27-Apr-2024 04:08 GMT.		 UNDER CONSTRUCTION
Anonymous, there are 134 items in your selection (but only 84 shown due to limitation) [1 - 50] [51 - 100] [101 - 134]
[Forum] OS 4.0 DEVELOPMENT PROGRESS REPORTANN.lu
Posted on 05-Mar-2002 23:22 GMT by 4pLaY134 comments
View flat
View list
Status: February 13, 2002 (Status as of February 1) Ben Hermans, LL.M managing partner Hyperion Entertainment Read on... 1.0 AMIGA OS 4.0 DESIGN GOALS

- 1 - Migrate OS 3.9 from 68K to PPC
- 2 - Untie the OS from the Amiga custom chipsets
- 3 - Introduce modern functionality
- 4 - Eliminate key performance bottle-necks
- 5 - Prepare the Amiga OS as a host-OS for Amiga DE

1.1 DEVELOPMENT UPDATE

* Exec Second Generation (Exec SG)
Exec is the kernel of the AmigaOS and is currently written in 68K assembly.
Exec will be re-written for in C and new functionality will be introduced to allow the deployment of OS 4 on any suitable PPC hardware including but not limited to the Amiga One, Blizzard PPC, Cyberstorm PPC, Shark PPC from Elbox, Pegasos from bPlan etc.
The following features are planned:
Hardware abstraction layer Virtual memory New library interface Resource tracking and management Optional memory protection WarpOS backwards compatibility
The following design documents regarding Exec SG were already written and are either finalised or being discussed:
- New library model (final draft)
- Hardware abstraction layer (first draft)
- Virtual environment (first draft)
All design documents are available as PDF files.
Both Dave Haynie and Heinz Wrobel have agreed to serve as technical consultants and the design documents for the HAL and Virtual Environment are currently being re-written to take account of the relevant feedback we collected.

Development status:

Actual coding on the higher level functionality of Exec is already under way.

Most of the “high level” functionality of Exec SG is already implemented (some 50 functions) which includes all list and tag functions and most notably the memory subsystem and parts of the new library scheme.

The memory system is for now limited to the “classic” memory system i.e. the MemHeader/MemChunk functionality. A more sophisticated system is planned and VM support will also be added down the line.

All of the implemented functions and especially the memory subsystem were successfully tested and withstood torture tests without problem.

Work is currently underway on the booting process.

The plan is to adapt a remote gdb kernel so that the whole debugging process can be cross-hosted on Linux thus making debugging substantially easier.

All in all the work is around 50% done.
* 68K emulation

The 68 emulation will be key to OS 4 with many parts of the operating system still in 68K form as well as a very large base of applications and games in 68K.
Development status of the JIT emulator:
Nearly all 68040 integer instructions are now emulated and work is about to start on FPU emulation, OS integration and further performance optimisation.

Emulated opcodes:
data transfer: "move" opcodes, moveq, lea, pea, movem, movec, clr, scc arithmetics: add, addi, adda, addx, addq, sub, subi, suba, subx, subq, neg, mulu, muls, divs, divu comparing: cmp, cmpi, cmpm, cmpa shifts: asl, lsl, rol, asr, lsr, ror, roxr, roxl flow control: bra, dbra, bsr, jmp, jsr, rts, rtr conditional branches: dbcc.x, bcc.x logic opcodes: eor, eori, not, andi, and, ori, or, btst, bset, bclr, bchg, tas miscellaneous: nop, exg, swap, tst, ext, extb, link, unlk

Adressing: all addressing modes which are supported in 68020 Flag emulation: negative, overflow, carry, zero, extended
During dynamic recomplilation a low-level flag data-flow analysis allows run-time optimalisations of recompiled code.

* PPC native TCP/IP stack & PPP drivers

Early tests already suggest that this is the fastest TCP/IP stack found on the Amiga.

Features are as follows:

- implemented as a single shared library - compatible with the Amiga standard "bsdsocket" API, as defined by the AmiTCP product - enhanced API for more control over the inner workings and configuration - built-in DHCP client - Internet Superserver (inetd) - IP filtering and networking address translation - drivers for asynchronous PPP (dial-up networking) and PPP over Ethernet

Development status:
The TCP/IP stack (“Roadshow”) and PPP drivers have been completed with the following issues outstanding:
- GUI (also required for the PPP drivers)
- SSL implementation/integration.
SSL V2 is very close to completion.

* Re-implementation of the Amiga file system (FFS2) for PPC

FFS2 is a fully backwards compatible re-implementation in C of the Fast File System. It supports media > 4 GByte, and a new variant of the file/directory name storage format which allows for long file names (up to 108 characters).
Development status:
Completed and successfully tested.
* Recovery and Salvage tools for FFS2 and SFS
Salvage, Undelete, Unformat, Repair, RDBSalv, ReOrg/Defrag, Check Integrity etc. with full support for FFS2 and SFS
Development status:

Filesystem Check, Salvage (recovery by copying to another partition), Undelete, Optimize ("ReOrg"/"Defrag"), Find Partitions and Unformat are done. The filesystem Repair function still needs some work. Localization and user documenation is not done yet.

