[Events] Notes about my presentation at the Alchimie 4 show | ANN.lu |
Posted on 29-Sep-2004 10:40 GMT by Stéphane Guillard | 427 comments View flat View list |
The html pages I made available was only a starter slideshow for my presentation.
Here are a few words about what I shown, and what I said.
Hi Gentlemen,
The html pages which are available at my site are just an introduction slideshow that I presented as a starter, it does not relate what I shown and said afterwards. Please find below a couple of notes about my presentation at Alchimie show.
Here is a quick summary of the presentation I did :
- I booted with boot delays set to 1 second for UBoot and for SLB (second level booter), thus less than 20 seconds after power on, we were with a fully loaded Workbench with Amidock.
- I started with the small html slideshow, presented on IBrowse. You can find those pages at the URL above. IBrowse loaded in 2 seconds with its About: page fully displayed. Browsing through the pages of the slideshow was instantaneous.
- While we were at it, I browsed the OS4 install guide, also as fast as can be, must say I also find this responsiveness impressive myself :)
Then I demoed as many things as I had time to during the 2 hours I had. Everything worked, fast & stable, and was smooth and impressed. I showed mainly:
- All os4 system, tools, utilities, prefs & stuff
- MUI, IBrowse, Amitradecenter
- Yam, SimpleMail
- ClassAction (M. Elsner's file manager)
- MakeCD
- MooVid running a DivX
- DVPlayer running an mpeg2
- AudioEvolution 4 with the demo project, cursors auto moving smoothly, the playback was smooth also, with mostly no CPU usage.
- AmiPDF with the AE4 manual and another heavy PDF file, very fast
- USB. I plugged a Microsoft IntelliMouse Optical into my USB hub, and we had 2 mice to control the system
- Crisot's slach5 winning demo (got applauses which it deserved)
- chip's rayrace realtime raytracer demo. When the juggler appeared, audience was impressed, but really took measure of what they were watching when I moved the mouse. Wait for the Altivec version !
- FPSE, AmiDog's PS1 emulator, with an oldskool game which ran nicely ('Cotton')
- some other visual toys I had at hand
The demos only grimreaped twice, and I anticipated the grims before they popped up :
- One was native glsokoban / w3d, when I launch it does a base page access (a “null pointer” bug in glsokoban). I didactically shown the disassembly which is available in the grimreaper window, was a store to r4, r4 was null etc. I clicked on continue, and it all went fine & fast.
- One was frying pan 0.3.1, I shown the app, and at one point I said 'now it should grim’ and it did. It still loaded fine though. I quit the app, clicked on reboot and less than 4 secs after, wb was up with amidock. That was the only reboot of the show.
I forgot to show (because of short time):
- Petunia... Almos, sorry, I had prepared something for that (side by side windowed jit & nonjit runs of voxelspace), but i both forgot, and was asked to stop at this point by the party organizers cause it was already 5:30 pm while I was scheduled until 4pm.
- ArtEffect
- USB with MassStorage (ie USB key or digital camera)
At the end, I had many fair questions to which I answered; my feeling is that the audience really appreciated the effort behind what I shown, and was conscious that we are not far away from a releasable 4.0.
Then came the expected question, 'and why doesn’t DMA work ?'
I said 'All what you saw was DMA since the 1st boot'.
I copied a few 100 MB files in a snap, with zero CPU (thanks to Pete Gordon for the clock/CPU docky, helps a lot). Then I switched to PIO, they saw 4 x slower and 80% CPU.
The audience understood that it was indeed DMA, and that was fast, and that was part of the overall smoothness of what I shown.
Then I explained the things below (this is the reference for my statements, please don’t rephrase or extrapolate or invent or whatever):
- IDE UDMA works on VIA and Articia on AmigaOne SE / XE / µA1 MK2 (as I demoed) ...
- ... except when the Ethernet chip goes online and is used.
- the Ethernet chip only triggers the problem, but it is not at all related to it (a test using a PCI Ethernet shows the same behaviour)
- We have made a driver for a Silicon Image 680 PCI IDE UDMA133 controller chip, this does UDMA 133 nicely, including when Ethernet is used at full speed.
- The fact that a PCI IDE controller solution works, shows that the problem is *not* related to Articia, since PCI DMA is *also* handled by the Articia, and that works.
- The full Alchimie show demo was done using UDMA, both from the VIA and from the Si680, without problem (but with Ethernet off, would the Ethernet have been turned online, I would have had to revert the VIA into PIO before).
- Things are currently under more investigation
In the meantime there are 2 options for existing A1 board owners:
- Use the VIA IDE controller in PIO mode when using Ethernet, and UDMA at other times,
- Purchase a faster (UDMA133) Silicon Image 0680 IDE PCI card (from around $20). This is my personally recommended option as the delivered speed is noticeable faster than the on-board VIA controller in UDMA mode.
|
|
Notes about my presentation at the Alchimie 4 show : Comment 301 of 427 | ANN.lu |
Posted by Fabio Alemagna on 30-Sep-2004 18:58 GMT | In reply to Comment 299 (Michal Schulz): Michal, of course the solution is to have the buffer aligned to the cacheline sizes? |
|
Notes about my presentation at the Alchimie 4 show : Comment 302 of 427 | ANN.lu |
Posted by Fabio Alemagna on 30-Sep-2004 18:59 GMT | In reply to Comment 301 (Fabio Alemagna): And of course buffer's size should be a multiple of the cacheline.
However, as Georg pointed out, this issue doesn't seem to be taken care of in all drivers. Certainly not in the AHI one. |
|
Notes about my presentation at the Alchimie 4 show : Comment 303 of 427 | ANN.lu |
Posted by Davy Wentzler on 30-Sep-2004 19:01 GMT | In reply to Comment 278 (Georg Steger): "Okay, but doesn't caching work on a page basis? And can't different allocations (in OS4) of different apps end up using the same page (also because IIRC of the rater big page size of 64K?) And while porting the GPL SBLive/Audigy AHI driver to a certain other OS a few weeks ago, I noticed that the OS4-specific version of linuxsupport.c/AllocPages() in there does a simple AllocVec(..., MEMF_PUBLIC). "
You forgot the important part there:
AllocVec(size + PAGE_SIZE, MEMF_PUBLIC); |
|
Notes about my presentation at the Alchimie 4 show : Comment 304 of 427 | ANN.lu |
Posted by Christophe Decanini on 30-Sep-2004 19:16 GMT | In reply to Comment 295 (MikeB): "You told your side of the story while I believe you should have refrained, at least if you really didn't want to see me clarify things in here with regard to your "selected facts". "
What I said about you posting anonymously here is right, there is no need for you to find lame excuses or to discuss it forever.
"- One fact is that despite I have never been moderated here on ANN you want to see me banned."
And this didn't happen without reason. It was not just about me. Out of the 3 moderators on this site 2 were sick of you taking over the site for your agenda/propaganda.
It looks that it also happened on other sites with other moderators.
