Home > Amiga & AROS > Reports from “OS4 on Tour 2003” in BathReports from “OS4 on Tour 2003” in Bath Mike Bouma 2003-12-16 Amiga & AROS 48 CommentsShow reports from last Saturday’s AmigaOS4 event in Bath(UK) are now available here, here and here. Also a short video from the event is available which shows Quake2 running hardware accelerated on AmigaOS4 and a real-time calculating fractal demo.About The Author 48 Comments 2003-12-16 7:32 pm when is it going to be released already ? 2003-12-16 7:57 pm A pre-release is planned for the coming month. This will enable current AmigaOne owners to finally use AmigaOS4 while waiting for the final release, but this version will only receive limited support as finishing the project currently deserves top priority.A final release will follow “When it’s done!” and the team is fully satisfied. The AmigaOS4 team doesn’t want to make predictions like they had done in the past, as there will always be some unknown factors to deal with and any disappointments may be (ab)used against the project and team effort by competitors. 2003-12-16 8:03 pm That concept was thrown out of the Amiga business model decades ago!The new managament appears to be a bunch of Underwear Gnomes (obligatory South Park reference), and their business plan goes something like:1. Create a computer ahead of its time2. Mis-manage resources and funds3. Alienate users and miss deadlines4. Pass technology from owner to owner.6. ???7. ProfitSimple, isn’t it!Seriously though… I don’t understand their logic. Why have monthly press releases, when you can’t even meet annual release schedules?I guess if their business is killing trees and spreading rumors, they’re doing pretty good, but a lot of press releases, and insanely optimistic viewpoints do not make up for missed deadlines, and no tangible product. 2003-12-16 8:11 pm Quake2? Yay. Not. I was a happy Amiga user. Pay respect to Amiga, and let it die. 2003-12-16 8:36 pm Could someone make those Amigans stop doing those “gotta-use-every-wipe-and-transition-available”-videos….Yuk! 2003-12-16 9:09 pm >A final release will follow “When it’s done!” and the team is fully satisfied.I look at Amiga now and I see the YellowTab 5 years from now. Frightening. 2003-12-16 9:12 pm They seem to have been touring around with this OS for more than a year! And it isn’t released yet… 2003-12-16 10:13 pm //event is available which shows Quake2 running hardware accelerated on AmigaOS4 //Quake 2.No, seriously. They’re proud of getting Quake *2* to run.“Too much time on my hands.” 2003-12-16 10:45 pm Afraid not. That said, the (utter lack of) availability of games for AmigaOS is the least of my concerns. 2003-12-16 10:57 pm @ Rockwell> No, seriously. They’re proud of getting Quake *2* to> run.No, Quake2 (running hardware accelerated) and Hyperion’s Warp3D technology is something which was new compared to the many previous demonstrations. DukeNukem3D and Quake had already been shown using software rendering. It’s another step forward. Quake2 and Warp3D were running well, e.g. you could move the Quake2 window around without noticing slowdowns, for open minded people this may pave the way for other Quake2-engine based games to be ported, as this should now be quite easy. For instance Hyperion has a huge library of Amiga game licenses lined up including Freespace, Heretic2, Soldier of Fortune, Sin, Worms Armageddon, Shogo, Midnight Racing and various yet unannounced titles. The Quake2 demo just demonstrates something which is now already done.As can be seen from previous demonstrations, AmigaOS4 will have a pretty decent library of software to start with. Everything from word processors, paint programs, spreadsheets, email/IRC/browser/PDF software, classic RTG games, advanced development and poweruser tools, etc have already been demonstrated to work with AmigaOS4.With regard to all the critics here I just want to add that I am personally amazed by how much the AmigaOS4 team has accomplished already with rather little resources, while the huge monopolistic OS vendors accomplish IMO relatively increbly little with mind-boggling resources! The AmigaOS4 developers come from all over the world and small companies from all over the world are co-operating to bring this project to a succesfull start. 2003-12-16 11:15 pm I’m an Amiga enthusiast from way back..I think its sad that the Amiga name is being kept alive like an undead zombie, can’t it be laid to rest with dignity? 