* PPC native RTG system (Picasso 96 V3)
The RTG system allows for the use of modern graphics cards such as the Matrox and ATI cards on the Amiga.
Development status:
Permedia 2 driver: 85% completed Permedia 3 driver: 60% completed Voodoo 3 driver: 100% completed Matrox G450/G550: work in progress ATI Radeon family: work in progress

Picasso 96 V3 will subsequently be ported to PPC with the use of the new Exec SG library model for optimal performance.
An arithmetically optimised PPC native version of layers.library is planned.
* Warp3D (3D driver system)

Warp3D is a powerful yet low-level 3D API which is both available on the classic Amiga and intent/Amiga DE which allows developers to rapidly migrate 3D content between both platforms.
Development status:

Permedia 2 driver: 100% completed Permedia 3 driver: work to start upon completion of 2D drivers Voodoo 3 driver: 100% completed Matrox G450/G550: work to start upon completion of 2D drivers ATI Radeon family: work to start upon completion of 2D drivers
* OpenGL 1.3 support (Mesa 4.0)
OpenGL is a cross-platform (Mac, Linux, Windows, AmigaDE etc.) high-level 3D API originally developed by Silicon Graphics.
Support for OpenGL 1.3 will be provided by porting the open source project Mesa (which now passes all SGI compliance tests) which will sit on top of Warp3D so that graphics-card functionality not currently offered by the OpenGL API may be supported nonetheless.
Development status:
Work will start once Warp3D V5 has been completed.
* PPC native RTA system based on AHI
Developer: Martin Blom and others
A Retargetable Audio System allows the use of plug-in soundcards (PCI or Zorro II).

