| [Forum] Try only to realise the truth | ANN.lu |
|
| Try only to realise the truth : Comment 251 of 335 | ANN.lu |
| Posted by Sammy Nordström on 06-Jul-2004 14:12 GMT | In reply to Comment 247 (Bernie Meyer):
>Let's put it this way --- unless and until some independent third party with a
>very technical background and a firm grasp of the issues involved has
>performed a range of hand-tailored stress tests, any claims of "no problem"
>should certainly be treated with caution.
No, let's put it this way --- unless and until some independent third party with a very technical background and a firm grasp of the issues involved has performed a range of hand-tailored stress tests, any claims of "hardware fault" should certainly be treated with caution.
I mean, it doesn't make logical sense to assume that something exists until proven otherwise. That's what believers do, not computer scientists. |
|
| Try only to realise the truth : Comment 252 of 335 | ANN.lu |
| Posted by Anonymous on 06-Jul-2004 14:16 GMT | In reply to Comment 251 (Sammy Nordström):
Teasing mode:
Sammy: soon you will pretend that the earth is flat, and that Galileo was wrong!
Bye |
|
| Try only to realise the truth : Comment 253 of 335 | ANN.lu |
| Posted by Johan Rönnblom on 06-Jul-2004 14:23 GMT | In reply to Comment 238 (Ferry):
Ferry wrote:
> [..] in scientific method the charge of the proof is *always* up to
> the person who states something [..]
Indeed. My proof that the A1/Articia S is buggy is that I can
demonstrate that every available DMA implementation results in bugs.
Now, why don't you ask the people who strongly assert that DMA does
work properly in OS4 for some proof of *their* theory? |
|
| Try only to realise the truth : Comment 254 of 335 | ANN.lu |
| Posted by Johan Rönnblom on 06-Jul-2004 14:28 GMT | In reply to Comment 248 (Bernie Meyer):
Bernie: Yes, it will be interesting to stress-test the supposedly
working OS4 DMA drivers with what you just mentioned in mind. |
|
| Try only to realise the truth : Comment 255 of 335 | ANN.lu |
| Posted by Johan Rönnblom on 06-Jul-2004 14:31 GMT | In reply to Comment 251 (Sammy Nordström):
Sammy wrote:
> I mean, it doesn't make logical sense to assume that something
> exists until proven otherwise.
Certainly so! So let's all agree then that until otherwise proven,
it's best to assume that the AmigaOne does not have working DMA. |
|
| Try only to realise the truth : Comment 256 of 335 | ANN.lu |
| Posted by Sammy Nordström on 06-Jul-2004 14:34 GMT | In reply to Comment 252 (Anonymous):
That's funny. I feel a bit like Galileo when he was standing in front of the Inquisition in Rome. :-P |
|
| Try only to realise the truth : Comment 257 of 335 | ANN.lu |
| Posted by Fabio Alemagna on 06-Jul-2004 14:34 GMT | In reply to Comment 248 (Bernie Meyer):
Given those considerations, it should be not difficult to build a test and check whether there are those theorized issues or whether they've been "fixed" with speed penalties, which would indicate a bounce buffer. |
|
| Try only to realise the truth : Comment 258 of 335 | ANN.lu |
| Posted by Fabio Alemagna on 06-Jul-2004 14:38 GMT | In reply to Comment 257 (Fabio Alemagna):
Also, disassembling the Cache#? functions would give a pretty good idea of what is going on behind the scenes. If it's not the Cache#? functions that are being used, whatever they're using will have to be made public, unless they'll require 3rd party developers to sign NDA's and whatnot to develop drivers, which would be the most absurd thing in the Amiga history ever. |
|
| Try only to realise the truth : Comment 259 of 335 | ANN.lu |
| Posted by Amon_Re on 06-Jul-2004 14:43 GMT | In reply to Comment 250 (Emeric SH):
I never denied the possibility, however, i refuse to take someone's word for it, that's my stance.
