20-Apr-2024 13:55 GMT.		 UNDER CONSTRUCTION
Anonymous, there are 122 items in your selection [1 - 50] [51 - 100] [101 - 122]
[Forum] Articia: Still DMA-problems?ANN.lu
Posted on 20-May-2003 11:16 GMT by Senex122 comments
View flat
View list
From a news-item at VGR.com it appears that the current Articia-northbridges might still have DMA-related problems: "Mr. S.W. who sends over screen shots of newer AmigaOne 800mhz G4 hardware, has sent over 2 screen shots of the new A1 with G4 running 'amigaone xe/g4 7451 800mhz, excalibur radeon 9000 pro 64mb ddr ram, valueram kvr133x72rc3/256 kingston'. He wants to know '.. if other aone users have that problem as well..', 'it never outputs the same checksum with dma enabled when testing an iso file which is larger than the swap space used'

You can see the Radeon 9000 Pro screen shots here: DMA On & DMA Off."
Articia: Still DMA-problems? : Comment 1 of 122ANN.lu
Posted by Anonymous on 20-May-2003 09:39 GMT
1) Mister S.W? Normally VGR shows full names.
2) Why ask at VGR (1):(you know the site that is not really independant in this case)? When there's an AmigaONE mailinglist. I checked and the question didn't show up there
3) Why ask at VGR (2): There is no possibility to reply the author of the question because there is no comments section; there's no email adres in the posting.
4) And who says it's an AmigaONE? Just because the prompt says so?
5) Someone already mentioned the timing, there was positive news about AmigaOS4
yesterday, so you can start counting the hours when some negative news would hit the news sites.

But anyway. I never heard of any data corruption with the AmigaONE XE ( and believe if there is any then you will notice a hell lot more then only a failing CRC) so I guess I can answer this question for you: No you're the only one.
Articia: Still DMA-problems? : Comment 2 of 122ANN.lu
Posted by Jupp3 on 20-May-2003 09:56 GMT
In reply to Comment 1 (Anonymous):
> 4) And who says it's an AmigaONE? Just because the prompt says so?

I prefer the name "Teron" too... But as so many want to call it "AmigaOne" let it be...
Articia: Still DMA-problems? : Comment 3 of 122ANN.lu
Posted by Anonymous on 20-May-2003 10:00 GMT
In reply to Comment 1 (Anonymous):
>But anyway. I never heard of any data corruption with the AmigaONE XE ( and >believe if there is any then you will notice a hell lot more then only a >failing CRC) so I guess I can answer this question for you: No you're the >only one.

I think noone cares if Mr. Anonymous did not hear of any corruption. It looks like its easy to verify if I look at the pictures, so just let some people try ....
Articia: Still DMA-problems? : Comment 4 of 122ANN.lu
Posted by Masta Boom on 20-May-2003 10:01 GMT
In reply to Comment 1 (Anonymous):
Everyone can repeat the very same test on his/her AmigaOne XE. Why don't we simply wait until someone else (who is willing to give his full name and owns an A1XE) confirms this, or not?

It doesn't matter whether there is positive AmigaOS4.0 news out or not. If Eyetech sells buggy hardware, we MUST know this.
Articia: Still DMA-problems? : Comment 5 of 122ANN.lu
Posted by Rik Sweeney on 20-May-2003 10:13 GMT
Sorry, but what am I looking for in these screenshots?
Articia: Still DMA-problems? : Comment 6 of 122ANN.lu
Posted by Ian Shurmer on 20-May-2003 11:03 GMT
In reply to Comment 5 (Rik Sweeney):
Okay just a couple of points:

a) What's to say the screenshots are even running on an AmigaONE?
b) What's to say that if it is running on an AmigaONE that it is not one of the unfixed SE models?
c) Why on earth would "Mr. S.W." post to vgr.com rather than e-mailing Eyetech, or posting on the AmigaONE mailing list?
d) Why haven't we heard about this from other XE owners?