AHI by Martin Blom is currently the de facto standard on the Amiga and a PPC native version will be offered with support for a wide variety of soundcards including but not limited to all current Amiga soundcards and the Soundblaster 128 and Live (EMU10K1) cards.
Development status:
Martin Blom has started work on the Soundblaster Live drivers.
* PPC native Intuition and Reaction
* Reaction is the BOOPSI based GUI system for the Amiga, introduced in OS 3.5 and extended in OS 3.9.
The functionality of Reaction will be extended by providing more classes and be integrating Reaction better within the OS.
The addition of the following classes is being contemplated:
-application.class - name not finalized. This class will be a "parent" and likely unify window class and arexx class management together, ease management of multi-window applications and support easy creation of applications as commodities.
-speedhint.class - a new help display class based on the speenthint code from window.class, however this will allow for a more powerful API and future growth path.
-popwin.gadget - similar to chooser.gadget but allows a popup display containing most any other gadget as a child - most useful for a popup listview. For this gadget to function, Intuition needs to gain some new ability. The safety of making these changes and therefore including the gadget will need to be evaluated via beta testing.
- prefs.class - a new baseclass that allows for custom plug-in preferences modules for any of the ReAction classes. Subclasses need to be written for the existing ReAction classes and supported in ReActionPrefs.
The following functionality will also be implemented:
Drag and Drop support.
New “ghosted” look
* Intuition will sport the following new features:
New DrawInfo pens, Enhanced window borders Resolution-adaptive system gadgets User-selectable styles for system glyphs and 3D frames with support for external plug-ins Configurable look for proportional gadgets New-style (3D recessed) “disabled” look for gadgets where applicable Gad-Tools enhancements (pop-up capability for cycle gadgets etc.) Full-user control of Workbench palette
Development status:
Work is expected to be completed in time for release of OS 4.0 with more work already planned for subsequent updates of the OS 4.0.
* SCSI drivers for onboard (UW) SCSI controllers
The current SCSI drivers for the onboard SCSI controllers of the Cyberstorm PPC and the BlizzardPPC are written in 68K code which cannot be emulated due to the usage of MMU instructions and the different MMU pagesizes of the 68K and PPC line of CPU’s.
It is therefore necessary to replace these drivers.
Support for other SCRIPTS based PCI-based SCSI controllers is planned to allow an easy migration of SCSI hardware to non-SCSI systems such as the AmigaOne.
Development status:
The SCSI driver framework is completed with work in progress on implementing the actual driver for the Cyberstorm PPC onboard SCSI controller.
Good progress is being made and the driver is expected to be finished in time.
* WarpInput API
WarpInput is an API for multimedia controller devices (akin to DirectInput on Windows) which allows a programmer to provide support through one API for a wide variety of input-devices such as keyboard, mouse, joysticks(analog and digital), track-ball, Playstation controller etc.
The design document is in its second draft and implementation has started.
Development status:
Work is underway to re-implement lowlevel.library in C.
* Minimal USB stack
A minimal USB stack would allow for the use of USB based keyboards and mice which would effectively aid in untying the OS from the custom chipsets.
Development status:
Design documents are drafted and implementation work is underway.
* PPC native datatypes
PPC native datatypes for common formats such TIFF, JPEG, PNG and BMP.
Development status:
Completed but still need to be recompiled for OS 4.
* New HDToolbox
HDToolbox is a utiliy which allow for the formatting and partitioning of storage devices.
The new incarnation of HDToolbox will offer substantially more functionality than the current version.
Development status:
- Lowlevel part ('the Engine'):
Some 80% of the written code was tested with a total of about 85% written. Succesfully tested code includes: removing/adding partition and filesystem, moving and sizing partition, saving RDB back to disk or to a file. The automatic error reporting and correction of RDB structures is still missing. This is the last remaining major item which remains to be implemented in the Engine.
- Highlevel part (GUI):
The partitioning window is 100% complete owing to a new gadget designed by Massimo Tantignone. So is the filesystem selection window barring the gadgets related to the correction/error reporting of RDB values. Two windows remain to be fully managed: the Installation window (already working but incomplete) and the lowlevel SCSI utilities window which is fully designed but not yet working. Also missing is a complete test of the GUI layout functionalities: the GUI will try to open a custom screen if the current screen is not big enough but this is not tested yet.
Still to do: support for localization and user documentation.
Overall the development status is estimated to be 85% completed.
* Support for TrueType/OpenType fonts
Currently the AmigaOS does not support TrueType fonts which is a serious deficiency.
A new font-engine as well as a reworked version of the bullet API, bullet.library, diskfont.library is called for.
Development status:
Work on the initial version for OS 4 is around 95% completed.
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 51 of 134ANN.lu
Posted by priest on 06-Mar-2002 14:58 GMT
In reply to Comment 50 (José):
>>"- 5 - Prepare the Amiga OS as a host-OS for Amiga DE"
>PLEASE EXPLAIN. That's the deciding issue for me. What does this means?
DE needs PPC -> porting of the OS.
DE might need more advanced host OS features -> being added to the OS