2003-12-16 11:17 pm Has the AmigaOne still got the hardware data-corruption bug? 2003-12-16 11:36 pm @ Mr. BannedThis is not a press release, these are show reports (and a video) from Amiga users who attended an “OS4 on Tour” event.The AmigaOS4 demonstrations organized this year in Australia (Brisbane, Canberra, Melbourne, Sydney), Austria, Canada, France, Germany, Hungary, Italy, Slovenia, Spain, Sweden, Switzerland, The Netherlands, UK (Bath, Basildon, Birmingham, London), US (Sacramento, New York) and the other events I forgot about were mainly organized by Amiga ethusiasts and usergroups. Of course Eyetech, Hyperion and Amiga Inc provided support, but most of the current AmigaOS4 news items are community related. The more mainstream targeted PR is planned to start roling when the AmigaOS4 project is completed.IMO this shows that the Amiga platform has an important ace in their hands: The great community spirit of the Amiga community! 2003-12-16 11:42 pm The VIA (DMA enabled) bugs exists in millions of mainstream PCs and don’t cause any such problems at all. And there will be no such problems when AmigaOS4 is released neither.For a more complete story with regard to these overhyped problems (by the competition) can be read within my recent interview with Alan Redhouse (Question 9):http://amigaworld.net/modules/features/index.php?op=r&cat_id=3&rev_… 2003-12-17 1:02 am We take for granted certain things on other platforms. Things like live window moving (not just an outline of the window), and a smooth, responsive UI. But this is new turf for the AmigaOne. This isn’t “old hat”, like it is in BeOS or MacOS X 10.3 or whatever.So, instead of yawning, like we couldn’t care less, let us see through the eyes of yesteryear… circa 1987 and after… when technological advances by democrews making fullscreen demos and processor speed going up from 8MHz to 16MHz… really made us go “Wow!”When we WEREN’T so jaded by multi-GigaHertz processors and tens of millions of polys per second video cards and 120 Gigabyte hard drives and multi-Gigabytes of DDR RAM…The Amiga is making a come back. The AmigaOne/AOS4 is a phoenix slowly rising from it’s ashes of long demise. Let us cheer it’s return, regardless how used to it’s “new features” we’ve become (and take for granted) on other platforms.The AmigaOne/AOS4 is a new platform. Let there be no mistake about that. And, while it may not be the fastest, nor the best computer on the market, it is still an Amiga… and that, in many cases, stands for more than just pure raw, processing performance and never-before-seen features. At least it should.Luposian 2003-12-17 1:21 am He’s probably not talking about the too-well-known VIA issues, as they have been fixed for ages in Linux already, but rather the issues with the ArticiaS, which we already discussed at length at an earlier thread here (forgotten already, have we?)…..and how on earth can you overhype something that can at any moment damage your data without your knowledge? No matter how little the risk may be in “normal everyday use” any sane person would consider it a great potential threat to datasafety… Trying to obscure the truth and sweeping it under the carpet however is a well-known tactic (IIRC it took VIA quite a while to admit that there was an actual bug, and issue drivers with workarounds).– CISC 2003-12-17 1:37 am Mike, I have never used nor seen an Amiga for myself, but I wish you the best of lucks.—- 2003-12-17 1:54 am LOL!!!That is stereotypical Amiga music, I just love it? It so classic. It reminds me of my A500 games.While the rest of the world is listening to Digitally Imported amigans listen to that music, is there a shoutcast stream for Amigans to listen to Amiga music? 2003-12-17 5:26 am No, I think the best possible respect to pay to the Amiga project (all platforms and so on are ultimately projects, created by teams of engineers) and the engineers that created the Amiga is to continue it, in the many forms it is now — Pegasos, AmigaOS 4, AROS, and so on. As a computer programmer, very little makes me happier to see my work continued by others. It is a form of respect, it means that, at the very least, your work was considered valuable enough to continue or improve upon.Erik 2003-12-17 9:06 am While the rest of the world is listening to Digitally Imported amigans listen to that music, is there a shoutcast stream for Amigans to listen to Amiga music?Nectarine. They play scene music generally, but there’s a lot of Amiga MOD’s. Can’t remember if it’s Shoutcast they are using, though…Link: http://www.scenemusic.net/ 2003-12-17 9:28 am @ UglyKidBillThanks. 