Submitted by “Back in the old days of the original AmigaOS, the system used to allocate areas of unused memory to new tasks was pretty simple. The old method served its purpose well enough at the time, but with the increased demands of modern computing – and of course the desire to bring this new version of the operating system to the cutting edge – AmigaOS4.0
has introduced a better way of doing things.”
Memory Management in AmigaOS 4.0 Explained
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
> Let us first consider the limitations of the old method.
> Imagine we have an area of free memory of one megabyte in
> size, and we allocate all of it in single-byte blocks.
> Then we free up every second block. In theory the list now
> offers half a megabyte of free store, but you cannot even
> satisfy a single request for two bytes, because nowhere
> in the list is there a block of free memory that is that long.
> […]
> the [new] slab allocator only allocates objects of a single
> size
So exactly how does the new allocator solve the problem above? To me it seems the new allocator makes the problem even worse, since now you have to free a whole “slab unit” of memory before any of that freed memory can be allocated. E.g., if a “slab unit” is 4 KiB then you could free 4095 of each 4096 bytes and you still couldn’t allocate even a single byte even though only 256 bytes is used in that 1 MiB (i.e. 99.98% of the memory is essentially wasted).
Granted, there are other, more important problems that the new allocator indeed solves, but this particular problem is one they mentioned.
I’d say that the number one problem related to AmigaOS 4 is to obtain hardware that runs it…
I’d say that the number one problem related to AmigaOS 4 is to obtain hardware that runs it…
Yep. It’s long overdue that they get over themselves and port it to x86.
Afaics there’s no technical reason to insist on PowerPC, as it’s no more compatible to the 68000 than x86. And WinUAE as well as Apple’s Classic and Rosetta show how to provide compatibility for old programs.
That leaves the old driver problem, but that could be tackled by specifying a small but commonplace set of supported components. You still couldn’t run it on any PC, but at least companies or individuals could easily build faster and cheaper AmigaOS computers than what’s currently available.
@nimble
There has been an “intel” based “Amiga-like” OS out for a while now in the form of AROS.
http://www.aros.org/
Personally speaking, I’m quite happy for the “Offical” Amiga OS to stay on PowerPC as that leaves the door open for interest in embeded systems etc which are big users of PowerPC chips.
AROS is both “intel” compatable but also “open Source” so still provides an Amiga evolution path in the event that the “official” OS stalls or fails for any reason.
To me, that is the best of both worlds.
Regards
Darren
“Yep. It’s long overdue that they get over themselves and port it to x86.”
http://www.aros.org is the closest there is to what you’re suggesting.
“Afaics there’s no technical reason to insist on PowerPC, as it’s no more compatible to the 68000 than x86.”
No, they’re purely historical reasons.
“And WinUAE as well as Apple’s Classic and Rosetta show how to provide compatibility for old programs.”
Indeed.
“That leaves the old driver problem, but that could be tackled by specifying a small but commonplace set of supported components.”
Well, the difference between proprietary PPC hardware (current Apple, the now non-available AmigaOne, Genesi’s Pegasos) and proprietary x86 hardware (future Apple) is small as you’ll only have to support a limited number of drivers.
“You still couldn’t run it on any PC, but at least companies or individuals could easily build faster and cheaper AmigaOS computers than what’s currently available.”
One of the chief reasons the AmigaOne and Pegasos systems are so outdated hardware-wise is the simple fact that unless you’ve got the money needed to design and produce custom chipsets like apple has done, there’s really not a lot to chose from when it comes to chipsets that support PowerPC (remeber folks, we’re talking desktop computers here, not the brakes of a BMW). Had they picked the x86-route, this would not be a problem.
There’s no technical reason for you to insist on x86 either.
I was not meaning to be overly critical, but as an OS enthusiast I am always wanting to try new operating systems. I am also not averse to buying hardware to run these systems on as well as the systems themselves. I am also very familiar with building computers, but every time I search for Amiga stuff I find manufacturer sites that are promising future Amiga part sales, and the Amiga site has been developing OS 4 for several years now with no end in site. If the platform is not supported with hardware and OS releases it will end up being so much vapor ware. The company has changed hands many times at this point making many, including myself, doubt it’s viability as a platform. I would love to see it emerge and grow as an alternate paradigm in modern computing
Edited 2005-12-14 16:39
The hardware already exists, it’s the OS that’s been stagnant for the last X years.