>How will native apps, and native apps programing be affected?
Because of DE? In no way.
IIRC: you can code for OS4 just like you would code for OS3.9, except a few new initialization calls when starting the app (or something like that)
>Thanks, really, thought I think noone will answer.
I often try... even though I had only a partial answer, like this time.
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 52 of 134ANN.lu
Posted by José on 06-Mar-2002 15:21 GMT
In reply to Comment 51 (priest):
Hey.
"IIRC: you can code for OS4 just like you would code for OS3.9, except a few new initialization calls when starting the app (or something like that)"
Hmmm, I don't remeber reading that. But I guess it's ok then, performance won't be affected.
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 53 of 134ANN.lu
Posted by Nicolas Sallin on 06-Mar-2002 15:24 GMT
In reply to Comment 43 (priest):
> Amiga Inc wants to have a AmigaOS that is under their control, so that they can develop it as it best suits to their needs, etc...
> It would be impossible with MorphOS (because it's under RS control).
AmigaInc nearly wanted to have the A\Box (i.e: the PPC AmigaOS compatible part of MorphOS) for free, this was this summer iirc.
So, you are wrong in this point :-)
> So, IMO: It's about control.
True, Mr Hermans and his friends want to have total control over the OS :-) even just to write games and 3d api&drivers.
But in reality, nobody needs kernel source to work on such things.
> I think the good part of AOS is that it is not tied to any particular HW like MOS seems to be tied to pblan & DCE HW.
> (hope some MOS supporter now tells that MOS will support wider range of PPC HW than just those)
MorphOS already runs on 4 differents boards: CybPPC, BlizzPPC, Peg proto, Peg final. The plan is to adapt it on others hardwares later. Macintosh in mind, a laptop is needed.
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 54 of 134ANN.lu
Posted by Johan "Graak" Forsberg on 06-Mar-2002 15:47 GMT
In reply to Comment 47 (Ole-Egil Hvitmyren):
I was there too (as you may have noticed Ole-Egil ;), and I talked to Martin Blom and tried out his Amithlon-machine there, but I was under the impression that he only had SoundBlaster 128 in it. And about that progress report, I just read the first lines of it and thought it was the old one =)
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 55 of 134ANN.lu
Posted by priest on 06-Mar-2002 16:00 GMT
In reply to Comment 53 (Nicolas Sallin):
>> Amiga Inc wants to have a AmigaOS that is under their control, so that
>>they can develop it as it best suits to their needs, etc...
>> It would be impossible with MorphOS (because it's under RS control).
>AmigaInc nearly wanted to have the A\Box (i.e: the PPC AmigaOS compatible
>part of MorphOS) for free, this was this summer iirc.
>So, you are wrong in this point :-)
I know they would have wanted to use it, but I think no one knows the truth, except those that were involved.
I have heard a lot of stuff about it, one being that AI would have wanted changes that were not ok for MOS guys.
But we are wise if we leave that subject to that. Perhaps MOS is better without AI.
>> So, IMO: It's about control.
>True, Mr Hermans and his friends want to have total control over the OS :-)
>even just to write games and 3d api&drivers.
IMO: It is AI who needs the control.
>But in reality, nobody needs kernel source to work on such things.
So it is AI who needs the control. Hyperion could as well code games for MOS, already.
>> I think the good part of AOS is that it is not tied to any particular HW
>>like MOS seems to be tied to pblan & DCE HW.
>> (hope some MOS supporter now tells that MOS will support wider range of
>> PPC HW than just those)
>MorphOS already runs on 4 differents boards: CybPPC, BlizzPPC, Peg proto,
>Peg final.
That's just pblan and DCE and one manufacturer (DCE).
>The plan is to adapt it on others hardwares later. Macintosh in
>mind, a laptop is needed.
Great, I really hope it happens.
(even though it might reduce pblan's income and part of RS's potential income via pblan ...)
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 56 of 134ANN.lu
Posted by priest on 06-Mar-2002 16:15 GMT
In reply to Comment 52 (José):
Hmmmm... I tried to check it,
I tried this search: http://groups.yahoo.com/group/amigaone/messagesearch?query=compatibility
And found this: http://groups.yahoo.com/group/amigaone/message/11193 (two lines)
and: http://groups.yahoo.com/group/amigaone/message/12101 (90% compatible)
and: http://groups.yahoo.com/group/amigaone/message/11194 (more about compatibility)
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 57 of 134ANN.lu
Posted by Trizt on 06-Mar-2002 16:19 GMT
Could be nice if FFS2 could be a filesystem with a journal like ext3/xfs.
I hope no one will get the idea that you have to shutdown the system before you can turn of the computer, IMHO a powerdown turning off is the best (simple and fast).
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 58 of 134ANN.lu
Posted by Ian on 06-Mar-2002 16:21 GMT
Has anybody got figures for how well the 68K emulator runs? Although all the new and extra features of the OS are very welcome, it would provide a good incentive to existing Amiga users if the (say) AmigaOne would cause your existing apps to run much faster. Since a good part of the OS will run under emulation you wouldn`t want to find that it's slower.
Also, how transparent will it be? Can I just click on an exe and it will auto-detect to run under emulation, or will we have to have some sort of "emulation environment" isolated from the rest of the OS?
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 59 of 134ANN.lu
Posted by Kronos on 06-Mar-2002 16:45 GMT
In reply to Comment 58 (Ian):
AFAIK the JIT is about 95% on a 604/233mhz. (has been mentioned
by H.J.F. somewhere, sometimes :o)
Add a cache, a descent design (which the P5 newer had)and a