😎@ Sigbjørn SkjæretThe high load incompatibility of the (x86 originating) Linux drivers with the ArticiaS’ ability to simultaneous access memory by the CPU and a PCI bus master has also been explained by non-AmigaOS4 involved developers like Bernd Meyer to be solvable through driver adjustments. The issues actually rarely occured with DMA enabled with the current boards and of course with DMA disabled there were no such problems at all.With overhyped I mean that the competition produced a seperate chip to partly address this issue and marketed this as if the problem was only solvable with this few cents worth chip, while FUDding Mai Logic/Eyetech’s competing solutions. Considering the competition was also dependent on Mai Logic’s ArticiaS norhtbridge this was not a good business decision (unsurprisingly the more major VIA bugs which had to be solved were never hyped like this) and provided Mai Logic enough reason not supply newer revision chips to this competitor and thus they had to move towards an alternative. (Which IMO is far from optimal considering this alternative currently cannot even reach AGPx1 performance levels with its AGP implementation.) Maybe this solution wasn’t all that neither (albeit we see not FUD spreading now ), as the product hasn’t been demonstrated in public running anything yet while it was said to be released months ago.A review of the AmigaOne motherboard undergoing some intensive testing can be read here:http://amigaworld.net/modules/features/index.php?op=r&cat_id=2&rev_… 2003-12-17 10:18 am AFAICR Bernd Meyer never said such a thing, he merely suggested that it was possible to circumvent most hardware issues by software, but at a price…FUDding? In what way was there any FUD (other than from you right now, and others before you), they were merely brave enough to go full out public about the issues, mind you this was *after* they had assisted Mai in pinpointing the problem, which they wouldn’t admit to being there, yet mysteriously there appeared a new revision which supposedly fixed these “non-existant” bugs (alas, it seems, not all).But I guess you are all too happy to FUD on your “side’s” behalf, like the sub-AGP 1x attempt there .. have you benchmarked the AGP performance on ArticiaS lately? Thought not…– CISC 2003-12-17 10:42 am >(unsurprisingly the more major VIA bugs which had to be solved were never hyped like this)Why would those problems have to be solved, as Pegasos uses different southbridge than they buggy (but easily fixable) VIA used on Teron boards? 2003-12-17 10:47 am @ Sigbjørn Skjæret> AFAICR Bernd Meyer never said such a thing, he merely> suggested that it was possible to circumvent most> hardware issues by software, but at a price…No Meyer’s posting ended with:“So with the right drivers performance shouldn’t be affected at all and it’s not an obscure trick…”> they were merely brave enough to go full out public> about the issuesNo, most modern hardware (complex) has some issues the drivers have to take into account. Bernd even uses classic AmigaOS as an example with regard to classic hardware.The “There’s no Mai without April” campaign was clearly of political nature.> mind you this was *after* they had assisted Mai in> pinpointing the problemNo, they have only recently pinpointed to the problem after the Linux driver issues were discovered by Mai Logic/Eyetech/Hyperion. They did not know this was an issue with the Linux drivers and therefor thought the VIA chips were to blame for all DMA issues. I don’t know if the competitor was even aware that the Linux drivers was the problem.> fixed these “non-existant” bugs (alas, it seems, not> all).Nobody claims there is any hardware ompletely without bugs or other issues. IMO that would be like living in a dreamworld as even the most tested identical specces complex devices (e.g. consoles, PDAs) have shown to have bugs for which tranparent (to the user) solutions need to be found (let alone mordern, ever changing PCs).Mai never said there would be no revisions or newer chips in the future. That would be like saying “AmigaOS/Windows/MacOS is perfect as so we will never have to address any bugs”.However at the time there were no major known bugs or issues with regard to Mai Logic’s ArticiaS Northbridge.> have you benchmarked the AGP performance on ArticiaS> lately?I have only seen the real life performance of the hardware (e.g. running demanding games), have you seen *anything* running with this competitor’s hardware? 2003-12-17 10:57 am @ J-P JokelaNeither the VIA 82C686B nor the VIA 8231 Southbridges are flawless. But there’s no need to FUD neither chip as can be seen in the PC world the issues can be solved transparently to the user. 2003-12-17 11:16 am @Mike Bouma>Neither the VIA 82C686B nor the VIA 8231 Southbridges are flawless.Different chip, different problems. You just made it shound like Pegasos had the same southbridge as Teron (aka A1) which it doesn’t. I wouldn’t call pointing out that “FUD”ok, I said it’s buggy (and easy to fix), but so did you. 2003-12-17 11:31 am @ J-P Jokela> You just made it shound like Pegasos had the same> southbridgeThat was not my intention. I wanted to point out that there have been and are various hardware issues with both solutions which needed (and need) to be solved and that the focuss on the DMA issues (mostly related to the VIA Southbridges) were being hyped due to political reasons.