It also look that people dislike you so much that some forums are perpetualy bashing you.
"- Another fact is that you have a personal interest, because of your involvement with a Pegaos dealer, for this platform to succeed. "
First of all it was an Amiga dealership which I own half. We refused to sell the Pegasos until the hardware was fixed and until having a decent version of MorphOS. We had the exact same requirements for the AmigaOne but unfortunately it turned out that AmigaOS4 was not available for a long time (and not working well with linux anyway) and that when it recently was we could not have enough interest from customers to start with an order.
So of course I want the Pegasos / MorphOS to be successfull but I also want the Amigaone / OS4 to be successfull. This is why I still care about these news.
Other than that I had not been interested financialy wise in this business even before we were selling the Pegasos. The reason of this was that the business could not sustain 2 employees anymore.
We have nothing against Hyperion, we are still selling their games (http://www.aps.fr/en/?page=amiga/games). We only have been bitter about too optimistic release dates that had for effect to have our customers wait for the new things and stop buying into the existing things. I also have a problem with B Hermans who took cheap shots at me as anonymous here (how familiar). We had problems with Genesi too with the limited availibility of the Pegasos. I think the main difference was that as soon as we started to sell the Pegasos we have been seen as a "traitor" by some people around AmigaInc and Hyperion.
In the end we lost countless time, effort and money to support this community while being attacked on a regular basis.
To success the Pegasos does not need the Amiga1 to fail. Most of the Amigans that had a need for a new machine already got one or left. With selling the Pegasos we kept them from going away and kept enough activity to survive with providing hardware (including eyetech), software (including hyperion) and services.
"- Another fact is that when I was asked to leave Genesi was involved here on ANN and later even hired its webmaster. "
Your bann had nothing to do with Genesi. As a matter of facts I knew that Christian was hired only several weeks after he was. Christian always wanted to keep ANN independent despite the fact he was hired. Having ANN moderators not knowing about his situation proove he did well in this respect.
"- Last but not least, you cannot stop insulting me and I believe that you would have moderated me if I would have shown the same behaviour against you. "
I don't call what I do insulting. I call it set things straight with you. I can not get any more lies from you.
I know that you have been part of mailing lists with Amiga Inc on image control etc and that Gary Peake lied about me with saying that I was under BB orders or something like this.
I do think that if you and some your friends would not have set up such an Amiga elitism where anyone enlighting a problem has to be excluded the communauty would do much better today.
"I don't understand you guys, it has been known for a very long time that the AmigaOS4/Linux kernel/drivers would have to take into account the way the current ArticiaS chips operates. The AmigaOS4 team has clarified on multiple occasions that this would NOT be an issue with AmigaOS4. "
And this was not clear at all from the start. You and you followers were the first to blame people enlighting such problems every time it happened.
That was just another "there is no problem" until another "it is feature that make it not possible but the next software / hardware release will fix it".
"How you guys could see positive news like this, that future AmigaOne boards won't have this problem with regard to standard unmodified Linux kernel/drivers as being so negative just beats me."
I see it as bad news as most of the people who already bought the hardware expected it to work without the need of adding 2 pci cards (1 more pci slot was suppose to be the A1 advantage over the pegasos).
It also has been said that new model have proper cache snooping so it makes me feel that all the line we have been lied about the articia working properly.
Knowing that we can wonder if we won't have another episode of new feature/bug that will prevent things to work properly and require additional hardware/ software fixes if possible.
You must be the most hypocrite Amiga supporter to see this as good news. I know as a fact that several people supporting A1 / OS4 have seen this news as a bad one at least for the current owners.
If I do like them it must be because I'm evil in your eyes or something. |
|
Notes about my presentation at the Alchimie 4 show : Comment 305 of 427 | ANN.lu |
Posted by Johan Rönnblom on 30-Sep-2004 19:26 GMT | In reply to Comment 304 (Christophe Decanini): This is bad news. It's just "sugared" with statements that there will
be good news later. The good news is when "OS4 with working UDMA is
released!" or "AmigaOne with working bus snooping is available!". |
|
Notes about my presentation at the Alchimie 4 show : Comment 306 of 427 | ANN.lu |
Posted by MikeB on 30-Sep-2004 19:33 GMT | In reply to Comment 304 (Christophe Decanini): > What I said about you posting anonymously here is right, there is no need for
> you to find lame excuses or to discuss it forever.
LOL, speaking of lame excuses, I guess you must be the expert. :-)
So I take over this website? Why? Because you were running out of arguments? LOL
IMO there is a rather constant group of vocal people here who seem to "support" mainly the MorphOS and think their attacks on figures relating to the AmigaOS4 project somehow would furhter their platform of choice. |
|
Notes about my presentation at the Alchimie 4 show : Comment 307 of 427 | ANN.lu |
Posted by Georg Steger on 30-Sep-2004 19:53 GMT | In reply to Comment 303 (Davy Wentzler): @Davy Wentzler:
> You forgot the important part there:
> AllocVec(size + PAGE_SIZE, MEMF_PUBLIC);
Oops, right. Sorry. How could I forget. As I took the same code and just fixed the known missing ("tbd") freeing code (memleak). Like this, BTW:
AllocPages:
address = AllocVec( size + PAGE_SIZE + sizeof(APTR), MEMF_PUBLIC );
if( address != NULL )
{
a = (unsigned long) address;
a = (a + PAGE_SIZE - 1 + sizeof(APTR)) & ~(PAGE_SIZE - 1);
((APTR *)a)[-1] = address;
address = (void *) a;
}
free_page:
if (addr) FreeVec(((APTR *)addr)[-1]); |
|
Notes about my presentation at the Alchimie 4 show : Comment 308 of 427 | ANN.lu |
Posted by Fabio Alemagna on 30-Sep-2004 19:55 GMT | In reply to Comment 302 (Fabio Alemagna): > However, as Georg pointed out, this issue doesn't seem to be taken care of in
> all drivers. Certainly not in the AHI one.
Mm, looking better at the code, it does align the buffer to the page size. However, to work correctly the alocated size must be a mutliple of the cache line, but the code in question doesn't do any check about it, which in itself is not wrong if the requested size is known to be always a multiple of the cache line.
As the comments in the code suggest, though, nothing is done (yet) to assure that the buffer is non cacheable. |
|
Notes about my presentation at the Alchimie 4 show : Comment 309 of 427 | ANN.lu |
Posted by Davy Wentzler on 30-Sep-2004 19:58 GMT | In reply to Comment 307 (Georg Steger): Ah, ok. :) |
|
Notes about my presentation at the Alchimie 4 show : Comment 310 of 427 | ANN.lu |
Posted by MikeB on 30-Sep-2004 19:59 GMT | In reply to Comment 304 (Christophe Decanini): > I know that you have been part of mailing lists with Amiga Inc on image
> control etc
It was an Amiga community advisory mailing list, used by Amiga Inc *and* its partners and several community representatives.
> and that Gary Peake lied about me with saying that I was under BB
> orders or something like this.