cpu one or two generations ahead runnig at at least 600mhz.
Make up your mind.
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 60 of 134ANN.lu
Posted by Kronos on 06-Mar-2002 16:48 GMT
In reply to Comment 58 (Ian):
Oh forgot that bit.
You should be able to use 68k without noticing in
every place (otherwise using 68k-PFS wouldn't be
possible).
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 61 of 134ANN.lu
Posted by priest on 06-Mar-2002 18:01 GMT
In reply to Comment 58 (Ian):
68k emulation in MorphOS is said to run 68k apps like 060 at half of the PPC clock speed.
That means that 600Mhz G3 would be (very very roughly) equivalent to a 300Mhz 060, so it will be aprox 6x the speed of 060, if everything would run through emulation. All native PPC parts will run 16x faster than 060/50.
I think it will take long time before Hyperion reach the same speed, but you get some clue from that.
The emulation will be transparent in OS4. You will just see speeeed!
(Amithlon at Athlon 1Ghz is said to run at around 040/450Mhz speed and 040 is a lot slower CPU than 060.)
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 62 of 134ANN.lu
Posted by Gunne Steen on 06-Mar-2002 18:04 GMT
In reply to Comment 12 (Ben Hermans/Hyperion):
Hello Ben and everybody else
The progress-report was available (printed) at the Swedish show (AmiGBG 2002)
last saturday. Because that the show come to be a bigger success then we all
had expected, eg more visitors then I think no exhibitor had expected, and
so many wanted to see the report, we decided to put it on our web for downloading
the day after.
Martin Blom was at the show, demonstrating AHI-driver for Soundblaster Live. He
also told me that he have some more possible benefits to AHI in mind for future.
Soo... be kind with Martin...:-)
Also there where a lot of others Exhibitors, like Onyxsoft, Kaliko etc...
A show-report is available at; http://www.amigbg.com
And if you like pictures please redirect your browser to; http://live.amigbg.com
We will also do some follow-up to the show, when we have cold down a little bit,
so keep an eye on the site...
Looking forward to OS 4... and to faster CPU-cards to the Amiga.
Cheers Gunne Steen
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 63 of 134ANN.lu
Posted by Anonymous on 06-Mar-2002 18:05 GMT
In reply to Comment 59 (Kronos):
>AFAIK the JIT is about 95% on a 604/233mhz.
>
*LOL* I would think not even the VP reaches 95% native CPU speed.
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 64 of 134ANN.lu
Posted by Anonymous on 06-Mar-2002 18:17 GMT
In reply to Comment 49 (cOrpse):
>Decent TCP/IP ,
>
You don't like the MOS TCP/IP? Roadshow will be not exclusive for OS4, you know?
>
>decent RTG (CGX sucks bawls believe me) ,
>
What does "sucks bawls" mean? To me either is as good as the other.
>
>Decent Filesystem ( that comes with the os ;) )etc.
>
SFS for MOS most likely. Even Ben Hermans admits FFS2 is only a polished FFS but
not the future filesystem solution.
>
>Morphos is in my view trying to reinvent the wheel , where as AOS has the wheels
>and the rest of the car and is cleaning it up and adding a body kit :).
>
Right, AOS has the wheels, but they have many angles and Hyperion's task is to
file it round so the car will run (maybe never perfectly though).
Now, MOS took a blank sheet of paper and a pair of conmpasses and draw a clean,
perfect circle for its wheels. :-)
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 65 of 134ANN.lu
Posted by Anonymous on 06-Mar-2002 18:20 GMT
In reply to Comment 22 (DET Nicolas):
BTW,the MOS team had to rewrite parts that the OS4 already had written. So the MOS team is reinventing the wheel, instead of licencing the AROS/OS3.1 original sources. The OS4 team has all the sources and they can use it, so it will take them less time. It's also a bigger team.
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 66 of 134ANN.lu
Posted by Kronos on 06-Mar-2002 18:24 GMT
In reply to Comment 63 (Anonymous):
Yes LOL :))
Typed to fast without reading it afterwards.
95% of an 060/50mhz was what i meant.
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 67 of 134ANN.lu
Posted by Alkis Tsapanidis on 06-Mar-2002 18:43 GMT
In reply to Comment 49 (cOrpse):
TCP/IP: *NEW* AmiTCP, with DCHP and several other stuff, ppc native, with GUI.
CGX sucks? Ae you kidding me? I've been using it for ages with no probs....
SFS is already PART of Morphos. It's in module.rom, PPC native. You just have
to partition your drives using the SFS 1.84.
About reinventing the wheel... Both teams HAVE to reinvent the wheel.
The problem is who will finish first. But then again what *I* care about
is to see both OSes finised at any time. Go use OS4 alone, I'll have the
advantages of using both oses and this means... more software.
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 68 of 134ANN.lu
Posted by Nicolas Sallin on 06-Mar-2002 18:47 GMT
In reply to Comment 65 (Anonymous):
You can have all the sources you want, that won't help you very much
when you need to write a kernel for a different processor, add a lot
of new concepts like a HAL, virtual memory spaces and so on...
Even less, when you need to add a JIT which was written for a
different architecture (warpos).
And all that doesn't help you in writting drivers for new chips: scsi,
gfx, snd,...
And at the end, when you have your base OS, you still need to rewrite
from scratch some components like the Workbench itself which sucks
badly and is a joke if you intend to simply "extend" it to add
multithreading...
You see, a load of work is needed to have something
decent, and that's only a part:-)
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 69 of 134ANN.lu
Posted by Targhan on 06-Mar-2002 19:18 GMT
Wow.. there are still people wanting to argue about MOS / AOS4. I
guess that I'll throw my two cents in, and I'm willing to be that ther
is more than just me who feel the same way.