Seems more than a little bit suspect to me... anyone from vgr.com, or Mr S.W. himself want to clarify the above questions?

Cheers,
Ian
Articia: Still DMA-problems? : Comment 7 of 122ANN.lu
Posted by Jürgen Schober on 20-May-2003 11:06 GMT
I've already added a statement at Amiga News about the issue.

In what way does this report mention the ArticaS ??

What it describes is a known issue with the VIA686B and IDE DMA modes with the current Linux drivers. This is worked on. That's it.

lg
Jürgen
Articia: Still DMA-problems? : Comment 8 of 122ANN.lu
Posted by Jürgen Schober on 20-May-2003 11:11 GMT
In reply to Comment 7 (Jürgen Schober):
Ah, and yes, to add to my own post.

I'll give that test a try this evening. I am running the G4 and the (933MHz) G3 since a while with UDMA IDE on, works fine so far. To be on the secure side, you can disable DMA by setting ide=nodma in the kernel settings in the UBOOT Firmware.

lg
Jürgen
Articia: Still DMA-problems? : Comment 9 of 122ANN.lu
Posted by mahen on 20-May-2003 11:22 GMT
I had the same problem with a pegasos without april. Fortunately april 2 fixed this, pfew !

Now, maybe it's something else...
Articia: Still DMA-problems? : Comment 10 of 122ANN.lu
Posted by Anonymous on 20-May-2003 11:22 GMT
In reply to Comment 6 (Ian Shurmer):
Exacly my point. Funny that mister anonymous in posting 3 doesn't understand/like my anonymous post and doesn't even understand that wether it's dispelled or not, people will always remember: Hey wasn't there a DMA problem with the AmigaONE XE? I'm sure I read it somewhere....

So posting this kind of "rumours", even with a very suspicious contents does always damage a product. And I guess that's exactly the purpose of that post.
Articia: Still DMA-problems? : Comment 11 of 122ANN.lu
Posted by Ian Shurmer on 20-May-2003 11:30 GMT
In reply to Comment 10 (Anonymous):
Exactly... "good" timing as well!

Ian
Articia: Still DMA-problems? : Comment 12 of 122ANN.lu
Posted by Olegil on 20-May-2003 12:02 GMT
In reply to Comment 5 (Rik Sweeney):
One window there shows that with DMA, he gets different checksums (means different file content read) each time he calls the checksum function.

Why this means Articia is broken is beyond me. Why can't it be the actual chip that handles DMA transfers, namely the VIA southbridge. Personally, I think this is a Linux driver issue (refer to a quote by Thomas Frieden I posted the other day), but I must admit to not being all that certain. I feel like a newbie some days :-)
Articia: Still DMA-problems? : Comment 13 of 122ANN.lu
Posted by Joe Pillow on 20-May-2003 12:17 GMT
In reply to Comment 12 (Olegil):
"Why can't it be the actual chip that handles DMA transfers, namely the VIA southbridge."

Because the very same problem happened on non-April PEGASOS mainboards, but doesn't on PEGASOS boards with an April fix. Keeping this in mind, it is likely that ArticiaS *is* the problem. However, I don't expect you or anybody else who is strongly involved in the AmigaOne project, to admit this.

It definetely would be interesting to see whether ordinary users of the A1XE have the same problem on *their* machine or not. The current information was posted by an anonymous person afterall.
Articia: Still DMA-problems? : Comment 14 of 122ANN.lu
Posted by Ben Hermans/Hyperion on 20-May-2003 12:23 GMT
In reply to Comment 13 (Joe Pillow):
The reason why we know it's a VIA 686B Linux driver related problem is simply because it doesn't show up when using a different IDE controller located on the PCI bus or a SCSI controller.