Don't remember. I do know the list was meant to be confidential though.
> I do think that if you and some your friends would not have set up such an
> Amiga elitism
It was an effort to see how things could be improved. Of course this is something some people don't wanted to see happening. (Including you?)
> And this was not clear at all from the start. You and you followers were the
> first to blame people enlighting such problems every time it happened.
False, we did not agree with Genesi's "There is no Mai without April" statement within which bbrv stated things like:
"We know what Eyetech pays for the Teron/A1; we know what the A1 distributors pay. We would be willing to match the prices all around less 5% and provide an immediate solution to the community."
"We can compensate Allan Redhouse through his success (something he is clearly about to loose with YDL Teron pricing). Hyperion too. Oh yes, you want to buy a G4 A1? Wake up people! Read the disclaimers on the YDL site. Do you understand what needs to happen before this will? There is a Pegasos G4 now."
"If we pulled eveyone together on this we might even honor those $50 coupons!"
"We will give Hyperion a board. We will support the A1 Betatesters as we have our own."
"Eyetech can certify the Pegasos to meet Amiga specifications -- they are positioned to do this. Hyperion and Eyetech can promote OS4."
I cannot believe you *really* think bbrv behaved constructively here!
> I see it as bad news as most of the people who already bought the hardware
> expected it to work without the need of adding 2 pci cards (1 more pci slot
> was suppose to be the A1 advantage over the pegasos).
You are talking about two seperate issues and you combine this into one. The ArticiaS problem has nothing to do with the conflict being talked about within this news item.
> It also has been said that new model have proper cache snooping so it makes me
> feel that all the line we have been lied about the articia working properly.
The current ArticiaS should work without problems with AmigaOS4! This is what you should have learned by now from reading this news item.
> You must be the most hypocrite Amiga supporter to see this as good news.
Yes, I consider being able to use Linux unmodified with future AmigaOne boards as something positive. No insult from you is going to change this fact.
> I know as a fact that several people supporting A1 / OS4 have seen this news
> as a bad one at least for the current owners.
I haven't. They seem to be dissapointed about the latest issue being discovered, not that others won't have to wait for modified Linux drivers/kernel to take full advantage of the hardware. |
|
Notes about my presentation at the Alchimie 4 show : Comment 311 of 427 | ANN.lu |
Posted by Davy Wentzler on 30-Sep-2004 20:02 GMT | In reply to Comment 308 (Fabio Alemagna): It's 4096-bytes aligned, therefore also 32-byte aligned.
"As the comments in the code suggest, though, nothing is done (yet) to assure that the buffer is non cacheable."
Doesn't need to. If you would have it non-cacheable these things would crawl. The interrupt code takes care of the cache flushing. |
|
Notes about my presentation at the Alchimie 4 show : Comment 312 of 427 | ANN.lu |
Posted by Christophe Decanini on 30-Sep-2004 20:36 GMT | In reply to Comment 306 (MikeB): And there are a lot of OS4 people and AROS people and others and I'am glad that they are here if they don't abuse ANN.
None of them got to the point you were when you where asked to leave. |
|
Notes about my presentation at the Alchimie 4 show : Comment 313 of 427 | ANN.lu |
Posted by Fabio Alemagna on 01-Oct-2004 02:54 GMT | In reply to Comment 310 (MikeB): > It was an Amiga community advisory mailing list, used by Amiga Inc *and* its
> partners and several community representatives.
Let's also say "some dealers", but not all of them. Dealers which had in advance the knowledge of some inside facts and could thus make business decisions on their basis, unlike the majority of others which had to decide whether or not to commit to Amiga inc. on the basis of their own hopes of success without any insight in what was really going on.
This topic has been talked about to death on Italian ML's thanks to the fact that Luca Diana was involved in it. Damage control was the rule, nothing of what was going on for real had to spread around the world, and if it accidentally did, anything possible to cover it up should have been done. That was the rule, and you can be sure I very well know it. |
|
Notes about my presentation at the Alchimie 4 show : Comment 314 of 427 | ANN.lu |
Posted by Fabio Alemagna on 01-Oct-2004 02:56 GMT | In reply to Comment 311 (Davy Wentzler): > It's 4096-bytes aligned, therefore also 32-byte aligned.
I know it's page aligned, what the code assumes is that the size of the allocated memory is a multiple of the cache line too, which may or may not be true, I dunno. |
|
Notes about my presentation at the Alchimie 4 show : Comment 315 of 427 | ANN.lu |
Posted by Fabio Alemagna on 01-Oct-2004 02:56 GMT | In reply to Comment 311 (Davy Wentzler): > It's 4096-bytes aligned, therefore also 32-byte aligned.
I know it's page aligned, what the code assumes is that the size of the allocated memory is a multiple of the cache line too, which may or may not be true, I dunno. |
|
Notes about my presentation at the Alchimie 4 show : Comment 316 of 427 | ANN.lu |
Posted by Michal Schulz on 01-Oct-2004 03:11 GMT | In reply to Comment 301 (Fabio Alemagna): Yes. But in this case that would mean that the buffer passed to the OS4's ATA driver has to be already aligned, which may or may not be true. It's one of the reasons why ata.device in AROS will abandon DMA transfter and will use PIO mode if the buffer is inproperly aligned (but in this case that means the bufer starts at ODD address :))) |
|
Notes about my presentation at the Alchimie 4 show : Comment 317 of 427 | ANN.lu |
Posted by Fabio Alemagna on 01-Oct-2004 03:40 GMT | In reply to Comment 316 (Michal Schulz): > Yes. But in this case that would mean that the buffer passed to the OS4's ATA
> driver has to be already aligned, which may or may not be true.
Yes, that's indeed true.
> It's one of the reasons why ata.device in AROS will abandon DMA transfter and
> will use PIO mode if the buffer is inproperly aligned (but in this case that
> means the bufer starts at ODD address :)))
Probably the solution in the AOS4 case is pretty similar: isolate the biggest aligned subbuffer in the given buffer, then only do DMA for it, whilst do PIO on the rest of the buffer.
Either that, or they're going to use a backbuffer. In both cases, performances will be suboptimal, as it's obvious. |
|
Notes about my presentation at the Alchimie 4 show : Comment 318 of 427 | ANN.lu |
Posted by Stéphane Guillard on 01-Oct-2004 05:52 GMT | In reply to Comment 278 (Georg Steger): Hi georg,
Hi Michael,
Here is an answer to your similar questions (what if cacheflushed buffer not properly aligned ?)
You have a valid point... that does not stand ;)
Indeed, caching is done on cacheline size granularity (32 byte for ppc) (and not on 64k !), and not on a byte per byte basis.
So, something bad could happen, if you cacheflush a memory area (get me right : try to flush those cachelines which map your memory area, which count from zero up to area size / 32 + a few if area mis aligned :), and this memory area is not 32 byte aligned at *either or both* ends.
