE 26 BITS SYSTEM CALLING SO... WE WILL HAVE THE SAME PROBLEMS OF WINDOWS 95/98 WITH OLD ISA CARDS (crash the system every time)

new app and old app, i am still asking how the old processor will manage a board that can have more RAM that what it can adress etc.. and how OMEGA could have two system 32 and 26 bits...

About OMEGA I WANT REMEMBER (like imago too) that this is motherboards PROJECTED for use the new processors but still not have it plugged in... remember RiscPC moderboards and StrongARM timing problems? well so don't just so easly in things that no one saw till now we want do "timing problems part two"?

SIBELLIUS etc similar... well SIBELLIUS is copyright so only is company know if it will be usefull or not for them develope it or not, so i agree with Lee.

I am still thinking that the old concept and the new concept of SA are too much different for not make problems but maybe i am wrong so this is only a my opinion.

Anyway, if MD can do it, so can others!

Paolo: the new motherboards have had the new processors plugged in.