If it really were an Articia S problem, you'd expect to see the same issues with all DMA transfers regardless of the controller.
Articia: Still DMA-problems? : Comment 15 of 122ANN.lu
Posted by Olegil on 20-May-2003 12:29 GMT
In reply to Comment 14 (Ben Hermans/Hyperion):
Does indeed remind me I have borrowed a scsi controller I should put to more use. Of course I can try that. But I don't have any really fasts disks... Nor silent ones :-)
Articia: Still DMA-problems? : Comment 16 of 122ANN.lu
Posted by Olegil on 20-May-2003 12:34 GMT
In reply to Comment 13 (Joe Pillow):
Ok, I see why I shouldn't be trusted, but please just concider this:

How do you know it's the same problem?

Because the data are being corrupted? If the VIA chip had a bug you could have data corruption (and I think I have seen pages proclaiming the SB to do this unless you have the right drivers, something I cannot guarantee we have managed to get into the Linux kernel, since the driver was written with x86 in mind and have shown to contain at least one endian issue). I really wish more people would get involved with the kernel programming...

What I'm saying is that if people I know and work with at Hyperion are saying it's not the NB, I personally choose to believe them over someone who does not even claim to know, just speculates about mine and their involvement in the project.

But I know it might be hard to trust an obviously blind and naive AOS4 supporter such as myself. But please be assured I'm actively giving the Friedens a hard time about this elsewhere so we could get some real evidence up and about :-)
Articia: Still DMA-problems? : Comment 17 of 122ANN.lu
Posted by Senex on 20-May-2003 12:42 GMT
In reply to Comment 7 (Jürgen Schober):
@6

Sorry, been just my guess it might be the Articia (see other comments, the prob sounded familar from un-Apriled Pegasosses). Indeed the original news-item did not mention the NB at all. My fault.

Just wanted to avoid writing: "AmigaOne might still..." - because then everyone would have shouted: "FUD". This way it's been more neutral while still spreading the information/question if there might be still issues with the Teron or Articia. (Well, Pegasos-Beta2-ML readers do know with the latter there are.)
Articia: Still DMA-problems? : Comment 18 of 122ANN.lu
Posted by Mike on 20-May-2003 12:46 GMT
In reply to Comment 14 (Ben Hermans/Hyperion):
@Ben Hermans

> The reason why we know it's a VIA 686B Linux driver related problem is simply
> because it doesn't show up when using a different IDE controller located on the PCI
> bus or a SCSI controller.
> If it really were an Articia S problem, you'd expect to see the same issues with all DMA
> transfers regardless of the controller. "

Then does it surprise you if I tell you that PCs using the same southbridge (VIA 686B) and Linux (even not the latest) doesn't have this problem? (Of course they have a different northbridge.)
Maybe you'll tell me this only happen on PPC. But I don't think so. The drivers are exactly the sames, and I don't see how the CPU can make the bug appear suddendly.

I think you conclude too quickly.
Articia: Still DMA-problems? : Comment 19 of 122ANN.lu
Posted by Christophe Decanini on 20-May-2003 12:52 GMT
While everyone looks to argue if it is an ArticaS or not problem I want to point out that if there is really such a problem on all AmigaOne it should be resolved ASAP before some people corrupt their data.

Another point is that some resellers are selling both the Pegasos and the AmigaOne without any notice about the missing parts / beta status.
Who is responsible if a user loss data because of dysfunctioning HW/SW ?

Sorry to sound negative but I think that these problems should have been addressed before selling systems to Joe User.
Articia: Still DMA-problems? : Comment 20 of 122ANN.lu
Posted by priest on 20-May-2003 12:54 GMT
In reply to Comment 13 (Joe Pillow):
"Because the very same problem happened on non-April PEGASOS mainboards"

But the equivalent of April(1) fix should be in the new ArtisiaS...
(no more jump wires & patch chip on the A1 board)
Articia: Still DMA-problems? : Comment 21 of 122ANN.lu
Posted by Christophe Decanini on 20-May-2003 13:04 GMT
In reply to Comment 20 (priest):
The articia situation is particularly unclear:
What was fixed with April 1 ?
What was fixed with April 2 ?
What was fixed with the wire fix for AmigaOne ?
What was fixed with the new Articia version ?