This *could* be bad, because you would have cached data cast outs to memory for all the 32 bytes of the cacheline, and maybe those are only 1 byte for your area, and 31 below (if flushing a cacheline that maps your misaligned area start) or above (cacheline maps area end).
Those castouts will go to memory (they can of course never send to unmapped areas and end up with a bus error, since virtual memory pages have much stronger alignment constraints), which is not a problem in itself... Except in *one* case : if the unwanted castouts go to a buffer which has just been dma'ed to by another driver, and which has not been invalidated *before* the read to ram by this other driver (this is a faulty design, according to my post 219, but like someone else pointed out, some possibly quickly ported ahi drivers might not take that into account, for example). In *this* case, your cache flush will trash the other driver's 'just dma'ed to' buffer.
In other words : if a dma driver does not invalidate his buffer before starting a dma read to it, and if by bad luck his driver is just adjacent to yours, and if the frontier between both is not 32 byte alined, and if he has just done a dma read to his buffer, and you flush the cacheline covering the frontier, you may overwrite what his dma device wrote to memory.
Nota bene : there is nothing specific to either A1 or OS4 here, this is only PPC cache rules.
Anyway, this situation should not happen in OS4 except for maybe drivers that were ported with overlooking this point :
- our ethernet driver,
- our usb driver,
- our ide drivers,
- our ahi drivers (at least daveae's sblive, cmi, etc, ross v. & davy w. will confirm)
(that is, all our dma drivers) have been designed with this point in mind.
Examples :
- for the ide drivers, they usually get buffers from the filesystems that are properly aligned. If that was not the case for whatever reason like a caller not being a filasystem but whatever application like a benchmark tool that feeds a CMD_READ with an even aligned buffer (!), this would still not be a problem, because in this case my code uses its own properly aligned buffer, dma's to it, and then memcpy()'s to the caller's buffer.
- for the ahi drivers, davy w and ross v also had some smart work to do, because the double dma buffer system used in ahi has frames which are not 32 byte multiples in size, so they had to tweak a bit, making sure that the memory around their flushed areas is their's.
- same goes for other drivers i mentioned.
By the way :
- the potential problem is obviously independant of hardware or software cache coherency : it is a matter of making sure that you are not 'side'flushing into a dumb (ie non invalidating) dma driver's buffers...
- the point 'and what if someone accesses your buffer while you're dma'ing into it' does not stand either, who would be so dumb as to program something like that, except to make a real random trash generator ?
Kind regards,
--
Stéphane |
|
Notes about my presentation at the Alchimie 4 show : Comment 319 of 427 | ANN.lu |
Posted by MikeB on 01-Oct-2004 05:56 GMT | In reply to Comment 312 (Christophe Decanini): @ Christophe Decanini
> And there are a lot of OS4 people and AROS people and others and I'am glad
> that they are here if they don't abuse ANN.
1) I don't think there are a lot of "AROS people" yet. The ones posting here are mostly seem to be directly involved in the project.
2) I believe a vast majority of AOS4 folk here behave pretty well. Even the much ridiculed Samface behaves pretty well on this board, mostly resorting to reasoning (most MOS folk don't agree with) instead of insults and rival FUD.
> None of them got to the point you were when you where asked to leave.
I nearly never resort to insults and try to reason with people (agree with my reasoning or not)
Most of my conflicts in the past have been directly or indirectly (through for instance people like you) bbrv related. I tried to warn the community based on my own experiences and criticized bbrv when they spread around their propaganda on the already established Amiga websites at the time. There were no good alternatives available at the time, gladly AmigaWorld.net has meanwhile grown into the biggest English languaged Amiga community portal and offers a place where topics can still be discussed without all the noise we see elsewhere.
IMO lots of the issues could have been avoided if people would have approached bbrv more cautiously. So with regard to many people I have still failed with regrd to what I tried to accomplish, they have now been burned themselves, but there were also people who took their smooth talking more catiously. In any case I only provided the information I knew, it's up to the individuals themselves to make their choices.
@ Fabio Alemagna
> Let's also say "some dealers", but not all of them.
IMO this just shows how little you actually know. There may have been one or maybe two dealers on the list.
> This topic has been talked about to death on Italian ML's thanks to the fact
> that Luca Diana was involved in it.
I agree with you for once. This topic has been talked about and FUDed about to death, no need for the outsiders to bring this up every now and then.
> Damage control was the rule, nothing of what was going on for real had to
> spread around the world
There was a large scala of different topics being discussed. The idea was for the community representatives to provide straight advise and criticism (while staying polite and reasonable) to help forward the Amiga platform. It had nothing to do with business decisions for dealers as you like to believe. There was a large variety of knowledge and experience on the list. The aim was to help advance the platform and companies involved while understanding the difficult and unfortunate situation they were in at the that poiny of time. |
|
Notes about my presentation at the Alchimie 4 show : Comment 320 of 427 | ANN.lu |
Posted by Stéphane Guillard on 01-Oct-2004 06:03 GMT | In reply to Comment 318 (Stéphane Guillard): (adding a bit to what i said)
obviously, we assume that the 'other dma driver' invalidates his buffer, otherwise it would be unusable, but it does so *after* and not *before* his dma read to ram, which is faulty and an open door to such 'side effect' cacheflush trash.
Regards,
--
Stéphane |
|
Notes about my presentation at the Alchimie 4 show : Comment 321 of 427 | ANN.lu |
Posted by Fabio Alemagna on 01-Oct-2004 06:39 GMT | In reply to Comment 319 (MikeB): > > Let's also say "some dealers", but not all of them.
>
> IMO this just shows how little you actually know. There may have been one or
> maybe two dealers on the list.
D'oh, your comprehension skills seem to improve each and every day, uh?
in what way, exactly, what you wrote contraddicts what I wrote? Isn't it rather a confirmation? Have you got my point at all?
You yourself confirm that "one or two dealers" were on the list, hence those were priviledged in respect of the other ones, and could make much more focused business decisions, including not making any decision at all when it was clear the business was not viable anymore, something other dealers couldn't do, and were thus disadvantaged by this.
Believe me, I know a lot more about it that you can imagine.
> > This topic has been talked about to death on Italian ML's thanks to the fact
> > that Luca Diana was involved in it.
> I agree with you for once. This topic has been talked about and FUDed about to
> death, no need for the outsiders to bring this up every now and then.
Oh, please... the master of FUD complaining about FUD. How more pathetic can it get?
> > Damage control was the rule, nothing of what was going on for real had to
> > spread around the world
> There was a large scala of different topics being discussed. The idea was for
> the community representatives to provide straight advise and criticism (while
> staying polite and reasonable) to help forward the Amiga platform. It had
> nothing to do with business decisions for dealers as you like to believe.
I don't "believe" it, it's simply true. Of course the topic was never "what business decision does this dealer have to make today", but it's an undeniable fact that the dealers involved in that group of elitists were advantaged over the ones who weren't part of it. The fact they could take more focused business decisions was a consequence of that fact, nothing else.