@ Anonymous
IMO the OS4 team are and have been making remarkable progress these last few years. The project isn’t been finalized yet for the mass market, but IMO that’s not the same as saying the project is “stagnant”.
Users aren’t left in the dark and have been receiving lots of progress updates, feedback and have been able to check out the progress themselves at various global events.
IMO developing a state of the art product by a relatively small team of developers and relativily small amount of resources (IMO somewhat made up with personal enthusiasm from lots of developers) takes time. Personally I say the project will take the amount of time that’s required and not any less, I am still young enough. 😉 Having said that, judging from a close perspective things are looking good, very good even!
I’d say it’s the other way around. Obviously the OS exists and has been developed for quite some time but there is no hardware for sale.
“There’s no technical reason for you to insist on x86 either.”
Technical as in technology? No, perhaps not when comparing a PowerPC and a x86 CPU per se, but when comparing the chipsets needed to turn respective CPU into a fully fledged desktop computer there’s every reason to pick an x86 over a PowerPC. Show me any other PowerPC based system than Apple’s that support tecnologies such as PCI-Express and DDR2 RAM. Heck, the AmigaONE and the Pegasos are still stuck at SDRAM memory, not to mention the fact that while the AmigaONE got a slow AGP-bus, the Pegasos haven’t even got a proper AGP-bus at all but is rather using some sort of turbocharged PCI-slot and an AGP-to-PCI bridge.
There is no way to run AmigaOS 4 on modern hardware and what’s worse, there’s no way to obtain hardware capable of running AmigaOS 4. Had it been developed for x86 this would most likely not be an issue.
The PegasosII does support DDR-RAM. Also the AGP is quite performant,
at least enough to get better 3D results running Linux than a PowerMac
G5 running Linux does.
Mr. DevL wrote:
>>>
but when comparing the chipsets needed to turn respective CPU into a fully fledged desktop computer there’s every reason to pick an x86 over a PowerPC.
>>>
It is only a matter of producing different solutions in chipsets.
The market will choice the best ones who will serve PPC motherboards and other hardware PPC based.
Sure unfortunately FIRST PPC chipsets were bugged, or unreliable, or had some problems (such as those from MAI, the infamous Articia chipset). These chipsets suffered some flaws, but actual chipsets for PPC are quite stable and reliable.
There are actually four different chipsets: Apple chipsets, Tundra chipsets, IBM chipsets, and Marvell Chipsets.
This fact gives a vaste opportunity to choice PPC chipset of reference for any kind of PPC motherboards.
PPC based solutions manufacturing firms for have now the chance to choice chipset they want to implement to.
>>>
Show me any other PowerPC based system than Apple’s that support tecnologies such as PCI-Express and DDR2 RAM.
>>>
Freescale based HPC I & II (High performance Computer) were adapted by Genesi to implement their Open Firmware and are now capable to run Pegasos based versions of PPC Linux and even MorphOS.
HPC I is based on Marvell III chipset.
HPC II is based on Tundra TS 108 chipset.
Both chipset support DDR2 RAM SATA and PCI-Express.
Evaluation motherboards are ready and solid silicon (true hardware)
Hope that one of these motherboards could satisfy enough Genesi engineers (expecially Hardware Guru Gerald Carda) to be elected as Pegasos III for the market.
You know it. Germans are very perfectionists and evaluate hardware quite a looong time before choicing it.
But take a look at the photos if you don’t believe me these motherboards exist and peek a spot on their bus slots and SATA connectors.
http://www.pegasosppc.com/image.php?id=1499
http://www.pegasosppc.com/image.php?id=1487
>>>
Heck, the AmigaONE and the Pegasos are still stuck at SDRAM memory, not to mention the fact that while the AmigaONE got a slow AGP-bus, the Pegasos haven’t even got a proper AGP-bus at all but is rather using some sort of turbocharged PCI-slot and an AGP-to-PCI bridge.
>>>
This shows that you are not informed at all on Amiga.
So I wonder why you posted these erroneous informations.