Which one of these may have data corruption ?
Articia: Still DMA-problems? : Comment 22 of 122ANN.lu
Posted by Ian Shurmer on 20-May-2003 13:06 GMT
In reply to Comment 19 (Christophe Decanini):
@Christophe

Of course, the problem should be resolved as quickly as possible IF there is one... Would any AmigaONE owners care to comment on any DMA problems? Could they carry out the same test as Mr. S. W. and post their screenshots?

Cheers,

Ian
Articia: Still DMA-problems? : Comment 23 of 122ANN.lu
Posted by Jürgen Schober on 20-May-2003 13:08 GMT
In reply to Comment 19 (Christophe Decanini):
I understand. And this is a serious issue. Believe me. That is also the reason why I mentioned we have known trouble (see posting above).

Please note. The board is a new product. We had to go public with it. These sort of things happen at startup. That's usual.

We (and Hyperion/Eyetech/Mai) do everything we can to get the issue sorted out for our customers ASAP. At least we will see a Firmware/Kernel update thru retail dealers as soon as we have a solution.

lg
Jürgen Schober
point.design
Articia: Still DMA-problems? : Comment 24 of 122ANN.lu
Posted by Christophe Decanini on 20-May-2003 13:27 GMT
In reply to Comment 22 (Ian Shurmer):
Looking at the answer from Jurgen it looks that the problem exists.
All users should be advised to disable their DMA if they want to keep their data safe.
Articia: Still DMA-problems? : Comment 25 of 122ANN.lu
Posted by corpse on 20-May-2003 13:41 GMT
"He wants to know '.. if other aone users have that problem as well..', 'it never outputs the same checksum with dma enabled when testing an iso file which is larger than the swap space used'"

The openBSD tars seemed to self destruct on my x86 machine, cause the MD5 sums would randomly go from valid to invalid and back again. Turned out to be: defective RAM and the connector of the ATA100 cable connected to the drive.

It seems the MAI chip is fussy about RAM so it may be one(or more)of the modules your using isn't up to being used in the AmigaOne or that its faulty. IIRC the AmigaOne uses the VIA686B which is known for data corruption with certain devices connected to the PCI bus(e.g. SBLive) and in heavey load conditions, Both can be fixed by using a driver set with the required fixes or by patching the PCI registers to extend the latency slightly.
Articia: Still DMA-problems? : Comment 26 of 122ANN.lu
Posted by David Scheibler on 20-May-2003 13:52 GMT
In reply to Comment 23 (Jürgen Schober):
>We (and Hyperion/Eyetech/Mai) do everything we can to get the issue sorted out
>for our customers

Just my two cents:

You should have informed your customers before they actually bought the
hardware. From various comments you made before (here and on amiga-news.de)
everybody got the impression that there is no known DMA bug. But if I
understood you correctly now, you already knew about it for quite a time?!
Articia: Still DMA-problems? : Comment 27 of 122ANN.lu
Posted by Georg Steger on 20-May-2003 14:08 GMT
In reply to Comment 14 (Ben Hermans/Hyperion):
> The reason why we know it's a VIA 686B Linux driver related problem is simply because it doesn't show up when using a
> different IDE controller located on the PCI bus or a SCSI controller.
>
> If it really were an Articia S problem, you'd expect to see the same issues with all DMA transfers regardless of the controller.

Hmm ... don't know that much about hardware stuff, but in
software area such conclusions unfortunately might not necessarily
always turn out to "work".

Own experience: software worked fine when compiled with gcc.
software crashed when compiled with egcs. egcs was known to
be pretty buggy. blamed crash on egcs. Later it turned out
that the crash was actually caused by a bug in the software.
nothing to do with compiler. It not crashing on gcc was pure
luck.