> > There was a large variety of knowledge and experience on the list. The aim
> > was to help advance the platform and companies involved while understanding
> > the difficult and unfortunate situation they were in at the that poiny of
> > time.
You are not at all contraddicting what I said. Your opinion about how to "help advance the plaform and companies involved" may very well include damage control, as in fact it does, as everyone can see, also right now. You always try to give a good spin to bad news, just like you've done in this tread all along, and that was why you were in that list in the first place: you're a master in such tactics. Luckily not everyone has been fooled. |
|
Notes about my presentation at the Alchimie 4 show : Comment 322 of 427 | ANN.lu |
Posted by Real Asemoon from England on 01-Oct-2004 06:47 GMT | In reply to Comment 319 (MikeB): "1) I don't think there are a lot of "AROS people" yet. The ones posting here are mostly seem to be directly involved in the project."
Check aros-exec.org
There is more potential and users than you tought. Many developers from the MorphOS and OS4 teams contribute to the AROS development. Not to mention there are many ex-Amiga users looking for an alternative solution. Windows suxx, Linux is crap, why not try AROS?
"2) I believe a vast majority of AOS4 folk here behave pretty well."
I believe a vast majority of non-OS4 folk here behave pretty well.
"gladly AmigaWorld.net has meanwhile grown into the biggest English languaged Amiga community portal"
Amiga is dying. Developer lists are shrinking if not empty, unpaid salaries in Amiga Inc/Genesi Inc/KMOS Inc, Amiga resellers dropping like flies and new blood is not coming in. Genesi Inc and Amiga Inc employees disappeared to other jobs. Ever wondered why Fleecy is quiet? He got a job and no longer works for Amiga Inc. KMOS is liquidating Amiga Inc assets soon. The whole community is dying and I mean the Amiga community as whole.
Don't say I'm not Asemoon... It is not a secret Asemoon was your alter ego. |
|
Notes about my presentation at the Alchimie 4 show : Comment 323 of 427 | ANN.lu |
Posted by Anonymous on 01-Oct-2004 06:51 GMT | In reply to Comment 322 (Real Asemoon from England): > Don't say I'm not Asemoon... It is not a secret Asemoon was your alter ego.
So you are mike bouma?
Or you are not Asemoon?
Eh? |
|
Notes about my presentation at the Alchimie 4 show : Comment 324 of 427 | ANN.lu |
Posted by Andrea Maniero on 01-Oct-2004 06:58 GMT | In reply to Comment 319 (MikeB): >> Let's also say "some dealers", but not all of them.
>
> IMO this just shows how little you actually know. There may have been one or
> maybe two dealers on the list.
Which is exactly what Fabio said. Surely one or two are not ALL the Amiga dealers. And those two (assuming they were two) had inside knowledge the others did not. Therefore they could make different (and safe) businness decisions, even with the chance to "damage" (since there is not a monopoly in Amiga dealership) the other dealers. Plain simple.
>> Damage control was the rule, nothing of what was going on for real had to
>> spread around the world
>
> There was a large scala of different topics being discussed. The idea was for
> the community representatives to provide straight advise and criticism (while
> staying polite and reasonable) to help forward the Amiga platform.
The idea was DOA, if the community representatives had to be you, Luca Diana and the likes, IMHO. And damage control IS still the rule. Look at this thread. Suddenly the "evil" Linux, too focused on x86 to take advantage of Articia's futuristic features, will work unmodified on the next batch of (bug fixed - when there was NO bug) A1s. And that's great news (well, honestly it should have worked this way from day 1)! But what about the thousand or so motherboards that we can finally aknowledge as bugged? And what about the lies (there's no bug - you'll see when AOS4 will have DMA)? Didn't Stephane aknowledge that lack of cache coerency is a bug indeed, in comment 219?
> It had nothing to do with business decisions for dealers as you like to
> believe.
But they could take businnes decisions based on what was not public knowledge. Fact.
Kind regards,
Andrea |
|
Notes about my presentation at the Alchimie 4 show : Comment 325 of 427 | ANN.lu |
Posted by Anonymous on 01-Oct-2004 07:11 GMT | In reply to Comment 324 (Andrea Maniero): > The idea was DOA, if the community representatives had to be you, Luca Diana
> and the likes, IMHO. And damage control IS still the rule.
In this case given advice is often solicited and rarely taken maybe you should not lay the blame at the "elite members" ( as those with a chip on their shoulder about it will look at it ) but at those that fail to listen even to their friends. However, advice is given always in the knowledge that the recipient does not have to take it.
Bouma and Diana do damage control on the forums perhaps, but gloves off in private is the norm for anyone involved in something like this or there would be no point in having a private list if these people are going to say the same thing in private as in public. This is no different for the political classes involved in any project be it MorphOS, AROS, AmigaOS4 or AmigaDE.
What information do you think was kept private away from the dealers that they would need to know or feel "disadvantaged"? That the 686b had a bug? Do you have evidence this was ever discussed on that mailing list? No, you do not.
Given you do not know the subject or content of the private list mailings you are in no position to judge or make such claims.
In the real world not all information is public, not all information needs to be shared and not all information that is undisclosed is pertinent to people that have not received it.
Or do you think confidential information should be handed out to everyone and then returned if they say "ok, that is not relevant to me, you don't need to keep this private"?
Deal with it, or start demanding freedom of information from all sources regardless of political position.
I do not see the point of this discussion anyhow, unless it is an attempt to drive a wedge between the dealers and those that are unable to defend themselves without breaking confidentiality. Or maybe it is an attempt to apply pressure to get at the contents of those mailing lists? |
|
Notes about my presentation at the Alchimie 4 show : Comment 326 of 427 | ANN.lu |
Posted by MikeB on 01-Oct-2004 07:25 GMT | In reply to Comment 321 (Fabio Alemagna): > in what way, exactly, what you wrote contraddicts what I wrote? Isn't it
> rather a confirmation? Have you got my point at all?
Nope, I will probably go beyond your reasoning capabilities. But here goes:
There are 28 official 3rd party AmigaOne dealers of which *none* I know of was part of this list. This *one* semi-dealer I am talking about happened to be an important editor for a very big magazine. And I believe this was the primary reason why he was invited onto this list.
So how could what you are suggesting possibly be correct?
> You yourself confirm that "one or two dealers" were on the list
Do you know what the word "maybe" means? Well, it means something similar to "perhaps" and "possibly".
> hence those were priviledged in respect of the other ones, and could make much
> more focused business decisions, including not making any decision at all when
> it was clear the business was not viable anymore, something other dealers
> couldn't do, and were thus disadvantaged by this.