Only to made some terrorism and scare potential buyers?
By the way Pegasos II supported DDR RAM since 2001.
I know it because last spring I got a Pegasos II G3 machine…
…And again to mention it, AGP slot on it is diverted from PCI.
This is just because ancient Marvell II chipset still does not support AGP natively.
But I remember you that even design of AGP slot on other platforms is a modify originated from PCI standard specifications.
Finally, still regarding the main problem you risen (lack of good chipsets aimed at PPC):
If you take a look on new PPC CPU such as 8641 from Freescale (which is about to be launched on the market) you could find that these new PPC processors have a chipset included into its die in a SoC (System on Chip) solution.
This avoids use of external chipsets which could reveal themselves less reliable than expected.
It has been stated that new Pegasi will made extensive use of SoC CPUs.
And Genesi demonstrated it keeps its promises.
Pegasos Efika sublaptop launced on the market just last week it is based on SoC PPC CPU MPC5200B.
Take a look on Efika specs right here on OSNEWS or directly on GenesiPPC site.
http://www.osnews.com/story.php?news_id=12824
http://www.osnews.com/story.php?news_id=12871
http://www.osnews.com/story.php?news_id=12923
http://www.genesippc.com/efika.php
It is a first step to avoid use of third made parties chipsets in the PPC world.
Edited 2005-12-15 12:47
“I’d say that the number one problem related to AmigaOS 4 is to obtain hardware that runs it…”
Well, maybe. But that dosn’t stop the Software developers from err.. developing.
To be honest I don’t really understand that website, but how does other OS do this?
@msundman
This was answered by the articles auther on the Amiga web site here: http://amigaworld.net/modules/news/article.php?storyid=2754&start=2…
(post #26 just in casse the link does not work)
Regards
Darren
Ah, OK. So the slabs don’t fix the problem, but the segregated free list allocation mechanism does (and this mechanism has almost nothing to do with slabs).
…when you can’t buy the a machine to run it. I am all of the time seeing all of this Amiga 4 stuff, but I see no vendors selling it. I wonder if this is a hoax that is being perpetrated on the Amiga community. Is there even an Amiga community? If so what computers do they use? The Amiga was a fine platform in it’s day, but this talking about new Amiga stuff like it is available is pointless. They need to sell the things or drop it.
@ Huehuecoyotyl
> I wonder if this is a hoax that is being perpetrated
> on the Amiga community.
Assuming you have genuine concerns, yes the Amiga community is alive and the AmigaOS4 project albeit still WIP is no hoax.
Hundreds of developers and beta-testers already own AmigaOS4 WIP systems. Ordinary users have access to a pre-release together with 3 updates.
Personally I went to an Amiga birthday event earlier this year with many AmigaOS4 enabled systems on display by developers, beta-testers and users.
Read my report (IMO should demolish any doubt): http://amigaworld.net/modules/features/index.php?op=r&cat_id=1&rev_…
Just because you people are too damn clueless to find out, it doesen’t mean there isn’t hardware to run AmigaOS 4.0 on.
Freetip: AmigaOne and MicroA1
“Just because you people are too damn clueless to find out, it doesen’t mean there isn’t hardware to run AmigaOS 4.0 on.
Freetip: AmigaOne and MicroA1”
…which aren’t in production and can’t be bought. Great tip.
I know where there’s one (yes, one)
http://www.stellardreams.co.uk/showitem.php?id=35
And before anyone starts, yes it is overpriced Hardware. We (Amiga owners) know.
Actually mikroA1’s can be bought from at least dealers in Scandinavia and Northern Europe. But I dare not say how many they have in total.
word missing: …at least >>3<< three dealers…
mA1 for sale? From which dealers?
> mA1 for sale? From which dealers?
The latest batch of mA1s went on sale at Amiga Pianeta 2005 in Italy, but were sold out almost immediately. After this, at the time there was still a unsatisfied in Italy for several dozen units…
You may want to keep track at the AmigaWorld.net classifieds section, but I it may be unlikely that current mA1 owners will sell their system at this point, so it’s may be best to wait for either a new batch to appear or for the Micro-ITX Troika boards which is pretty similar in terms of specifications.
