AmigaOne DMA Problem

[Hooligan_DCS]

The original poster himself "drew" Pegasos into this thread.. sort of... too...

[itix]

Don't know, I can't speak for PegI owners  . All I know is that I can view tv (yes, before you ask, with sound) on my A1 under Linux.

It indicates Articia used on Peg1 and A1 is buggy.

[itix]

everything i read seems to make me lean towards a pegasos2 rather than an amigaone.

There is no working Linux for AmigaOne. They have been trying to implement working UDMA for A1/Linux for 2 years and counting.

but then i also read somewhere that MorphOS might be released for the Mac, in which case I might just be better off getting a cheap one of those.

It is not going to happen very soon but if you can wait an year or two...

[Bodie]

itix wrote:

It indicates Articia used on Peg1 and A1 is buggy.

I thought the April chip fixed all the issues for the PegI :-?.

With reagrds to the tv cards, quite a few users (of those that I know) seem to have them working fine with their A1's. You can draw your own conclusions (and something tells me we won't agree :lol:).

[KimmoK]

itix wrote:

All I know is that I can view tv (yes, before you ask, with sound) on my A1 under Linux.

It indicates Articia used on Peg1 and A1 is buggy.

 :-D

Makes me wonder what it would indicate if TV viewing would not work.

[KimmoK]

captainmoomoo wrote:
everything i read seems to make me lean towards a pegasos2 rather than an amigaone.

Buying amigaone you would support AmigaOS platform. Buying pegasos2 you support MOS or at least Genesi & Mr Buck.
Today, peg2 might give you more Amiga"fun", though.

but then i also read somewhere that MorphOS might be released for the Mac,

IIRC, genesi has not said they will release MOS for Mac.

MOS is developed by a HW company to help to sell pegasos HW.
What ever some Genesi head says, releasing MOS for Mac would hurt pegasos2 sales. (a little bit like releasing it for CSPPC/BPPC)

IIRC, Buck has said that there could be CD's with some MOS runtime to run some MOS application/game (from that CD) on Mac HW.


On the other hand... AmigaOS4 is being developed by a SW company. Their plan/strategy is to get it running on several third party HW platforms, as long as it makes market sense in general (more in the KMOS interview). Untill AOS4 has matured a year or two I do not think we will see AOS running on other platforms than A1 and classic PPC equipped Amigas.


Check through the alternatives, listen to your conscience etc... and then get the best HW for you.

If you want to use Linux, most likely neither A1 or Pegasos is the best option (and especially not A1).
If you want to use  AOS4.x, A1 would be the only option currently.
If you want to use MOS, Peg2 would be the only option...


About the DMA problem. It seems that it's not a problem for AOS4.
Only time will tell if the "SW cache coherency" affects AOS performance. So far AOS4 DMA drivers have not been released outside beta tester group (betatesters indicate that it works well), so it's too early to tell.

Looking outside AOS world: MAI really messed up. They have pretty much lost all their Linux market possibilities for ArtisiaS equipped HW.

[takemehomegrandma]

@ HotRod

I've read the specs for the Atricia S since some months back (allmost a year) and I haven't seen DMA mentioned for the Atricia S.

You would have to go much further back in time than that. A year ago, both the April1 and April2 patches had been released, and also the Articia based Pegasos 1 motherboard had officially been dropped altogether in favour of the DiscoveryII based Pegasos II. These issues was put forward to MAI much long before that.

Besides, I think it's rather funny that you claim that DMA is there and it's working flawlessly, when the manufacturer themselves does not even put this among the features for this particular product (they certainly do for their other (vapor since 4 years) northbridges, since DMA is a too important feature to leave out of a product description if you want to sell northbridges today). Why do they not put DMA among the features for the Articia S northbridge?

It's handled in software, not in hardware.

:-o :-o :-o

Uh, DMA in -->SOFTWARE<--?!??

The AOS 4 driver works. It is 100% reliable.

I actually have an A1 with OS4 pre-release right here, right now. Perhaps this rather unique "software implementation of DMA" would explain the "speed" in file transfers? :-D


@ Bodie

Don't know, I can't speak for PegI owners  . All I know is that I can view tv (yes, before you ask, with sound) on my A1 under Linux.

...

With reagrds to the tv cards, quite a few users (of those that I know) seem to have them working fine with their A1's. You can draw your own conclusions (and something tells me we won't agree ).