MorphOS looks pretty cool, and if they can port it to the hardware
that I'll be upgrading to (SharkPPC)--GREAT! AmigaOS4 if it will work
on the Shark (hoping that the Shark goes into production)--GREAT. If
they BOTH become available for SharkPPC (and, I'm willing to bet the
prometheous users will say the same thing about their upgrade
path)--Double GREAT! Bottom line, if I *can* use them both--I
probably will.
I certainly wouldn't want to hope that EITHER effort was in vein, and
will give both a chance when the time comes.
Regards,
Targhan
--
infinitiv Towered A1200, 040/extraslow MHz, mediator, Voodoo3
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 70 of 134ANN.lu
Posted by Don Cox on 06-Mar-2002 19:26 GMT
In reply to Comment 58 (Ian):
"Has anybody got figures for how well the 68K emulator runs? Although all the new and extra
features of the OS are very welcome, it would provide a good incentive to existing Amiga users if
the (say) AmigaOne would cause your existing apps to run much faster. Since a good part of the OS
will run under emulation you wouldn`t want to find that it's slower. "
It will have to be a great deal faster than a 68060, otherwise why
would anyone buy it? If Amithlon runs your programs at 12 times the
speed of a 68060, there's not much point in buying something that is
much slower.
IMO the crucial deciding factor for these new Amigas will be which
runs 68k software best, as that will be the vast majority of the
software available.
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 71 of 134ANN.lu
Posted by Martin Blom on 06-Mar-2002 20:32 GMT
In reply to Comment 41 (Anonymous):
(According to <URL:http://www.rvo.mxm.cx/gbg/AmiGBG-021.jpg>, it actually
blows more than it sucks ... ;-) )
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 72 of 134ANN.lu
Posted by Darin on 06-Mar-2002 20:34 GMT
In the posting, Ben mentioned that the design documents are available in pdf format. Does anyone know where we can actually find these documents?
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 73 of 134ANN.lu
Posted by Christophe Decanini on 06-Mar-2002 20:47 GMT
In reply to Comment 71 (Martin Blom):
"(According to <URL:http://www.rvo.mxm.cx/gbg/AmiGBG-021.jpg>, it actually
blows more than it sucks ... ;-) ) "
WOW what a typo !
Anyway i like Martin sense of humor.
He deserves a big "THANK YOU" for the incredible amount of work he did for the Amiga. I hope he is part of the AmigaOS 4 / MorphOS development :)
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 74 of 134ANN.lu
Posted by Gabriele Svelto on 06-Mar-2002 21:32 GMT
Nobody knows if this wonder will work on new/old Macs too? AFAIK they are CHRP machines too so it shouldn't be *that* hard, I really would like to see one of them running AmigaOS ;-)
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 75 of 134ANN.lu
Posted by Quantum-3 on 06-Mar-2002 21:56 GMT
In reply to Comment 61 (priest):
>(Amithlon at Athlon 1Ghz is said to run at around 040/450Mhz speed
>and 040 is a lot slower CPU than 060.)
Actually on my XP1800 the 040 is reported as just slightly below 2 ghz.
It feels around that speed too.
Cheers :-)
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 76 of 134ANN.lu
Posted by Ben Hermans/Hyperion on 06-Mar-2002 22:09 GMT
In reply to Comment 75 (Quantum-3):
This contradicts what Harald Frank said that you get an 040 of 60% of the native performance (at best).
An XP 1800 is what? An Athlon 1500 Mhz?
That would put the "real" 040 speed at considerably slower than 2 Ghz.
People simply don't appreciate the enormous overhead involved in emulation.
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 77 of 134ANN.lu
Posted by Anonymous on 06-Mar-2002 22:45 GMT
In reply to Comment 71 (Martin Blom):
Which of your "stuff" did you actually show, Mr "Blow"? *gg*
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 78 of 134ANN.lu
Posted by Anonymous on 06-Mar-2002 22:50 GMT
In reply to Comment 75 (Quantum-3):
>Actually on my XP1800 the 040 is reported as just slightly below 2 ghz.
>
Sure, and an XP2000+ would be slightly below 040-3GHz? *lol*
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 79 of 134ANN.lu
Posted by Anonymous on 07-Mar-2002 00:49 GMT
In reply to Comment 75 (Quantum-3):
Let me quess ? You used SysInfo ? Try quess why I ask it :)
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 80 of 134ANN.lu
Posted by Anonymous on 07-Mar-2002 00:53 GMT
In reply to Comment 78 (Anonymous):
And you used SysInfo too ? :)
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 81 of 134ANN.lu
Posted by HammerD on 07-Mar-2002 01:14 GMT
In reply to Comment 80 (Anonymous):
On my AMD Athlon 1266MHz with PC-133 memory SysSpeed (under OS 3.9/Amithlon) reports:
68040 at 1578 MHz
MIPS : 403
MFLOPS: 609
In comparison a 060/66Mhz gets:
MIPS : 66
MFLOPS: 26.50
And a PPC (200MHz) gets:
MIPS : 271
MFLOPS: 179
I would expect a 600MHz G3 to get maybe 3x the mips/mflops figure of the 200Mhz 604e.
HammerD
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 82 of 134ANN.lu
Posted by NeRP on 07-Mar-2002 01:25 GMT
In reply to Comment 70 (Don Cox):
I honestly couldn't care if my current Amiga system with cyberstorm
060 and 233mhz PPC runs 68K software slower than amithlon when I run
my Amiga under OSv4
As long as the 604 PPC can emulate 68K software close enough to the
real thing, then I've lost nothing! I'm not looking for 68K software
anyway, I'm waiting for PPC native apps to come out. Current 68K
software speeds isn't an issue as long as it's about 80 % of the same
speed.
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 83 of 134ANN.lu
Posted by HammerD on 07-Mar-2002 01:57 GMT
In reply to Comment 82 (NeRP):
Well it would depend what your current 68K cpu is...
I am used to a 68060/66Mhz, which is pretty fast for Amiga.
When I see Amithlon run its just blindingly fast..its really quite
amazing - way faster than the 060 even.
We'll have to wait and see how well the OS 4 emulator is. But the Amithlon