Pure Bullshit as usual. |
|
Notes about my presentation at the Alchimie 4 show : Comment 327 of 427 | ANN.lu |
Posted by Anonymous on 01-Oct-2004 07:33 GMT | In reply to Comment 325 (Anonymous): PS: If the discussion degenerates into a case of "you would say that Bouma you were on the inside and you are just covering your arse" from people who clearly were not on that list you know its predicated on bullshit. If they produce message excerpts from the list and message headers that can be verified to back up their statements of "here is the evil purpose of that list" then it is a different matter. |
|
Notes about my presentation at the Alchimie 4 show : Comment 328 of 427 | ANN.lu |
Posted by Fabio Alemagna on 01-Oct-2004 08:01 GMT | In reply to Comment 326 (MikeB): > > in what way, exactly, what you wrote contraddicts what I wrote? Isn't it
> > rather a confirmation? Have you got my point at all?
>
> Nope, I will probably go beyond your reasoning capabilities. But here goes:
Mike, stop it already, your derogatory attitude doesn't impress anyone... you look like a 3 years old fighting with his kindergarden-mates.
> There are 28 official 3rd party AmigaOne dealers of which *none* I know of was
> part of this list. This *one* semi-dealer I am talking about happened to be an
> important editor for a very big magazine. And I believe this was the primary
> reason why he was invited onto this list.
You are thus confirming that there were people who used to make money from the Amiga market that were part of the list, whilst some others (the majority) weren't part of it. You call it a "semi-dealer", which is a term you should first define, and you say he was for sure an important editor of a big magazine, without realizing that the fact he sold magazines rather than machines makes no difference whatsoever for the point I was making.
I know there was at least one Italian amiga dealer involved in the list, though, via another Italian guy who did nothing but lead some Italian mailing lists.
> So how could what you are suggesting possibly be correct?
Are you implying I should disregard my knowledge of the facts for the one you're pushing forward? Was that meant to be a joke or what?
> > You yourself confirm that "one or two dealers" were on the list
>
> Do you know what the word "maybe" means? Well, it means something similar to
> "perhaps" and "possibly".
Ah, so you are at least confirming I could possibly be right, then? Hey, your own words, which however contraddict the ones you yourself wrote at the beginning of the message I am replying to.
So, let's put it straight: was it or was it not any Amiga dealers or any other entity making money in any way from the Amiga market, involved in that list?
You say "maybe", thus I'm "maybe" right, thus you don't know more about this than I do, thus your rants are completely useless.
> Pure Bullshit as usual.
Nice, concise and well worder reply, not to mention full of solid facts. :-)
So, where did the rest of my message go? |
|
Notes about my presentation at the Alchimie 4 show : Comment 329 of 427 | ANN.lu |
Posted by Anonymous on 01-Oct-2004 08:12 GMT | In reply to Comment 325 (Anonymous): "In the real world not all information is public, not all information needs to be shared and not all information that is undisclosed is pertinent to people that have not received it. "
Apparently, according to the usual physically sickening Mike Bouma rhetoric, he among others were "community representatives". If that revolting and ludicrous expression is to be taken seriously, they should per definition not have kept the discussion away from the view of the people that they self-allegedly "represented".
Just an observation.
Then again, nobody in his right mind would take Mike Bouma seriously, even if they didn't bother testing what he says against observable reality and logic. Someone in this thread said he's a master of spin, or words to that effect. I think he's invariably and embarrassingly transparent, which along with the dishonesty and hypocrisy adds to him being generally seen as a particularly repulsive character, which aren't desirable traits of a skilled spin doctor. |
|
Notes about my presentation at the Alchimie 4 show : Comment 330 of 427 | ANN.lu |
Posted by MikeB on 01-Oct-2004 08:25 GMT | In reply to Comment 328 (Fabio Alemagna): This is my last reply to this thread. Most people here know near to nothing about the true purpose of the Amiga Community Advisory Group as it was intended and evolved. Some people here are only "fishing" in the hope you find anything worthy to cry outrage about. IMO these are the same people who cry outrage about loads of other issues not really concerning themselves.
> You are thus confirming that there were people who used to make money from the
> Amiga market that were part of the list
No, I'm not confirming this. However I am pretty confident that some people on the list who for instance contributed to Magazines earned money.
> call it a "semi-dealer", which is a term you should first define
Someone I believe was NOT a dealer at the time and still am not sure about if he is one today, although he had such intentions.
> and you say he was for sure an important editor of a big magazine, without
> realizing that the fact he sold magazines rather than machines makes no
> difference whatsoever for the point I was making.
It does. It shows you are just "talking out of your arse", figurally speaking.
> I know there was at least one Italian amiga dealer involved in the list,
> though, via another Italian guy who did nothing but lead some Italian mailing
> lists.
Don't know, I do know Luca Diana was part of the list. But he quit when people like you tried to drag this constructive effort into a bad daylight. I believe he even left the Amiga community for good. (Well done guys...)
It even took a long time (after people like Luca & Wayne left the list) for even to have Alan Redhouse to be on the list. In fact if it wouldn't have been for my efforts he probably would never have been!
It is so incredibly obvious you nearly know nothing about what acag was used for. IMO you are only dragging your own integrity through the dirt with this kind of behaviour.
Back to: http://www.amigaworld.net ! |
|
Notes about my presentation at the Alchimie 4 show : Comment 331 of 427 | ANN.lu |
Posted by Gabriele Favrin on 01-Oct-2004 08:38 GMT | Interesting: out beloved troll is back. No matter he has been asked
to give his excuses to italian users and leave, he is back
to annoy all of us with his flames. Not bad. Tell me Mike,
have you read my pretty story about you? ;)
Now I'm going to be even more clear here: what happened with
ACCM is simply ILLEGAL. Here in Italy there are four Amiga
dealers (at that time two, iirc). One of them was present on
the list and has been able to know FACTS about Amiga Inc
financial situation. Ehy, Mike, do you recall the long post
where McEwen told all about the legal actions of Thendic
that had HIS OWN back account being interested? Do you? This
allowed a dealer (or maybe more than one) to know something
about the Amiga financial situation while other dealers
didn't have that information available to let them decide if
support Amiga inc projects or not. And that's not enough: on
the list there was also people from some printed magazines,
giving to their editors an economic advantage, being them
able to KNOW (not to PREDICT as others) the Amiga situation.
In a word this is UNFAIR COMPETITION and in a biggest market
this would have lead to legal actions. It surprises me that
a graduate in economic stuff as Luca is created such a
thing. Maybe he has been forced from people with a lot less
experience?!
Btw, italian users can read my long analysis of the Amiga
situation and social problems created by Amiga Inc
behaviour (the ACCM thing is covered too). It's on
http://www.amigapage.it/commenti.php?modo=v&idnotizia=N30092004-001
Flames and Amiga talebans are not welcome. Civil discussion
is. I wrote it yesterday without knowing about the new ACCM
thread here.
Btw, and you and Luca never have been elected to represent
the Amiga Community. If there is still an Amiga community,
you have nothing to do with it. Amiga means support betweenn
users and search for the best option, not fud, hidden
agenda, censorship and so on. |
|
Notes about my presentation at the Alchimie 4 show : Comment 332 of 427 | ANN.lu |
Posted by Fabio Alemagna on 01-Oct-2004 09:06 GMT | In reply to Comment 330 (MikeB): > It is so incredibly obvious you nearly know nothing about what acag was used
> for. IMO you are only dragging your own integrity through the dirt with this
> kind of behaviour.