I am happy for you! :-)

Just a quick note about this; from what I understand, some TV-cards use IRQ/CPU to transfer data from the TV-card to the Graphics card, while other uses direct PCI->AGP DMA for its transfer (which completely leaves the CPU out of the picture). Some cards can perhaps even use both ways, I don't know. I am glad that your particular TV card is usable on your system! :-)

I thought the April chip fixed all the issues for the PegI

Sure, it fixes most of the issues. However, it can not "fix" things that was not implemented in the first place (like PCI->AGP DMA for instance). Anyway, the absolutely best fix against the Articia issues was the Marvell DiscoveryII (and I am SOO glad for that one)!  :lol:  :-D
 

[takemehomegrandma]

KimmoK wrote:

If you want to use Linux, most likely neither A1 or Pegasos is the best option (and especially not A1).

It depends on the application! If you want to run Linux and *real silence* is important, but without compromising too much with horse power, for instance in a media box for your living room, and real fast network speed with one (and possibly two?:-o) Gigabit Ethernet lines straight into the fast Northbridge (without going through the PCI bus) and one "traditional" 100Mbit PCI based ethernet (how is that for a home server/router/firewall for your 24/7 home network/Direct Connect efforts? ;-)  :-P ), then the Pegasos is an EXCELLENT choice! All major Linux distros runs fine, and it's only getting better all the time! :-)

[Hooligan_DCS]

Talking of bridges.. was I the only one who was VERY sceptic towards Marvell-Pegasos2 solution back then? Really... I thought it would never work.
One could say now, afterwards, that it was a good decision to switch to Marvell's product..

(it's us sceptics who make the world worse ;-))

[DrBombcrater]

Why do they not put DMA among the features for the Articia S northbridge?

When you see 'DMA' mentioned on the spec sheet of a North Bridge it is referring to a seperate controller used to run high-bandwidth devices (such a Gb ethernet) that would saturate the PCI bus. This was't a popular feature when the ArticiaS was designed, so it doesn't have one.

DMA over the PCI bus is handled by the North Bridge's PCI bus mastering unit. The BM unit in the ArticiaS can handle up to 5 bus masters at one time and I've seen nothing to indicate that it doesn't work just fine.

Uh, DMA in -->SOFTWARE<--?!??

Well, no, not really. Actual DMA transfers on the ArticiaS work just the same as they do on any another North Bridge. The 'software' part comes when the transfer has completed. Most systems have hardware that will flag the CPU to do a partial cache-flush at ths point, in order to make sure the caches reflect the new data just DMAed into memory. The ArticiaS can't do this, so after DMA activity the driver must explicitly call an OS function that does the cache updating.

This isn't exactly an elegant approach (I'd go as far as to say it was a rather stupid design decision on the part of Mai) but for OS4 is should be a moot point. Performance and stability should be just as good as a system with hardware cache-coherency provided the driver writers are paying attention.

Just a quick note about this; from what I understand, some TV-cards use IRQ/CPU to transfer data from the TV-card to the Graphics card, while other uses direct PCI->AGP DMA for its transfer (which completely leaves the CPU out of the picture). Some cards can perhaps even use both ways, I don't know. I am glad that your particular TV card is usable on your system!

Most TV cards are based on BT848/849 chips which do use DMA to push picture data into memory. I've seen one of these cards (a Hauppauge WinTV-Go to be exact) working on an A1 under Linux without any apparent issues.

Anyway, the absolutely best fix against the Articia issues was the Marvell DiscoveryII (and I am SOO glad for that one)!  

The Marvell is certainly a nice North Bridge and a better choice than the ArticiaS if you want a system capable of running multiple operating systems, but for a machine intended to run just one OS that is specifically designed for it the ArticiaS works okay.

My only real gripe about it is the memory compatibility problems. It's rather insane that there are no widely available Dimms guaranteed to work with it, and even if you buy the exact module recommended by Eyetech there's still a chance it won't work.

[smithy]

I actually have an A1 with OS4 pre-release right here, right now. Perhaps this rather unique "software implementation of DMA" would explain the "speed" in file transfers?

How about writing a review, comparing MOS and OS4? :-)

[Piru]

for a machine intended to run just one OS that is specifically designed for it the ArticiaS works okay.

Marvell is over 100% faster with HD DMA than ArticiaS. Same southbridge, same HD.

I am no expert, but to me it seems as if Articia was seriously crippled.

[takemehomegrandma]

DrBombcrater wrote:

Actual DMA transfers on the ArticiaS work just the same as they do on any another North Bridge. The 'software' part comes when the transfer has completed. Most systems have hardware that will flag the CPU to do a partial cache-flush at ths point, in order to make sure the caches reflect the new data just DMAed into memory. The ArticiaS can't do this, so after DMA activity the driver must explicitly call an OS function that does the cache updating. This isn't exactly an elegant approach ...

This doesn't sound too healthy to me. BTW, isn't the point of DMA to keep the CPU out of the transfer process?

Anyway, perhaps both "feature+driver" and "bug+workaround" could be used to describe this (depending on your point of view I guess)! ;-)

Most TV cards are based on BT848/849 chips which do use DMA to push picture data into memory. I've seen one of these cards (a Hauppauge WinTV-Go to be exact) working on an A1 under Linux without any apparent issues.