And right now i don't give a flying fuck anymore, most of the people here are just here to cause flamewars, i'm trough with it, these people can all drop dead for all i care. |
|
| Try only to realise the truth : Comment 260 of 335 | ANN.lu |
| Posted by Sammy Nordström on 06-Jul-2004 14:45 GMT | In reply to Comment 255 (Johan Rönnblom):
You're right in the sense that this feature has yet to be enabled by the software. You know, these word-games is not making this discussion any more constructive. |
|
| Try only to realise the truth : Comment 261 of 335 | ANN.lu |
| Posted by Gunne Steen on 06-Jul-2004 14:46 GMT | In reply to Comment 246 (Anonymous):
Good afternoon Mr anonymous :-)
You can boot from a LS 120 floppy if you want. With LS 120 you can also use 120MB floppys :-)
Gunne |
|
| Try only to realise the truth : Comment 262 of 335 | ANN.lu |
| Posted by Fabio Alemagna on 06-Jul-2004 14:48 GMT | In reply to Comment 259 (Amon_Re):
> I never denied the possibility, however, i refuse to take someone's word for it,
> that's my stance.
But you are taking someone's word for it, it's just that you're taking the word of someone who denies the problem exists, rather then the ones who confirm it does exist. |
|
| Try only to realise the truth : Comment 263 of 335 | ANN.lu |
| Posted by Fabio Alemagna on 06-Jul-2004 14:50 GMT | In reply to Comment 260 (Sammy Nordström):
> You're right in the sense that this feature has yet to be enabled by the
> software.
It's not a feature, it's a lack of a feature, it's lack of cache coherency. Since when having to do by hand - and theoretically with poorer results - what you can be done automatically - and with optimum performances - by the system is a feature? It's perhaps time to say "Only Amiga makes it possible"? :-) |
|
| Try only to realise the truth : Comment 264 of 335 | ANN.lu |
| Posted by Sammy Nordström on 06-Jul-2004 15:02 GMT | In reply to Comment 262 (Fabio Alemagna):
>But you are taking someone's word for it, it's just that you're taking the
>word of someone who denies the problem exists, rather then the ones who
>confirm it does exist.
No, you don't have to take anyone's word in order to assume that a product will function as advertised. When I look at products in a store that I consider buying, I simly assume that they work without someone's word for it. I mean, it wouldn't be very rational to assume that a product does not function as advertised until otherwise has been proven, now would it? We have consumer laws and product warranties for a reason, you know. |
|
| Try only to realise the truth : Comment 265 of 335 | ANN.lu |
| Posted by Sammy Nordström on 06-Jul-2004 15:05 GMT | In reply to Comment 263 (Fabio Alemagna):
Even if the DMA is lacking a feature, the AmigaOne is not lacking a DMA and the DMA is all I was talking about. Unlike some other people, I refrain myself from commenting on the technical details until this whole mess has been sorted out properly. |
|
| Try only to realise the truth : Comment 266 of 335 | ANN.lu |
| Posted by Darth_X on 06-Jul-2004 15:09 GMT | In reply to Comment 261 (Gunne Steen):
shame the LS 120 is discontinued product. :/ |
|
| Try only to realise the truth : Comment 267 of 335 | ANN.lu |
| Posted by Amon_Re on 06-Jul-2004 15:17 GMT | In reply to Comment 262 (Fabio Alemagna):
Wrong Fabio, i don't take somebody's word on it, i find it hard to believe a company would bring out a chip with the faults attributed to it, and that the same company (who designed the bloody thing in the first place) would be unable to fix their own design.
That is the root of my stance, it just doesn't add up, plain & simple. |
|
| Try only to realise the truth : Comment 268 of 335 | ANN.lu |
| Posted by Amon_Re on 06-Jul-2004 15:18 GMT | In reply to Comment 261 (Gunne Steen):
Those still exist? I tought they were extinct by now ;) |
|
| Try only to realise the truth : Comment 269 of 335 | ANN.lu |
| Posted by Fabio Alemagna on 06-Jul-2004 15:21 GMT | In reply to Comment 264 (Sammy Nordström):
> No, you don't have to take anyone's word in order to assume that a product
> will function as advertised.
Dear Sammy, when I, 3rd party, get confronted with 2 other parties which say one the opposite of the other, and I decide to believe one rather than the other one, I am taking someone's word.
But don't mind replying, I know damn too well that you'll keep babbling, I'm sayin this just for the record.
> When I look at products in a store that I consider buying, I simly assume
> that they work without someone's word for it.
You are taking productor's word for it. It puzzles me you fail to undersatand so simple logic.
When you are at a cross, you can do eithr one of the following things: take one of the new paths, or go back from where you come. You've chosed to take a new path. You could have as well said "f#?t it, I don't care, keep fighting about who's right and who's wrong, I'll simply move elsewhere". That would have been the equivalend of going back from where you came.