Mike, Luca was my friend. We used to share a lot of information, and he used to tell me things I wouldn't have otherwise known. I'm beginning to wonder whether I know more than you actually know about this issue... |
|
Notes about my presentation at the Alchimie 4 show : Comment 333 of 427 | ANN.lu |
Posted by Lamont Cranston on 01-Oct-2004 09:38 GMT | In reply to Comment 322 (Real Asemoon from England): >unpaid salaries in Amiga Inc/Genesi Inc/KMOS Inc,
only Amiga Inc and Genesi have former employees who haven't been paid. That (as of right now) is not the case with KMOS. |
|
Notes about my presentation at the Alchimie 4 show : Comment 334 of 427 | ANN.lu |
Posted by Andy Hall on 01-Oct-2004 09:59 GMT | In reply to Comment 332 (Fabio Alemagna): *You*.....had..a.........friend?!? Wow! I bet that didn't last long.
BTW it's good to see you're more interested in AmigaOS 4.0 and fighting with MikeB than addressing concerns about the viablity and usefulness of AROS |
|
Notes about my presentation at the Alchimie 4 show : Comment 335 of 427 | ANN.lu |
Posted by Anonymous on 01-Oct-2004 10:09 GMT | In reply to Comment 334 (Andy Hall): What usefulness?
Ok, that one was very easy, but I just couldn't resist ;-) |
|
Notes about my presentation at the Alchimie 4 show : Comment 336 of 427 | ANN.lu |
Posted by Fabio Alemagna on 01-Oct-2004 10:23 GMT | In reply to Comment 334 (Andy Hall): Yeah yeah, as if anyone cared about "Andy Hall"'s unsubstantiated beliefs about my private life.
If it amazes you so much I have friends, then perhaps you'll get an heart attack by knowing that I actually even have a girlfriend!
How was that for a shock, uh? Still on your chair? Need assistance? |
|
Notes about my presentation at the Alchimie 4 show : Comment 337 of 427 | ANN.lu |
Posted by Christophe Decanini on 01-Oct-2004 11:02 GMT | In reply to Comment 330 (MikeB): "It even took a long time (after people like Luca & Wayne left the list) for even to have Alan Redhouse to be on the list. "
Which is another lie. Luca was still on the list when Alan got in.
Your presence on the list and your attitude he showed that your role was to control the opinion and attack the users that would either prefer another solution or criticize (objectively or not) the Amiga side.
Your propaganda here, your attacks on users and moderators, your effort to link all news threads you posted to another news site you controled participated in the fact that you were asked to leave this site.
As far as I'm concerned it was a great decision. |
|
Notes about my presentation at the Alchimie 4 show : Comment 338 of 427 | ANN.lu |
Posted by Christophe Decanini on 01-Oct-2004 11:14 GMT | In reply to Comment 331 (Gabriele Favrin): It looks that not despite the fact that Mike think that everyone arguing with him do it because they have an agenda with Genesi we see a lot of people like
Gabriele, Fabio who are not directly involved with Genesi.
It confirms my idea that with Mike attitude of elitism and elimination of the people not joining it Amiga has lost prominent figures in the community.
Fortunately there are alternatives out of the "Bouma world" or "Bouma and friends". |
|
Notes about my presentation at the Alchimie 4 show : Comment 339 of 427 | ANN.lu |
Posted by Don Cox on 01-Oct-2004 11:18 GMT | In reply to Comment 285 (Agima): "Sure, you're right there are none. Huh, I wonder why the dual kick start boards that let you have 1.3 and 2.0 were so popular if there was no software that didn't work under 2.0 that did work on 1.3 machines then."
I never installed or needed a dual kick start board. It was only needed for games.
Give examples. I'm sure there were one or two, but I have about 600 floppies of original Amiga programs here, and I can't think of any that stopped working in 2.0. |
|
Notes about my presentation at the Alchimie 4 show : Comment 340 of 427 | ANN.lu |
Posted by Don Cox on 01-Oct-2004 11:19 GMT | In reply to Comment 286 (MikeB): "Anyway amongst the reasons why features like Virtual Memory and Resource Tracking were never implemented in classic AmigaOS releases in the past was because of software compatibility issues this would cause. "
Virtual memory has been available for Amigas for at least 15 years. |
|
Notes about my presentation at the Alchimie 4 show : Comment 341 of 427 | ANN.lu |
Posted by Christophe Decanini on 01-Oct-2004 11:46 GMT | In reply to Comment 340 (Don Cox): "Virtual memory has been available for Amigas for at least 15 years."
You are wrong ;)
It has been available as a third party utility. |
|
Notes about my presentation at the Alchimie 4 show : Comment 342 of 427 | ANN.lu |
Posted by Johan Rönnblom on 01-Oct-2004 12:01 GMT | In reply to Comment 333 (Lamont Cranston): Lamont Cranston wrote:
> Only AmigaInc and Genesi have former employees who haven't been
> paid. That (as of right now) is not the case with KMOS.
Wrong. First, AmigaInc is part of KMOS, and thus KMOS has taken over
all the responsibilities of AInc. This has been expressly confirmed
by Garry Hare of KMOS - but it is of course a legal requirement if
they indeed bought AInc.
Second, Bolton Peck has confirmed on MooBunny that KMOS have not paid
him in part or in full, only a few days ago. It seems reasonable to
assume that the same is true for others. |
|
Notes about my presentation at the Alchimie 4 show : Comment 343 of 427 | ANN.lu |
Posted by Luca Diana on 01-Oct-2004 13:36 GMT | In reply to Comment 313 (Fabio Alemagna): Time goes by but Fabio Alemagna never grows up. I wonder if you'll ever realize that there's more to life than petty bickering. I'm not too sure you do.
I left the community long time ago because of people like YOU, there's only so much a mature person is willing to put up with.
I'm here just because Mike brought your email to my attention and you can be certain that once I said what I have to say I will not reply, nor read, any of the spoiled-kid-stomping-his-feet-on-the-ground insults you will throw at me. Again, the Amiga community to me is dead.
What was discussed on ACCM was far less important than you think, obviously your conspiracy theorist mania dictates differently. The rest of your claims come from your fear that there's alays someone out there to get you.
And I still don't understand what's this obsession you have with me, it's been what...? 2 years since we stopped talking? Just drop me and get on with your life.
To Christophe Decanini:
Totof, I was never in ACCM when Alan Redhouse joined (did he?), although I did invite him at least twice and he never replied. That's all I remember, as I said, I stopped caring a long time ago and all details are, at best, blurry.
Overall, once again let's give way to sensationalism because that's all the "community" (all 20 of you) wants. Infinite threads over the same old arguments, not even the C64 vs Amstrad or Amiga500 vs Atari ST wars were so boring.