> ok, I said it’s buggy (and easy to fix)The fixes for solving the VIA southbridge problems are well documented and implemented in the PC world. However without proper access to this documentation this was not an easy fix.Both teams have very competent hardware specialists amongst their team. I don’t blame them for the political decision coming from higher hand. But IMO especially Mai Logic has more of them within their ranks and I believe this benefit will show off in the future. 😎IMO Mai Logic deserves much praise for making both ArticiaS solutions possible to exist today. 2003-12-17 11:38 am There is no Linux driver issue .. as discussed earlier here, the Linux drivers have been modified to circumvent corruption issues in the ArticiaS, and it does this by stalling the cache at various strategic (but not very logical) places .. the fact that the authors of the “fix” don’t really know exactly what it does, does not bode well…Ofcourse the “No Mai without April” was political .. they were stonewalled by Mai (and they were not the first) when trying to bring up the issues so that there eventually could be a chip without these very specific problems, even when they sent their lead technician over to Mai to demonstrate the issues to them they were very reluctant to fix them.There were no “major known bugs at the time” because Mai refused to acknowledge them .. they had already been informed several times of the issues by several of their customers!What so-called “demanding games” are these, Quake2? No games that run on an A1 at present date comes anywhere near saturating an AGP 1x bus (266MB/s) .. Quake2 f.ex. just need to upload a bunch of textures before each level, then just keep feeding the card with vertices, this takes barely any bandwidth at all…– CISC 2003-12-17 12:04 pm @ Sigbjørn Skjæret> Ofcourse the “No Mai without April” was political ..> they were stonewalled by MaiWith you being a developer for this rival platform I believe you would have been told something which suits the agenda of this rival company’s management.But just to give an example I know about claims made by employees that the relationship between Mai Logic and this company were just excellent, meanwhile in fact there have been several threats of costly and time consuming lawsuits made by the management against Mai Logic at the time (as shared with me by both teams, including the top management from the competitor). To me it’s pretty clear who has been the bully. So IMO it’s best to take this “information” with a bucket of salt, (just like the previous performance claims, corporate structure claims, IP ownership claims, etc). 😉> What so-called “demanding games” are these, Quake2?You could for instance try to multitask Quake2 and MOL running a game like Myth 2: Soulblighter. Demanding enough for you? 😉Seriously for as far as I can tell the AmigaOne hardware performs just as well as can be expected (e.g. running Linux on Mac or AmigaOne) based on its specifications. 2003-12-17 12:46 pm Just because I’m a developer doesn’t mean I get told propaganda fodder (why would I care?), however I’m quite capable of doing my own research and put two and two together .. and sometimes you have to be a bully to get things done (esp when people go out of their way to ignore you)…Ehm, what are you talking about? Quake2 puts next-to-no strain on the AGP bus, and last I checked MOL doesn’t even support 3D (not that it matters, as Myth2 is even less demanding than Q2), so all it would have to do is write its software rendered bitmap to the gfxcard (and since you claim multitasking, you’d have to have both in a window == even less data transferred), where’s the demanding part?– CISC 2003-12-17 1:28 pm Sadly it’s a little hard to understand as the microphone was nearer to the crowd than Alan…http://www.rycochet.com/However better quality recording may become available as Paul recorded the public part of the presentation as well, with the Microphone in Alan hands. 2003-12-17 1:36 pm @ Sigbjørn Skjæret> Just because I’m a developer doesn’t mean I get told> propaganda fodderNearly everyone involved I know has been, so your situation seems to be a rare exception.> and sometimes you have to be a bully to get things doneI don’t agree, bullying does not make you friends and should IMO only come as a last resort if you need to protect yourself.