> I mean, it wouldn't be very rational to assume that a product does not
> function as advertised until otherwise has been proven, now would it? We have
> consumer laws and product warranties for a reason, you know.
Infact the rational thing to do, at first, is to do take productor's word for it. However, when others bring some _undisputable_ facts that put the product in a bad shape, and you decide to dismiss them, then you are taking, again, productor's word for it. Of course you're free to chose, at that point, which one's word to take for it, but others are equally entitled to take questioner's word for it as you are to take productor's word for it. At that point, at the cross, all paths are equally valuable.
if you question the word "undisputable", feel free to prove those facts are disputable - after all, you yourself said that there's no reason to believe something is wrong (in this case, there's no reason to believe the fact is disputable) unless someone proves it is (in this case, unless someone proves it is disputable).
It works both ways, Sammy :-) |
|
| Try only to realise the truth : Comment 270 of 335 | ANN.lu |
| Posted by Johan Rönnblom on 06-Jul-2004 15:23 GMT | In reply to Comment 264 (Sammy Nordström):
Sammy wrote:
> No, you don't have to take anyone's word in order to assume that a
> product will function as advertised.
Then where did those words in the advertising come from, then, if not
from "anyone"?
There's every reason to assume that the A1 does *not* work as
advertised, because there's nothing available to support the claim
that it works except - words. |
|
| Try only to realise the truth : Comment 271 of 335 | ANN.lu |
| Posted by Fabio Alemagna on 06-Jul-2004 15:26 GMT | In reply to Comment 265 (Sammy Nordström):
> Even if the DMA is lacking a feature, the AmigaOne is not lacking a DMA and the
> DMA is all I was talking about.
Then you've been talking about the wrong thing all along, Sammy.
No one ever said the AmigaONE doesn't have DMA, what's been said is that DMA is not working properly, meaning that the whole system does not work properly when DMA is enabled, because the NB is lacking an essential feature which all modern drivers for desktop operating systems expect. It can be worked around in SW but, as Bernie explained eloquently, the only way to work it around in SW and</b< avoid all race conditions is to use a bounce buffer, in which case I'm sure even you, who claim to be a programmer, will acknwoledge that the whole purpose of DMA is defeated. |
|
| Try only to realise the truth : Comment 272 of 335 | ANN.lu |
| Posted by Fabio Alemagna on 06-Jul-2004 15:31 GMT | In reply to Comment 267 (Amon_Re):
> Wrong Fabio, i don't take somebody's word on it, i find it hard to believe a
> company would bring out a chip with the faults attributed to it, and that the
> same company (who designed the bloody thing in the first place) would be unable
> to fix their own design.
Thus, you're taking productor's word for it. You'll admit that your refusal to accept or even try to understand the word of people who
1) know a lot more than you about programming and hw design
2) are not directly involved in this mess, from neither side
3) did tests and showed results
on the basis of your assumption that no one would put on the marked a broken product (as if that had never happened before... where are you living?) is quite laughable, to say the least! |
|
| Try only to realise the truth : Comment 273 of 335 | ANN.lu |
| Posted by Sammy Nordström on 06-Jul-2004 15:54 GMT | In reply to Comment 269 (Fabio Alemagna):
No. When I enter a store and buy a CD-Player, I don't ask for anyone's word that it will function as advertised, I *assume* that it will function as advertised. If two people walks up to me in that very same store, one that claims that it won't work and one that claims that it will, unless any of them has anything factual to back it up with, I will ignore *both* and go back to my initial assumption that it will function as advertised. Since my assumption is NOT based on anyone's words, I'm not taking anyone's word. Simple, isn't it?
The same goes for these DMA problem claims; not believing in these claims does NOT equal to taking the words of those claiming that there is no DMA problems. From my point of view, I've made an assumption that the product functions as advertised long before anyone had ever heard of these claims of problems with the DMA and for as long as noone is able to actually prove those claims, my assumption remains. The claim that the DMA does function is just a well expected response that has had no impact what so ever on my initial assumption. The fact that it does support my initial assumption does not change the fact that the assumption was not based on it. |
|
| Try only to realise the truth : Comment 274 of 335 | ANN.lu |
| Posted by Fabio Alemagna on 06-Jul-2004 16:02 GMT | In reply to Comment 273 (Sammy Nordström):
Sigh. Didn't I say you'd keep babbling? ;-) |
|
| Try only to realise the truth : Comment 275 of 335 | ANN.lu |
| Posted by Sammy Nordström on 06-Jul-2004 16:03 GMT | In reply to Comment 271 (Fabio Alemagna):
>Then you've been talking about the wrong thing all along, Sammy.