one is a custom written, 2nd generation JIT that is quite good (IMHO).
HammerD
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 84 of 134ANN.lu
Posted by logain on 07-Mar-2002 01:59 GMT
Hello Mr. Hermans,
I personly think this progress report is really nice and gives a good inside in what is archived or being worked at. Will you give us a new Progress Report in the near future?
Regards
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 85 of 134ANN.lu
Posted by Joe "Floid" Kanowitz on 07-Mar-2002 07:34 GMT
In reply to Comment 33 (Mika Hanhijärvi):
Mika said,
>Yep. PFS3 has somekind of journaling system, I don't know technical issues
>more. They call it Atomic commit, PFS3 can go only from one valid state to
>another valid stae. If system crashes when you are writing to disk, you will
>only loose changes to that file and disk will return to state where it was
>before writing to disk started.
That's the softupdates concept- you only commit to disk in an order that you can safely recover from. The softupdates concept is concerned with preserving data integrity; checking the filesystem still take a long time, though presumably it keeps the filesystem from getting to a state that *requires* checking before accesses (and further writes) can be performed safely.
Journaling is a little different, and is concerned with speeding recovery *after* a catastrophe occurs. The system keeps a log of the last accesses in one location, rather than spread around the disk, and can scan through the log to find incompleted commits and know where the bad spots may lie:
http://www.linux-mag.com/2000-08/journaling_03.html
I don't think there's anything that prevents *combining* softupdates and journaling, which would provide both best integrity and fastest recovery, but the softupdates (atomic updates) side will argue that journaling impacts performance, while the journaling side will complain that softupdates impact performance...
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 86 of 134ANN.lu
Posted by Anonymous on 07-Mar-2002 08:04 GMT
I really don't understand.
Morphos is working for two years now. We saw it running on a real Pegasos in Köln.
We never saw OS4 running, even less on an AmigaOne. The work on OS4 started 3 or 4 months ago, and people are expecting that OS 4 will ship in 2 months on an AmigaOne, where it took approx. 2 years to bPlan to have Morphos on demonstration on a live Pegasos.
How the hell can you imagine than OS 4 will be available THIS YEAR ?
Calendar of operations :
- Spring 2k2 : release of the Pegasos with Morphos
- Summer 2k2 : Alpha of OS 4 on CSPPC
- Winter 2k2 : Beta of OS 4, maybe on an Aone
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 87 of 134ANN.lu
Posted by priest on 07-Mar-2002 08:23 GMT
In reply to Comment 86 (Anonymous):
>How the hell can you imagine than OS 4 will be available THIS YEAR ?
Less to do and more man power ... ?
But MorphOS surely does have the possibility to be far more mature product this summer.
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 88 of 134ANN.lu
Posted by priest on 07-Mar-2002 08:29 GMT
In reply to Comment 81 (HammerD):
>On my AMD Athlon 1266MHz with PC-133 memory SysSpeed (under OS 3.9/Amithlon)
>reports:
>68040 at 1578 MHz
>MIPS : 403
>MFLOPS: 609
So, it seems that the emulation has improved.
I wonder why they market it as só much slower.
Now it would be interesting to get the same MIPS rating from MorphOS running 68k code.
>In comparison a 060/66Mhz gets:
>MIPS : 66
>MFLOPS: 26.50
Is this with 060 optimized code?
IIRC: According to motorola 060/50 should reach 80MIPS
>And a PPC (200MHz) gets:
>MIPS : 271
>MFLOPS: 179
This must be native PPC code, but using what? (PowerUP/WarpUP/MorphOS?)
>I would expect a 600MHz G3 to get maybe 3x the mips/mflops figure of the 200Mhz 604e.
It's little over 1300Mips according to IBM (IIRC).
(IIRC, Duron600 gets about 200Mips more)
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 89 of 134ANN.lu
Posted by Anonymous on 07-Mar-2002 08:52 GMT
In reply to Comment 87 (priest):
There are more to do and maybe yes more manpower.
But how about HW dependencies ?
Where is the kernal ?
The JIT emulator ?
They told us it's done or under way and we saw nothing running on an Amiga PPC so far.
I'm sure they can deliver, and they will do it. But not in the timeframe they claim.
I'm a bit upset by years of promises and still nothing to play with.
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 90 of 134ANN.lu
Posted by anonymous on 07-Mar-2002 09:29 GMT
In reply to Comment 88 (priest):
I wonder why no-one question MIPS as a performance benchmark?
MIPS have *nothing* to do with real-life speed, esp when using JIT
emulation...
You should do some real work, and time that instead. This will give
you some real comparision figures, and most likely the emulation
doesn't look that fast anymore...
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 91 of 134ANN.lu
Posted by priest on 07-Mar-2002 09:37 GMT
In reply to Comment 90 (anonymous):
I agree that we need a decent "performance measurement suite".
For example demo versions of hand full of applications and AREXX scripts that would stress the system using those apps.
But as we do not have anything like that, MIPS gives at least some very very rough estimates of how RAW number munching gets sped up.
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 92 of 134ANN.lu
Posted by Solar on 07-Mar-2002 10:25 GMT
In reply to Comment 91 (priest):
MIPS as a measure of "raw number crunching"? Oh, I see, you're joking...
(And, by the way, how much "raw number crunching" are you doing over a day, and how much /real/ work? So what's the speed that matters? Correct.)
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 93 of 134ANN.lu
Posted by priest on 07-Mar-2002 10:27 GMT
In reply to Comment 89 (Anonymous):
>There are more to do ...
Hyperion has more to do for OS4 than what MorphOS team has done in three years?
I do not think so.