--
Georg Steger
Articia: Still DMA-problems? : Comment 28 of 122ANN.lu
Posted by Anonymous on 20-May-2003 14:10 GMT
In reply to Comment 26 (David Scheibler):
..there are bugs in almost all motherboards. even the one i am using now has a nice workaround built into the Linux kernel - and even the windows drivers for
the chipset have workarounds.

all people running VIA boards also have to get their latest VIA 4-in-1 fix each month
Articia: Still DMA-problems? : Comment 29 of 122ANN.lu
Posted by Olegil on 20-May-2003 14:11 GMT
In reply to Comment 26 (David Scheibler):
Actually, the bug we know about is solvable (because it's been solved in x86 drivers by VIA), but not yet fully solved. The thing is, a lot of people NEVER stop to think that it might be a driver issue, and all they base it on is that "well, people who would say that are saying it, so it must obviously be a lie".

_That_ is what bites me.

There's a new kernel coming up, based on 2.4.21. Those who are interested can already get it from the cvs server (sourceforge). It's been largely done by MAI, and some after-effects added by Ross Vumbaca. So we're not just saying we're working on it, there's actually some work going on. But with 7 out of 10 people on the _developer_ developer list working on OS4 at the moment, it is taking a bit more time than it ought to. I admit to not being all that good at linux kernels, ok? :-)

I have a good understanding of the build system, but some of the coding seems like pure gibberish to me. Maybe it's because it IS gibberish, but I tend to blame myself :-)
Articia: Still DMA-problems? : Comment 30 of 122ANN.lu
Posted by Olegil on 20-May-2003 14:14 GMT
In reply to Comment 28 (Anonymous):
Excactly. It's just that we're really stretched thin here. We don't have a huge crowd of people working on it, we're just a couple of guys. And I'm working full time just answering user questions... :-(

Imagine if VIA had an operation like that, eh? :-)
Articia: Still DMA-problems? : Comment 31 of 122ANN.lu
Posted by Johan Rönnblom on 20-May-2003 14:24 GMT
To Olegil, Ben et al:

How can you be so damn sure that the problem is with the VIA chipset
if you can't even fix the problem yet?!

How can you be so quick to refute alternative theories - put forth by
reputable hardware and software experts - when you *still*, after
several months, don't know enough about the problem to solve it?
Articia: Still DMA-problems? : Comment 32 of 122ANN.lu
Posted by Olegil on 20-May-2003 14:31 GMT
In reply to Comment 31 (Johan Rönnblom):
Because certain people we work with has put forth evidence we believe in.


That's my final word on this for today, now I'm gonna play Civ3 for a bit...
Articia: Still DMA-problems? : Comment 33 of 122ANN.lu
Posted by DaveP on 20-May-2003 14:33 GMT
In reply to Comment 31 (Johan Rönnblom):
Politician: Ah, lets see, theres something wrong with this bridge, cars keep driving over it
and 1 in 10 lose tyre tread.

Engineer: the bridge is flexing, visible fault lines appear in the surface when certain traffic patterns pass by.

Politician: What should we do resurface it?

Engineer: We could, temporarily but we are going to have to keep resurfacing it, the real problem is in one of the supporting struts which is defective. This needs to be replaced.

Politician: How can you be so damn sure thats the problem until you've fixed it?

Engineer: By observation and logic as well as a basic knowledge of engineering.
Articia: Still DMA-problems? : Comment 34 of 122ANN.lu
Posted by Ferry on 20-May-2003 14:40 GMT
In reply to Comment 7 (Jürgen Schober):
Useful info about VIA 686B bug on

http://alive.singnet.com.sg/tech/via.htm

This page has also other links (Via FAQs and forums, The Register, etc.)