No, you have. You jumped into a discussion I had with Johan Rönnblom and I recommend you read what I replied to (comment #255) before making such claims. |
|
| Try only to realise the truth : Comment 276 of 335 | ANN.lu |
| Posted by Sammy Nordström on 06-Jul-2004 16:07 GMT | In reply to Comment 274 (Fabio Alemagna):
You jump into other people's discussions without reading what the discussion is about, you say that I'm babbling without addressing the issues I raise in my posts... Congratulations, you've just been added to my ignore list. |
|
| Try only to realise the truth : Comment 277 of 335 | ANN.lu |
| Posted by Fabio Alemagna on 06-Jul-2004 16:08 GMT | In reply to Comment 275 (Sammy Nordström):
> No, you have. You jumped into a discussion I had with Johan Rönnblom and I
> recommend you read what I replied to (comment #255) before making such claims.
Johan never said the AmigaONE has no DMA, as he will confirm to you. Perhaps it's time for you to take a rest from computing, Sammy, |
|
| Try only to realise the truth : Comment 278 of 335 | ANN.lu |
| Posted by Sammy Nordström on 06-Jul-2004 16:10 GMT | In reply to Comment 277 (Fabio Alemagna):
>Johan never said the AmigaONE has no DMA
Neither did I. |
|
| Try only to realise the truth : Comment 279 of 335 | ANN.lu |
| Posted by Fabio Alemagna on 06-Jul-2004 16:11 GMT | In reply to Comment 275 (Sammy Nordström):
Btw, from the comment you mentioned, Johan wrote (emphasis is mine):
> it's best to assume that the AmigaOne does not have working DMA.
So what now, Sammy? Will you run away, or admit you've barked at the wrong tree all along?
Moreover, since you yourself said that you weren't arguing whether it worked well or not, but rather if it existed at all (claiming others said it doesn't even exist), do you acknowledge that it exists, but doesn't work as it should? |
|
| Try only to realise the truth : Comment 280 of 335 | ANN.lu |
| Posted by Gunne Steen on 06-Jul-2004 16:14 GMT | In reply to Comment 268 (Amon_Re):
Hi Amon_re & Darth_X
Might still be possible to get your hands on if you want .-)
Gunne |
|
| Try only to realise the truth : Comment 281 of 335 | ANN.lu |
| Posted by Amon_Re on 06-Jul-2004 16:14 GMT | In reply to Comment 272 (Fabio Alemagna):
Thus, you're taking productor's word for it.
No, i take the explication that's the most plausable in my eyes.
You'll admit that your refusal to accept or even try to understand the word of people who
1) know a lot more than you about programming and hw design
Proof to me right here right now that i refuse to even try to understand people's explication, i dare you, right here right now, if you fail, go crawl under a rock.
People who offer a good explication as to why they think it's broken or not have had good conversations with me on this matter in the past, and they still can in the future, the people who i dismiss alltogether are people like Johan what's his name. There's a difference in repeating what other people tell you, and understanding what they tell you, you of all people should realise that with all your wordgames.
If you fail to prove your point above, this conversation is over.
2) are not directly involved in this mess, from neither side
3) did tests and showed results
They showed tests, or said it was broken, they never offered a valid technical explication as to why, except that "dma is broken", guess what, that's not enough for me, because as we know now, it IS possible to use it with DMA, however, it doesn't work as the average northbridge in other desktop computers, and that it even might be a crappy design, crappy != broken.
on the basis of your assumption that no one would put on the marked a broken product (as if that had never happened before... where are you living?) is quite laughable, to say the least!
Spare me your morale, because my point wasn't that they released it in a inperfect shape, but that they would have been unable to solve/fix the problem, while someone from outside the company, who has no knowledge of the chip's internals would be. That would eighter make the guy a genious (and he'd be wasting his talents in the wrong market) or the other guys (Mai) complete idiots, also unlikely.