I am in no way an expert on TV cards, but as you said, it seems like many of them are based on Brooktree/Rockwell/Connexant chipsets Bt848, Bt849, Bt878, Bt879 (at least according to the "overview" at http://www.tv-cards.com/faq.php, click on "Are all PCI TV-Cards the same?"), and at least the Bt848, Bt848A and Bt849A seems to support DMA transfer. But they should all still be able to (at least it looks that way in my ignorant eyes) fall back to IRQ/CPU transfer, righ (DMA is of course the best, but not *required*)?

"What is the minimum PC specification to use a PCI TV-Card?
 
TV-Cards based on the Brooktree/Rockwell/Connexant chipsets require the following minimum spec:

- PCI slot
- Pentium 133Mhz Processor
- DirectX Compatible sound card
- Graphics card supporting DirectDraw and Overlay
- Available IRQ

TV-Cards usually require an IRQ and a memory range but NOT an I/O address or DMA channel."

(This was from the same site (http://www.tv-cards.com/faq.php), click on "What is the minimum PC specification to use a PCI TV-Card")

How do you know that the A1 you saw was really using DMA and not falling back to using the IRQ to ask for CPU-time for it's transfer? How did you tell the difference?

BTW, "Frostwork" in this thread has a BT878 based TV card working fine in his Pegasos 1 too, so ... ;-)

Anyway, thank you for your clarifications regarding the Articia DMA. When some people talk about it, it really sounds to be a great chip! Many people seems to be really happy about it (Amiga - back for the future ;-)), and I don't want to be a party pooper here. If you people are happy about it and thinks that it's a solid ground to build an Amiga future(?) on, then I won't disturb you! But personally, I am very glad that the Pegasos moved away from it. ;-)

Have a nice day people! :-)

[takemehomegrandma]

smithy wrote:

I actually have an A1 with OS4 pre-release right here, right now. Perhaps this rather unique "software implementation of DMA" would explain the "speed" in file transfers?

How about writing a review, comparing MOS and OS4? :-)

Hmm, that sounds like a plan, doesn't it? ;-)

[KennyR]

KimmoK wrote:
Buying amigaone you would support AmigaOS platform.

How?

The AmigaOS platform is now the users: UAE, Amithlon, MorphOS, OS4, original Amiga, AROS. How is buying an AmigaONE going to support anything? The hardware is owned by Eyetech and Mai, the software by Hyperion and "KMOS"*, the label by the now-for-all-intents-and-purposes defunct AInc. I really don't see how buying an A1 supports the 'Amiga platform' any more than buying a Pegasos or a copy of Amiga Forever, or downloading AROS.

* No proof that KMOS is an actual company yet, and likewise no proof that they are not just a cynical Amiga Inc. investor trick. Supporting them could be the same as supporting Amiga Inc - DAMAGING the Amiga platform.

[DrBombcrater]

@Piru

Marvell is over 100% faster with HD DMA than ArticiaS. Same southbridge, same HD.

Since the UDMA driver for OS4 is still in development and not publically available, I fail to see how you can state this with any authority.

The speed of UDMA transfers depends mostly on the efficiency of the PCI controller anyway (which is why Intel chipsets often worked faster than VIA ones in years past) but the software/hardware coherency issue should not be much of a factor.

@takemehomegrandma

This doesn't sound too healthy to me. BTW, isn't the point of DMA to keep the CPU out of the transfer process?

The CPU is kept out of the transfer process. It's only after the transfer has completed that a cache update needs to be performed. The only difference between the two architectures is that on software coherency systems like the A1 you need to make an OS call after the transfer. The overhead for this is minimal and probably not detectable even using specific benchmarks.

I may be wrong (crappy memory) but I'm pretty sure 68k Amigas use a software cache coherency system.

How do you know that the A1 you saw was really using DMA and not falling back to using the IRQ to ask for CPU-time for it's transfer? How did you tell the difference?

Because it can't work without PCI DMA. The 'DMA' the TV card spec sheet talks about is a legacy ISA DMA channel and is quite different from DMA on the PCI bus. In the days of ISA PCs a controller on the motherboard (its implmented in the South Bridge now) was used to pull data out of expansion cards, and every card needed to be allocated its own channel.

The PCI bus doesn't use this hardware as it has its own bus mastering system, and you'll never see 'DMA' demanded on the specs of a PCI card because it is an integral part of the PCI specification.  The ISA DMA channels are just a hang over and don't get used on PPC systems at all unless they implement an LPC bus. Even on a modern PC they mostly get ignored: the sound card, graphics card and HD controllers on my Windows box all function without a legacy DMA channel.