Ferrán
Articia: Still DMA-problems? : Comment 35 of 122ANN.lu
Posted by David Scheibler on 20-May-2003 14:40 GMT
In reply to Comment 7 (Jürgen Schober):
>What it describes is a known issue with the VIA686B and IDE DMA modes with the
>current Linux drivers

Yes, but only when copying data from one IDE channel to the second under high
PCI bus last (read the c't article I posted to amiga-news.de).

However if you read what this AmigaOne owner wrote and how he made the test
(only one IDE HD, etc.), then it doesn't sound like the bug above. Of course it
might be a different VIA bug which is currently unknown and only happens on
AmigaOne systems (and unfixed Pegasos boards).
Articia: Still DMA-problems? : Comment 36 of 122ANN.lu
Posted by Leki on 20-May-2003 14:50 GMT
"Your account is currently unavailable.
Please contact your support representative"

thats all im gettin on vgr so far
Articia: Still DMA-problems? : Comment 37 of 122ANN.lu
Posted by David Scheibler on 20-May-2003 14:54 GMT
In reply to Comment 36 (Leki):
You can still download the provided pictures.
Articia: Still DMA-problems? : Comment 38 of 122ANN.lu
Posted by Ferry on 20-May-2003 14:54 GMT
In reply to Comment 34 (Ferry):
More info...

http://www.au-ja.org/review-kt133a-1-en.phtml

Ferrán.
Articia: Still DMA-problems? : Comment 39 of 122ANN.lu
Posted by Snowfox on 20-May-2003 15:17 GMT
I had one of these buggy x86 via686 boards and used it under linux. You can only test that with big files and doing the same thing over and over again and then compare results. And no, it was not only from one channel to the other channel. I had same prob when working on only one harddisk as well. I remember i only formated one harddisk and had filesystem errors right after that. I didn't copy anything. So i dunno, all i can say is it is very annoying and i just threw it away as i found out what exactly was causing the errors. new board - problems gone. so i dunno i heard a new bios would fix it as well. who knows hehe.
and btw.: harddisk access without dma access is damn slow. so turning it off is no real option.
Articia: Still DMA-problems? : Comment 40 of 122ANN.lu
Posted by Leki on 20-May-2003 15:22 GMT
AmigAOne user: i spilt water on my HD and now it doesn`t work properly.

Morphos fanatic: it has to be because of Articia DMA problems.

Olegil: umm i think its probably due to the water being split on it.

Morphos fanatic: dont listen to olegil he is involved with AmigaOne and so must obviously be lying and cant be trusted because of that.

Bill Buck: remeber folks, theres no mai without april and G3 is faster than G4.


.....heh i was bored ;)
Articia: Still DMA-problems? : Comment 41 of 122ANN.lu
Posted by Alkis Tsapanidis on 20-May-2003 15:29 GMT
In reply to Comment 40 (Leki):
1) There's a CONVERSATION here, *NOT* a flamewar... YOU are about to start a
flamewar...
2) "Bill Buck: remeber folks, theres no mai without april and G3 is faster than G4."

You can be sued for puting word in people's mouths... That's NOT what he said.
He said that a G3 *SYSTEM* is faster than a G4 *SYSTEM* in practice.
This doesn't have to do with CPU speed, it's well known that the G4 is faster.
And about "there's no April without Mai"... Prove the opposite... With the old
Articia, the April was the ONLY perfectly working fix. With the new one, we'll
see when the boards mature.
Articia: Still DMA-problems? : Comment 42 of 122ANN.lu
Posted by Frodon on 20-May-2003 15:41 GMT
Hello,

I read on the latest URL posted some comments before:

" It is rumoured that a new revision of the chipset has been released to fix the problem." And this was written in 2001.

Did VIA made a new version of this southbridge? I find quite hard to believe that they didn't provided a fixed revision yet as this was already a known bug in 2001.

In the case they did provide a fixed version, then why AmigaOnes don't use it (apparently)?
Or if the AmigaOne indeed use a recent version, then why is there still the bug if it's this VIA bug?