I don't have harsh feelings toward anyone (after all, we all took a side, even those who claim they didn't. Everybody had his beliefs and acted accordingly, it's just human nature. I don't even have bad feelings toward Gabriele Favrin, I might not understand him, but I know he has a big heart), only exception is Bill Buck. He lied and he plotted with malevolent intentions since the beginning and ultimately I hold him responsible for splitting the community and eating resources on both sides just like a cancer, resources that would have made a huge difference in the medium and long terms, but everyone was too busy throwing mud to notice his plan, which ultimately led to the distruction of both sides, sides that could have peacefully coexisted. Ultimately I compare him very much to George Bush.
Go on with your lives guys, let the dream remain a dream before it's too late, there's nothing wrong with it. |
|
Notes about my presentation at the Alchimie 4 show : Comment 344 of 427 | ANN.lu |
Posted by Luca Diana on 01-Oct-2004 13:37 GMT | In reply to Comment 343 (Luca Diana): BTW, I got a package in the mail from Amiga a few days ago. Haven't opened it yet but I believe it's a t-shirt ;-)
Oh, anybody wants to buy a Massachusetts license plate that reads "AMIGA" :-D |
|
Notes about my presentation at the Alchimie 4 show : Comment 345 of 427 | ANN.lu |
Posted by Luca Diana on 01-Oct-2004 13:43 GMT | In reply to Comment 313 (Fabio Alemagna): >Let's also say "some dealers", but not all of them.
There was not a single dealer in the list, I just checked the subscribed members page. |
|
Notes about my presentation at the Alchimie 4 show : Comment 346 of 427 | ANN.lu |
Posted by Bernie Meyer on 01-Oct-2004 13:45 GMT | In reply to Comment 318 (Stéphane Guillard): Stephane, you have just singlehandedly destroyed a whole lot of confidence in the technical savvy of the OS4 driver writers (or at least the IDE driver writer). Because your explanations in that comment were utter crap!*If* you have a non-cacheline-aligned DMA buffer, you run into all sorts of fun (not the "that was hilarious" kind of fun, though, but rather the "oops, I guess we fucked up" kind of fun) when reading.Scenario 1: You yourself say that before a DMA read, you *invalidate* the affected cache lines, rather than flushing them, stating that flushing would be a waste of time, seeing as the data is going to be DMA'ed over, anyway.... But if parts of the cachelines *aren't* going to be DMA'ed over, you just invalidated potentially valid and dirty data. Oops.Scenario 2: You manually pre-DMA invalidate the cache. Then some other thread writes to a variable that shares a cacheline with your DMA buffer, thus pulling a copy back into cache and dirtying it. Then DMA happens; You now have a cacheline which has *some* valid data in main memory (the DMA'ed stuff) and some valid data in cache (the variable). And as far as the CPU is concerned, the cached copy is authorative, so the DMA'ed stuff is lost.Scenario 3: You do a DMA read, and for whatever reason, between the time you invalidate the cache and the time DMA actually happens, some process looks at some part of your buffer, or something close enough to share a cacheline with your buffer. Boom, once again, the line gets pulled into the cache, and becomes authorative. This one you can (mostly) deal with by invalidating the cache again, after the DMA read, but see Scenario 2 --- if you do that, you end up choosing the DMA stuff over the variable stuff.No, the only thing you can do if you don't have hardware cache coherency (which would deal with all of these scenarios without problem) and are provided with non-cacheline-aligned buffers (which, judging by my experience of writing the zero-copy IDE stuff for Umilator, is most of the time), is to use bounce buffers. Bounce buffers, of course, are rather unpleasant from a performance point of view, not because the CPU-driven copy itself takes all that long (although it isn't nice), but particularly because of what they do to the caches.The other alternative, of course, would be to simply disallow any memory accesses between the time you invalidate (or flush, for partial cachelines) your buffer, and the time DMA has completed. But that would go against the whole point of DMA, which is that the CPU is supposed to be able to do something else. |
|
Notes about my presentation at the Alchimie 4 show : Comment 347 of 427 | ANN.lu |
Posted by Don Cox on 01-Oct-2004 13:58 GMT | In reply to Comment 341 (Christophe Decanini): ""Virtual memory has been available for Amigas for at least 15 years."
You are wrong ;)
It has been available as a third party utility."
That's right. I didn't say it came free. |
|
Notes about my presentation at the Alchimie 4 show : Comment 348 of 427 | ANN.lu |
Posted by Christophe Decanini on 01-Oct-2004 14:02 GMT | In reply to Comment 343 (Luca Diana): "To Christophe Decanini:
Totof, I was never in ACCM when Alan Redhouse joined (did he?), although I did invite him at least twice and he never replied. That's all I remember, as I said, I stopped caring a long time ago and all details are, at best, blurry."
When you presented the ACCM list to me you told me about who was participating there and you showed me that Alan was posting there.
Even if we had arguments with you in the past I agreed to disagree with you and understood that contrary to me your interest in Amiga was maily for AmigaDE. It was your choice and I respected it while I told you I didn't like Amiga Inc decisions and way of proceedings.
Mike is a complete different story. He is not straight like you, me, fabio or Gabrielle. Maybe it has something to do with our Italian blood ;)
I don't like Mike's way to see everything as great and spin things around when the situation is really deteriorating and when he participate to sectarise (spelling ?) the rest of the people who went through all the downfalls of the Amiga.
You must have seen that I was one of the most fervent Amiga supporters back when you came home. Now it looks that for some people (including Mike) I don't deserve to be an Amigan anymore.
I will still be an Amigan in the spirit and will still use "Amiga like solutions". I will eventualy go back to Amiga when the conditions will be different, which mean fully working hardware at decent price, support independent from Mike Bouma & friends.
Luca, you are welcome anytime at my place to have a drink and have a friendly discussion. |
|
Notes about my presentation at the Alchimie 4 show : Comment 349 of 427 | ANN.lu |
Posted by Bernie Meyer on 01-Oct-2004 14:04 GMT | In reply to Comment 346 (Bernie Meyer): Oh, and BTW --- I may be mistaken here, but I'd expect the following to have a deterministic and obvious result, which it does not have if you *invalidate* rather then *flush* the cache prior to DMA:(a) allocate large-ish buffer (b) fill it with data using CPU-writes (c) initiate DMA into the buffer (d) [this is the important part] encounter an I/O error, cutting short the DMA transfer (f) read the data from the bufferWhat one would expect to get in step (f) is all the data that *was* DMA'ed, and whatever was written to the buffer by the CPU for those areas that DMA did not happen for. What you'd get by using cache-invalidate pre-DMA would be all the data actually DMA'ed --- but the memory where DMA did not take place may contain *anything* at all. Some parts may contain the data written by the CPU, others may not. There'd be no guarantees at all. |
|
Notes about my presentation at the Alchimie 4 show : Comment 350 of 427 | ANN.lu |
Posted by Christophe Decanini on 01-Oct-2004 14:05 GMT | In reply to Comment 344 (Luca Diana): Did you move back to Italy ? |
|
|
- 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.
|