> Ehm, what are you talking about? Quake2 puts next-to-no> strain on the AGP busLike I pointed out this was not a serious example. With regard to wild claims IMO the burden of providing solid evidence is always on those who make those claims.To differentiate my mention of the slower than AGPx1 AGP bus speed is solely a fact based on the official specifications.> MOL doesn’t even support 3D (not that it matters, as> Myth2Myth2 runs fine on MOL. I am sure some more demanding games will see the light on the AmigaOne, until then your claims prove nothing. First priority is now to finish the AmigaOS4 project. 2003-12-17 2:21 pm Well, maybe I am a rare exception, maybe I’m not, maybe it’s all in your head, who knows… Yes, and this was exactly what it was, a last resort to protect themselves (why else would they be crazy enough to alienate their only southbridge chip supplier if it wasn’t evident they were getting nowhere?).The official specifications says nothing about being slower than 1x, that was your snide remark .. the official specs says rather confusingly AGPx1, but given what they have to work with it probably means 1x, however there’s no indications whatsoever of it being subpar .. however, the official specs for the ArticiaS say AGP 2x, but anyone who’s actually put that to the test knows that is far from the truth…Yes, I don’t doubt Myth2 runs fine on MOL, however, since MOL does not support 3D (unless something happened very recently (no mention of it on the homepage though)) you are most likely running it in software render mode .. either way it makes no difference, bandwidth usage is minimal…– CISC 2003-12-17 3:15 pm @ Sigbjørn Skjæret> maybe it’s all in your head, who knows… I never like it when discussions begin to become too personally tainted, so I will leave this as my last message in reply to you.Much of the propaganda is publicly available at forums as written by the management and some of the employees. So it’s not all just in my head.> The official specifications says nothing about being> slower than 1xI was talking about the Northbridge specifications, not what is advertised by this competitor.> the official specs says rather confusingly AGPx1<just kidding>Maybe they did this on purpose (?) and will later come up with the excuse: “We told you our hardware only comes supplied with one APG bus, why would you ever think we were referring to bandwidth?”.</just kidding> LOL CYA 2003-12-17 4:21 pm >No, they have only recently pinpointed to the problem >after the Linux driver issues were discovered by Mai>Logic/Eyetech/Hyperion.Mike. It is not a problem of the Linux drivers. What MAIare currently (kernel 2.4.22) trying in their ide driver isflushing the caches at every DMA access (“cache gives only a 2-3% improvement…”). However that stilldoesn’t solve the problems. They are currently searchingfor other workarounds but still haven’t found a workingone. That’s also one reason why there isn’t the Linuxkernel 2.4.23 for the AmigaOne yet.Also they have to fix the remaining other issues (noworking AGP/PCI GART etc.). 2003-12-17 4:26 pm @Mike Bouma>> The official specifications says nothing about being>> slower than 1x>>I was talking about the Northbridge specifications, not what is advertised by this competitor.I’m not familiar with this AGP standard “slower than 1x” care to explain more? I know only AGP1x, AGP2x, AGP4x and so on…But with most apps speed doesn’t matter, as long as it’s AGP2.0 (and if speed DID matter, 2x would likely bee too slow aswell) 2003-12-17 5:49 pm @ Anonymous> That’s also one reason why there isn’t the Linux> kernel 2.4.23 for the AmigaOne yet.If you really want to know about this I would recommend to ask the Hyperion team. Mai Logic hired Hyperion to port Linux to the AmigaOne platform. However it’s no secret that AmigaOS4 development currently deserves Hyperion’s top priority, so I imagine this may have something to do with this.Or ask Ross Vumbaca:http://amigaone-linux.sourceforge.net/@ J-P JokelaThe Marvell chipset only supports PCI-X mode and if an AGP slot is electrically implemented on a PCI-X bus this only runs with the 66 MHz PCI (not PCI-X) protocol and is therefore significantly slower than even a single speed AGP.I believe you are correct to note that the mininal assumed AGP bandwidth by graphics card manufacturers would be 1x.The ArticiaS supports 2x. 2003-12-17 6:28 pm >If you really want to know about this I would recommend to>ask the Hyperion team. Mai Logic hired Hyperion to port>Linux to the AmigaOne platform.The Linux adaptions are being done by MAI. Check thesources.