Now, stick to my points & don't try to deflect or twist my words, or this conversation is over |
|
| Try only to realise the truth : Comment 282 of 335 | ANN.lu |
| Posted by Amon_Re on 06-Jul-2004 16:16 GMT | In reply to Comment 280 (Gunne Steen):
Nooo! Keep them away from me!!! :P |
|
| Try only to realise the truth : Comment 283 of 335 | ANN.lu |
| Posted by Fabio Alemagna on 06-Jul-2004 16:18 GMT | In reply to Comment 278 (Sammy Nordström):
> >Johan never said the AmigaONE has no DMA
> Neither did I.
Ehum... so what? Did I say you said that the AmigaONE has no DMA? :-o
Sammy, are you aiming at the ignobel prize, by any chance?
What you said, is this:
> Even if the DMA is lacking a feature, the AmigaOne is not lacking a DMA and
> the DMA is all I was talking about.
Thus, you:
1) acknowledge that DMA is lacking a feature
2) say the AmigaONE is not lacking DMA
3) and say that you were only talking about DMA (I assume here you mean you were talking about the mere existence of DMA, since everyone here was talking about the DMA topic).
Thus, 1) means that you agree with everyone that DMA on the AmigaONE is not working properly (lacks a feature, you say, a feature that is practically mandatory, experts say), 2) confirm that you were arguing about the wrong thing, because all that people have argued about is whether DMA works or not, not whether it exists. |
|
| Try only to realise the truth : Comment 284 of 335 | ANN.lu |
| Posted by Sammy Nordström on 06-Jul-2004 16:45 GMT | In reply to Comment 279 (Fabio Alemagna):
Fabio, Fabio, Fabio...
Let's take this one step at a time.
1. Johan claims that the AmigaOne has no working DMA.
2. I say that this feature (ie the DMA) has yet to be enabled by the software.
3. You say "it's not a feature, it's lacking a feature".
You see, in the discussion I had with Johan, I refered to the DMA as a feature of the AmigaOne. Then you jumped into the discussion and tried to redefine feature into being about features of the DMA instead, which explains why I made the "the AmigaOne does not lack DMA" comment. I was annoyed by your attempt to steer the discussion into something else and decided to not let you do that.
Furthermore, your inability to comprehend what is said in these forums is really starting to annoy me. Please stop responding to my posts because I don't want to discuss anything with you anymore. |
|
| Try only to realise the truth : Comment 285 of 335 | ANN.lu |
| Posted by Sammy Nordström on 06-Jul-2004 16:56 GMT | In reply to Comment 283 (Fabio Alemagna):
>1) acknowledge that DMA is lacking a feature
No. You as a programmer should be able to understand the concept of the word "if".
>2) say the AmigaONE is not lacking DMA
Exactly.
>3) and say that you were only talking about DMA (I assume here you mean you
>were talking about the mere existence of DMA, since everyone here was talking
>about the DMA topic).
My god... Grasp this: You said that the DMA was not a feature but rather lacks a feature. I tried to explain to you that I don't care about the technical details behind the claims that the AmigaOne would have a malfunctioning DMA and that I was refering to the DMA rather than a feature of the DMA when I said that the software has yet to enable this feature.
Now please go away. |
|
| Try only to realise the truth : Comment 286 of 335 | ANN.lu |
| Posted by Amon_Re on 06-Jul-2004 17:06 GMT | In reply to Comment 285 (Sammy Nordström):
It's not even a malfunction, it's an inheritance from the embedded platform, BenH said that it would make more sence in the embedded market with the 4XX etc series of PPC's, but that it isn't the best implementation (according to him, and he'd know better then me) to run linux on an G3/G4.
This isn't a malfunction, nor does it imply it's broken, rather that it's just different from the rest out there. |
|
| Try only to realise the truth : Comment 287 of 335 | ANN.lu |
| Posted by Johan Rönnblom on 06-Jul-2004 17:14 GMT | In reply to Comment 285 (Sammy Nordström):
Sammy wrote:
> I tried to explain to you that I don't care about the technical
> details behind the claims that the AmigaOne would have a
> malfunctioning DMA and that I was refering to the DMA rather than a
> feature of the DMA when I said that the software has yet to enable
> this feature.
Ok, so you mean that as long as it's DMA then you're happy? The
"details" in question are of course whether
a) the DMA speeds up the transfer
b) the DMA frees CPU resources
c) the DMA delivers error-free data.