Regards
Articia: Still DMA-problems? : Comment 43 of 122ANN.lu
Posted by hooligan/dcs on 20-May-2003 15:43 GMT
In reply to Comment 40 (Leki):
Not to say, maybe a little tired aswell?
Articia: Still DMA-problems? : Comment 44 of 122ANN.lu
Posted by Leki on 20-May-2003 15:45 GMT
In reply to Comment 41 (Alkis Tsapanidis):
1) from reading the posts thats the jist of the conversation i get.

2) i disagree....but we both wont come to an agreement about that so lets leave it at that rather than reall starting a flame war.
Articia: Still DMA-problems? : Comment 45 of 122ANN.lu
Posted by Johan Rönnblom on 20-May-2003 16:03 GMT
In reply to Comment 44 (Leki):
I'd say it's more a bit like this:

(rumour) There's a dma bug in A1 and P.
P camp: (silence)
A1 leaflet: There is a bug!
P engineers: The bug is in A.
P boss: We'll replace all P's to fix A.
A1 lawyer: For the record, there is no bug in A.
P boss: There's no A without B.
A1 boss: There's no bug.
A1 engineers: (silence)
(all P's are replaced, at great cost)
P boss: P didn't work! We had to fix it!
A1 boss: We've fixed the A1.
A1 lawyer: The new A1 has the A fixed.
(rumours: the A1 is not really fixed)
anonymous: The A1 still has a bug, look!
A1 lawyer: We've known this all along, we'll fix it. We're sure it has
nothing to do with A, though.
P supporter: How can you know this, when P's engineers says it's in
A?
A1 supporter: A1 engineers says it has nothing to do with A, it must
be so!
P supporter: Ok, but P engineers say it does, and after all
they already fixed it, unlike the A1 engineers. Isn't this worth
considering?
A1 supporter: What, you don't trust A1 engineers!?
Articia: Still DMA-problems? : Comment 46 of 122ANN.lu
Posted by Alkis Tsapanidis on 20-May-2003 17:10 GMT
In reply to Comment 44 (Leki):
You disagree about what? That Buck didn't say what you wrote? It's a fact, not
something to disagree about.
That we can't be sure that the NBGE fixes everything, yet? That's not something
we can disagree, we can't possibly know, till it has matured.
Articia: Still DMA-problems? : Comment 47 of 122ANN.lu
Posted by Olegil on 20-May-2003 17:53 GMT
In reply to Comment 41 (Alkis Tsapanidis):
Oh come on. That was funny.
Articia: Still DMA-problems? : Comment 48 of 122ANN.lu
Posted by Olegil on 20-May-2003 17:54 GMT
In reply to Comment 45 (Johan Rönnblom):
Wow, now I'm confused ;-)
Articia: Still DMA-problems? : Comment 49 of 122ANN.lu
Posted by Jürgen Schober on 20-May-2003 19:17 GMT
In reply to Comment 40 (Leki):
> and G3 is faster than G4.

But my G3 _is_ faster than G4 ... at 933MHz in the G3/XE =:->
Articia: Still DMA-problems? : Comment 50 of 122ANN.lu
Posted by Mayonaise on 20-May-2003 20:12 GMT
In reply to Comment 42 (Frodon):
"Or if the AmigaOne indeed use a recent version, then why is there still the bug if it's this VIA bug?"

Bugs exist in alot of the VIA chipsets including the latest ones :(, VIA will only admit theres a problem when heavy duty evidence is presented to them as happened with the 686B(VIA are meant to have released a datasheet detailing the problem and how to solve it in firmware, but I've yet to find it). Bugs aren't the only thing wrong with VIA; The 686B and a few others(at least) have really serious performance issues also i.e. burst access in VIA chipsets is done in sections rather then in one long burst as chipsets from SiS , ALi and intel do.

The positives of VIA chipsets seem to be that they're cheap and offer heavy integration...
Anonymous, there are 122 items in your selection [1 - 50] [51 - 100] [101 - 122]
Back to Top