I do believe there’s still a new AmigaOne-XE system on sale in Portugal. Also some beta-testers and developers are running AmigaOS4 on classic PPC-enabled Amiga systems.
AmigaOne-XE? Wooooh .. They’re like impossible to get. I think there’s one micro for sale in GB, and a couple in Sweden yet. And least there was in the end of october.
Contact these guys:
http://www.centroamiga.com/index2.htm
Hi,
I was the lucky winner of a development uA1 running Amiga OS ver 4 pre-release 3 at the Big Bash `BB3′ in Peterborough this year.
It flies and is extremely responsive.
With a 750Fx @800 MHz cpu module on a standard AmigaOne MegArray connector, and a single SODIMM socket populated with 256MB.
FSB speed 133 MHz
PCI slots 32 Bit, 33 MHz, optional riser cards
IDE controller VIA 686B ATA 100, 40 & 44 way headers
Graphics are provided by an onboard Radeon 7000 with 32MB memory.
Ethernet 3COM 10/100 controller
Sound CMI8738 6-channel 5.1 surround sound with CD-ROM, front audio on headers
USB 1.1 2 rear panel sockets, 2 on headers
Legacy I/O PS/2 mouse & keyboard, game, parallel ports on back panel, 1 x serial on header
(The final production µA1-C will have a MagArray 300 connector for PPC processors made by IBM and Motorola and will handle G3/G4 CPU’s up to their current clock limits.)
Regards, Michael
aka rockape
There is nothing Amigalike in the new PPC based systems.
They use fairly generic and standard parts.
They also make Apple computers look cheap and over powered.
Switching to X86 would make tons of sense.
All the people who already own a PC could run the OS then. The only problem is the OS would end up on Bit torrent nobody would buy it, everyone would say it should be opensourced and the developers would go bust.
The only problem is the OS would end up on Bit torrent nobody would buy it, everyone would say it should be opensourced and the developers would go bust.
This is precisely why AmigaOS won’t be and shouldn’t be ported to x86.
The current Amiga market should remain closed and with controlled hardware, otherwise it’d be impossible to get any foothold in any kind of market using AmigaOS.
It’s so tempting to say that you could just port it to x86 and be done with it, but there are way too many competitors on x86.
BeOS died (or stagnated seriously) that way.
Generic PPC hardware availability is a problem, but it’s a much smaller problem than having an OS that has had a lot of money poured in it over the past 5 years, vaporize due to piracy or simply becomes a curiosity and gets a “nice toy” effect and would be forgotten a year after it was ported.
Anonymous wrote:
“Switching to X86 would make tons of sense. All the people who already own a PC could run the OS then. The only problem is the OS would end up on Bit torrent nobody would buy it, everyone would say it should be opensourced and the developers would go bust.”
Are you kidding? They’ve been working on the PPC version for over 5 years and they don’t have everything running on it yet. Switching to X86 would put the launch of Amiga OS 4.0 sometime in the fourth or fifth millennium.
i am curious.. just an idea.. since most probably sp2, nintendo and xbox will come up with harddisk and really fast configuration.. wouldnt be a nice idea to export their work to one of those platforms?
Since amiga 500 was less or more was like consule, and thats how we loved it…
The Amiga had a tight integration between video hardware and the GUI, to give you goodies like draggable screens—which even to this day is not doable with a generic PC. If you know or choose the graphics hardware in the user’s PC, then you can start using the video hardware as part of the GUI, a la MacOSX.
For this reason (and for driver issues) I’m surprised that people care so much about the choice of CPU and so little about the other hardware. It’s not the CPU that makes an Amiga-type OS possible, it’s the video and audio subsystem—or to be more specific, having the OS developers *know* the video subsystem. The CPU is only the heart of a generic PC. You don’t want a generic PC. In an Amiga, the “heart” is somewhere else.
I would imagine that if I and my friends got together to design a “new Amiga,” we’d not care about the Mobo or CPU, but standardize on a single graphics chipset so we can rely on its hardware features to run the GUI.
Actually draggable screens is possible and has been implemented in AmigaOS4 in combination with mainstream graphic cards as well.
What’s not possible with modern graphic cards is to display different screenmodes simultaneously. Hopefully at some point this is about to change.