>The Marvell chipset only supports PCI-X mode and if an AGP>slot is electrically implemented on a PCI-X bus this only>runs with the 66 MHz PCIThat’s the same implementation as the Articia does it.Checkout the technical documentation available at mai.com. 2003-12-17 11:39 pm That was not a personal remark (unlike yours), and ooops, I meant northbridge ofcourse, my bad…The ArticiaS does most certainly not support 2x, and their implementation of the AGP bus is exactly the same (except they don’t have the additional features to work with)...and Hyperion most certainly is not doing Linux (they did the UBoot port however)…– CISC 2003-12-18 12:12 am @ Anonymous> That’s the same implementation as the Articia does it.That’s incorrect the ArticiaS does support AGP 2x, unlike the used Marvell northbridge. I recommend reading the ArticiaS specifications a little better:http://mai.com/products/chipset.htmAlso the ArticiaSa and ArticiaP are in the pipeline, which will support AGP 4x.Maybe you are confused because the chip *also* supports a 66 Mhz PCI card. 🙂 2003-12-18 4:25 am Nope, I’m not confused at all .. has it ever occured to you that what’s printed on paper (or webpage) isn’t necessarily the truth, the whole truth, and nothing but the truth?Only Intel chips have “real AGP”, all the other chipmakers simply create something that works in a similar fashion using a overclocked PCI bus and some extra logic .. the gfx cards are nonethewiser, and thus the users as well…It can claim it supports AGP 2x all it wants, but when the *real* speeds are nowhere near even 1x it doesn’t do the users much good, now does it?– CISC 2003-12-18 4:30 am @ Mike Bouma> That’s incorrect the ArticiaS does support AGP 2xDoes AmigaONE with ArticiaS reach AGP 1x speed? How about AGP 2x?I’d say that answer to these questions would be the most convincing proof whether the chip actually performs as promised in the marketing material (which is the question here I might add: Does ArticiaS provide the performance MAI claims it does).Anyone upto some benchmarks? It should be trivial to check the framebuffer writespeed. 2003-12-18 1:11 pm >That’s incorrect the ArticiaS does support AGP 2xI wrote about the implementation amd that is exactly thesame. You can either use the 66MHz PCI 0 bus of theArticiaS as a PCI or as an AGP bus. The implementation isnot different at all.>Maybe you are confused because the chip *also* supports a>66 Mhz PCI cardNo you are confused with the bus design.Also note that Linux on ArticiaS based machines uses theAGP bus only as AGP 1x. I think nobody cared aboutit/noticed it so far.PS: Do you have an AmigaOne yourself? 2003-12-18 2:07 pm “The implementation is not different at all.”Well the bus protocolls and what they can request is certanly different.“Also note that Linux on ArticiaS based machines uses theAGP bus only as AGP 1x.”I do belive you are confusing the A1 with the Pegasos-I, the Articia falls back into AGP x1 if you clock the memory-bus or fsb under 133Mhz, which is the case with the Pegasos. The A1 can allso fall back into AGP x1 speeds if the 66Mhz PCI slot is used simulanously as the AGP slot. AGP “emulation” on the Articia is most likely acived using it’s internal crossbar achitecture(floating buffer), and thus while not fully adhering to the original architectural thoughts of the AGP bus, it’s still able to reach full AGP x2 speeds.The Peg-II, however when reading the above comments, looks like it’s forcing the graphics cards to use there PCI falback mode, with an glued on voltage regulator to make shure that it does not fry the AGP-cards. Which is what it’d do if they’d be using the standard PCI voltage settings. 2003-12-18 2:57 pm @ EZ“The A1 can allso fall back into AGP x1 speeds if the 66Mhz PCI slot is used simulanously as the AGP slot. AGP “emulation” on the Articia is most likely acived using it’s internal crossbar achitecture(floating buffer), and thus while not fully adhering to the original architectural thoughts of the AGP bus, it’s still able to reach full AGP x2 speeds.”The question is: Does A1 reach even AGP 1x speed? If not, then all this theoretical talk is pretty much in vain.Wouldn’t it be funny if Pegasos-II would have fast AGP? I guess we will see soon. 2003-12-18 3:45 pm >The A1 can allso fall back into AGP x1 speeds if the 66Mhz>PCI slot is used simulanously as the AGP slot.Try enabeling any feature such as AGP GART. The firstyou’ll see is a fall back into AGP x1 (also if there is noPCI card used in PCI bus 0!) on the A1. What’s the point inhaving “AGP x2” if it falls back to x1 if you try to enableany specific AGP feature on the A1? 2003-12-18 10:00 pm That might just as well be a BIOS issue you know… 2003-12-18 11:13 pm >That might just as well be a BIOS issue you know…AGP GART gets compiled into the Linux kernel. But then itmight just be yet again another Linux issue. Or anArticia ‘feature’ which hasn’t been implemented so far. Toobad.