IMO: There is less to do to go from 3.1/3.9 to 4.0, than from zero+Quark to MorphOS1.0.
I think Hyperion needs to do a lot less reverse-engineering than MorphOS team has done.
(note that I do not know the full history of MorphOS)
>But how about HW dependencies ?
They've said they will go step by step.
And they'll have the bridge to AGA HW anyway.
>Where is the kernal ?
That's something they have more to do, while MorphOS uses Quark.
>The JIT emulator ?
I think also MorphOS guys had to do that.
>They told us it's done or under way and we saw nothing running on an Amiga PPC so far.
Read the report. That's what they have said and done.
If one uses one's brains one should see that you will need a kernel to run anything at all.
(Quark in MorphOS, execSG in OS4)
It has been know all the time that execSG is not ready.
>I'm sure they can deliver, and they will do it. But not in the timeframe they claim.
I think the discussion was if they would deliver THIS YEAR.
I think they will do "several deliveries" of OS4 during this year (a,b,0, and plenty of bugfixes).
Perhaps not 4.1, though.
They have not said what features are in 4.0 and what are in 4.1, I think if there are delays,
features will be postponed to the next relese (very common prosedure for SW professionals
to meet the deadline when nothing else helps, been there, seen that).
I think no-one should expect OS3.1 level quality in OS4.0, untill later this year.
There is plenty of room for Hyperion to make surprices. I hope we get more positive than negative ones.
>I'm a bit upset by years of promises and still nothing to play with.
Perhaps I've become numb when regarding those. ;)
But we are in a wholly different situation now, things are being worked at by serious
teams and there always is the backup plan, planb (or should I say bplan). ;)
(as well as those x86 Amiga alternatives)
I do not "believe", I consider the facts available and draw conclusion what is possible.
It is ALWAYS possible that shit happens. When you accept that, you will not get hurt,
it might kill some of one's enthusiasm, but it protects "your soul". :)
Current state of thing is: EXTREMELY POSITIVE ! 8)
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 94 of 134ANN.lu
Posted by priest on 07-Mar-2002 10:54 GMT
In reply to Comment 92 (Solar):
>MIPS as a measure of "raw number crunching"?
No,
>Oh, I see, you're joking...
no.
I said:
"But as we do not have anything like that, MIPS gives at least some very very rough estimates of how RAW number munching gets sped up."
Pay attention to the "very very rough estimate" and "munching", not crunching... ;)
>(And, by the way, how much "raw number crunching" are you doing over a day,
>and how much /real/ work?
ok, OFF-TOPIC, but:
Raw crunching with the Amiga... almost nothing, one reason is lack of time,
the other is lack of RAW power, third is that I have not BOUGHT THOSE apps yet,
fourth is that there are so much other things to do with it... ;)
>So what's the speed that matters?
General/total speed of the system. For me, the nbr1 is the fluency of multitasking.
The ultimate benchmark system for me would be a multitasking stress test (FP and INT
load) while some realtime multimedia effects are played simultaneously. It would be
interesting to see those results. One hicup in audio/video vould result 0-points. ;-)
Also thigs like 3D video rendering and 2D game playing should run without hickups.
That's "Amiga" -like stuff.
(I plan to get back to fractal landscape rendering, 3D rendering as well as to new joys like
NLE and MPEG4 video packing.)
(I'm scared to how my world as Amiga hobbyist might change when there will be
immense speed jump in every aspect, no matter if I buy native MOS or AOS platform.)
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 95 of 134ANN.lu
Posted by Anonymous on 07-Mar-2002 10:56 GMT
In reply to Comment 93 (priest):
Small update:
"I think no-one should expect OS3.1 level quality in OS4.0, untill later this year. "
Change to "...untill later this year, earliest."
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 96 of 134ANN.lu
Posted by Anonymous on 07-Mar-2002 11:01 GMT
In reply to Comment 94 (priest):
argh, another update:
Also thigs like 3D video rendering and 3D game playing should run without hickups.
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 97 of 134ANN.lu
Posted by John Block on 07-Mar-2002 11:32 GMT
We will need utilities to slow things down.
Try scrolling this page 1200% + faster!
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 98 of 134ANN.lu
Posted by John Block on 07-Mar-2002 11:35 GMT
In reply to Comment 97 (John Block):
That's 1200 percent faster :)
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 99 of 134ANN.lu
Posted by Ian on 07-Mar-2002 12:53 GMT
In reply to Comment 91 (priest):
MIPS are meaningless in the context of a benchmark loop. Even more meaningless is the reports that Sysinfo gives of estimated CPU and MHz. (My '060 is a 300MHz '040 apparently).
The best measure we've got is that MUI package, can't recall the name, which calls ImageStudio, LHA, graphics.library, etc.
The issue is not what stupid effective clock speed we can claim, but whether the real-world use of OS4 on a modest PPC will offer a performance increase sufficient to justify upgrading an existing Amiga. Yes, in future we hope to see PPC programs but just now the large software base *and a lot of OS4* is in 68K.
How much faster will Imagine/Real3D/Cinema4D render? How much smoother is TurboCalc? Or Voyager or IBrowse? How quick does Multiview decode a jpg? How quickly does xyz Image Processor blur an image. Real-world tests.
OS 4.0 DEVELOPMENT PROGRESS REPORT : Comment 100 of 134ANN.lu
Posted by Johan "Graak" Forsberg on 07-Mar-2002 13:15 GMT
In reply to Comment 99 (Ian):
The program you're looking for is SysSpeed.
Anonymous, there are 134 items in your selection (but only 84 shown due to limitation) [1 - 50] [51 - 100] [101 - 134]
Back to Top