But I guess, you do not care for such "details", right? |
|
| Try only to realise the truth : Comment 288 of 335 | ANN.lu |
| Posted by Sammy Nordström on 06-Jul-2004 17:33 GMT | In reply to Comment 286 (Amon_Re):
Nor did I claim that it would be. I've even repeatedly said several times in this thread that I assume that the product will functions as advertised until proven otherwise. The claims that there would be a malfunction in the DMA are not mine and I will keep ignoring such claim until someone manages to actually prove it. |
|
| Try only to realise the truth : Comment 289 of 335 | ANN.lu |
| Posted by Don Cox on 06-Jul-2004 17:39 GMT | In reply to Comment 264 (Sammy Nordström):
"No, you don't have to take anyone's word in order to assume that a product will function as advertised. When I look at products in a store that I consider buying, I simply assume that they work without someone's word for it."
Do you? I don't. Experience teaches otherwise, especially with anything digital. |
|
| Try only to realise the truth : Comment 290 of 335 | ANN.lu |
| Posted by Don Cox on 06-Jul-2004 17:45 GMT | In reply to Comment 247 (Bernie Meyer):
"So when reading, not only do you need to invalidate all caches covering the area *after* the DMA, but also *before* you initiate the DMA. And strictly speaking, you need to make sure that they stay invalidated during the whole period of the transfer (which is where fine-grained MMU tables for non-cachable areas come in handy --- although they typically work on page-sized memory areas, which is *not* the granularity with which you want to disable caching in your data segment. "
What is the granularity in a PPC ? |
|
| Try only to realise the truth : Comment 291 of 335 | ANN.lu |
| Posted by Johan Rönnblom on 06-Jul-2004 18:12 GMT | In reply to Comment 290 (Don Cox):
Don Cox wrote:
> What's the granularity in a PPC?
The pages are for every 4kb, which is why you don't want to mess with
the settings for these too often. The data for this is stored in
normal ram of course.
Then there are Batch Array Tables, BATs, which is what Ben H talked
about in the quote above. These allow to set up MMU/cache options for
a range between 128KB and 256MB.
I don't know what would be the least costly way to solve this problem
on the PPC, but it's quite obvious that the penalty is rather heavy -
although I guess it might be possible to achieve better results than
by not using DMA at all. Maybe we'll see. |
|
| Try only to realise the truth : Comment 292 of 335 | ANN.lu |
| Posted by Sammy Nordström on 06-Jul-2004 19:30 GMT | In reply to Comment 289 (Don Cox):
It's not like I've never experienced recieving malfunctioning products. You see, despite my assumption that the product will function as advertised, I'm fully aware of the risk that it won't. I mean, there is no such thing as a way to be 100% sure that the product will function as advertised, hence the assumption. Whenever I happen to end up in a situation of having bought a product that is not functioning as advertised, I make use of my legal rights as a consumer and return the product to the distributor/seller. It's as simple as that, really. |
|
| Try only to realise the truth : Comment 293 of 335 | ANN.lu |
| Posted by Johan Rönnblom on 06-Jul-2004 20:06 GMT | In reply to Comment 292 (Sammy Nordström):
It's not that simple for two reasons. First, the warranty may only be
valid for one year in some countries, in Sweden it's always two years
for electronics though. But even if it is two years, you have to
complain as soon as the fault appears. As in this case, the fault has
been there from the start, the seller could (and most likely would)
argue that since you didn't complain when you got the product, you've
lost your right.
The second reason is that the seller may want to claim either that the
fault is not vital (say, because you can disable DMA or caches) or
that you should have known that this fault existed before you bought
the product (as in this case, we're discussing this fault everywhere
repeatedly, after all).
For both of these reasons, you're not exactly better off having closed
your ears humming "lalala" when people tried to warn you. And in any
case - unlike you, I think there are people who would rather not have
to spend the money and energy to do a warranty claim. |
|
| Try only to realise the truth : Comment 294 of 335 | ANN.lu |
| Posted by 3seas on 06-Jul-2004 22:16 GMT | | just trying to find out what this link leads to |
|
| Try only to realise the truth : Comment 295 of 335 | ANN.lu |
| Posted by syrtran on 06-Jul-2004 23:57 GMT | In reply to Comment 226 (Chris Hodges):
@Chris
(quote)
As the hardware DMA support is assumed for most systems, hardly anybody will care for such manual cache consistency in their drivers. So although this might be possible to fix for Linux, I don't think every driver will be checked for necessary CachePreDMA and CachePostDMA calls.
Even if one would not call the ArticiaS bugged related to DMA, it is clearly a missing feature for transparent DMA operations.
(/quote)
I agree that the missing 'dirty' signal is a design flaw, however there is one more term in this equation. BenH stated that he used only the BAT registers to map the entire LinuxPPC address space and completely ignored the page tables. IMO, this is a bad software design. I'm not the one, however, that's done kernel programming, so take my opinion for the little it's worth.
In order to invalidate a cache entry, a driver designer would have to invalidate a huge block of memory addresses (up to 256MB). This would probably be outside the addressing for a single device's I/O buffers and could seriously hamper performance across the board.
I've read Moto's documentation and an old Que PowerPC book I have. I get the feeling that the BAT registers were in the Power architecture solely to accomodate OS/400's design as one large database. Moto's own documentation states:
[Moto]
The block address translation (BAT) mechanism in the OEA provides a way to map ranges of effective addresses larger than a single page into contiguous areas of physical memory. Such areas can be used for data that is not subject to normal virtual memory handling (paging), such as a memory-mapped display buffer or an extremely large array of numbers.
[/Moto]
Seems to me that, if someone wants DMA-enabled Linux on an A1, all they've got to do is rewrite the memory-management routines using page tables and throw out all the work done by BenH, an acknowledged kernel guru. |
|
| Try only to realise the truth : Comment 296 of 335 | ANN.lu |
| Posted by Anonymous on 07-Jul-2004 01:42 GMT | In reply to Comment 295 (syrtran):
"BenH stated that he used only the BAT registers to map the entire LinuxPPC address space and completely ignored the page tables. IMO, this is a bad software design."
Of course it is a bad software design because it doesnt work in Amiga!!! Oh btw you can't have a dual CPU card in AmigaOne now... |
|
| Try only to realise the truth : Comment 297 of 335 | ANN.lu |
| Posted by Anonymous on 07-Jul-2004 02:59 GMT | In reply to Comment 283 (Fabio Alemagna):
Fabio, if you're such an expert on DMA, why is it that AROS still (after what, ten years and a bounty?) STILL doesn't have a DMA-capable IDE driver? could it be that you're incapable of writing one, and all you're good at is writing 'hello world' and BSing in forums? |
|
| Try only to realise the truth : Comment 298 of 335 | ANN.lu |
| Posted by hammer on 07-Jul-2004 08:33 GMT | In reply to Comment 223 (Johan Rönnblom):
With Pentium Pro(P6), it has microcode updates. But there's a limit to how P6 microcode updates can fix certain "issues" e.g. early Pentium III @~1.1Ghz i.e. issues with recompiling Linux kernels. |
|
| Try only to realise the truth : Comment 299 of 335 | ANN.lu |
| Posted by hammer on 07-Jul-2004 08:39 GMT | In reply to Comment 230 (Ferry):
I don’t mind a near free Pegy 1 to be sent to me. ;) |
|
| Try only to realise the truth : Comment 300 of 335 | ANN.lu |
| Posted by Fabio Alemagna on 07-Jul-2004 09:00 GMT | In reply to Comment 297 (Anonymous):
> Fabio, if you're such an expert on DMA, why is it that AROS still (after what,
> ten years and a bounty?) STILL doesn't have a DMA-capable IDE driver?
Because no one ever felt the need to write one until now? The driver is going to be DMA capable in the next few days.
> could it be that you're incapable of writing one, and all you're good at is
> writing hello world' and BSing in forums?
No, it couldn't simply because knowing how DMA works and what it's supposed to do, doesn't automatically make anyone an HW bashing programmer. I'm more of an high level programmer, there are other people who focus on the HW part of AROS.
Thanks for the cheap shot :-) |
|
|
- User Menu
-
- About ANN archives
- The ANN archives is powered by #AmigaZeux. It was updated daily (news last: 22-Oct-2004; comments last: 18-May-2005).
ANN.lu was created, previously owned and maintained by Christian Kemp, www.ckemp.com.
- Contribute
- Not possible at this time!
- Search ANN archives
- Advanced search
- Hosting
- ANN.lu was hosted by Dreamhost. Sign up through this link, mention "ckemp" as referrer and he will get a 10% commission on any account you purchase.
Please show your appreciation for any past, present and future work on ANN.lu by making a contribution via PayPal.
|
