On 5/24/05, Dennis SCP <[EMAIL PROTECTED]> wrote: > I must say I'm rather disappointed after reading the announcement of > the parting with the hardware manufacturer. I had forgotten all about > the negative comments on slashdot and googled the project after > thinking about how much I wanted it. I got really exited when reading > other discussions. But I guess that after evaluating the online poll > (less then 10 000 people in the close knit OSS society could be > bothered to vow their support, even if over 60% of them actually buy it > we'll have a big loss on our hands) they called it a day.
I too was disappointed by the turn-out at the poll. But experience has shown that people often cannot be bothered to do this sort of thing. You'll recall the recent article about a Linux-only HDTV-in card. The designers polled a California users' group, and only 2% responded positively. They expected to sell 2000 cards a year (IIRC), but when they produced it, they sold 2000 in the first 3 days. > > I think this really underestimates the potential market. I knew I > wanted mythtv and bought the hardware everybody talked about, no not > the Haupauge pvr 250 as that one is almost unavailable here in Europe, > I bought the more expensive haupauge pvr 350, more expensive but > everybody says its the most compatible. And I did not pledge to buy it > on beforehand in any mythtv forum. I have not seen or signed your ogp > poll but had decided some days ago to wait for june and order one at > 200 euro. But in my excitement I wanted to be amazed some more and read > the mailing list and found sad news. Well, that's what happens when projects get dropped. Those who actually believe in it have to do it themselves, on a slower schedule. > So what is my profile as a guaranteed buyer? > I wanted Linux / KDE / Mythtv > I know they look great but I heard they can be a bitch to configure > hardware wise. > So I buy what the community says is well supported for safety. > I was forced to buy an fx5200 for almost 100 euro as, oops, nvidia > nforce does not mean that minimal VGA support is always integrated. > Now, I read about an fx5200 like video card codenamed ogp with out of > the box linux support and a bright future of support and likely some > amazing tweaks for just double the money. Yeah I'd buy that, we'd all > buy that; a great future protecting against obsolesce for less then a > high end video card. > > But now I read it'll cost 5 times a minimal card, I can never defend > such an expense. Hello, Bebox / Amiga / research fan sales, that does > not sound like a large target market. And for us followers there will > be a static card later on, that is popping the dream of a bright future > of endless possibilities. The $500 is not for a graphics card. It's for a hobbyist card that is the development platform for the OGP graphics card. > What do I want instead? > A cool name and a dream backed with some facts for a discardable > premium. > > The name > Freeforce X11 sounds like a proper graphic card name and has many winks > to its OSS market, although you might be scared of a nvidia attack over > the nforce name now that you're on your own. > > Freeforge X11 I guess would go well with the OSS crowd but marketing > wise forge is not a beginners class English word, making it harder to > remember for international consumers and has some resemblance to the > negative word forgery. It is best not to be marketed as the open source > community ripping of capitalist industrial designs and laughing them in > the face with it. > > Freefox X11, sounds as fast as name one and has OSS written all over > it. I like it. > > The dream > A Dual-DVI port is one part of the dream. Maybe inkjet printing > displays or OLED will cause a revolution or my income will increase and > I can afford something like Apple's biggest screen, resolution wise > this card can support it. > > FPGA is another part of the dream. First of all it allows buying > without thinking as any bugs wil later be fixed or possible features to > be expected from such card will eventually be supported by the army of > OSS. It allows for some wild dreams like small optimisations to allow > speeding up SVG display or even live theora encoding / decoding. Quite > funny that both projects called each other vaporware because of a lack > of endusers while both have made great progress and are going strong. There's only so much logic that you can fit in the chip, and even with SVG support in the hardware, somehow, you've got to expose that to applications through an X11 extension, and THEN you've got to get the applications to support that extension. Do you understand the magnitude of the infrastructure that would need to be developed here? > > VRAM expansion is another way to get linux wannabe power users excited. > I do not use my 768MB RAM but I think I may use it and that it may keep > my computer from slowing down because of swapping issues. A VRAM DDR > slot will help sales and not just to uncertain / bet on the safe side > buyers like me or CAD / fluids emulation researchers but also for > desktop linux users on outdated hardware. How much expense do you want to add to this thing? It would cost as much to produce a card with ZERO memory and slots than it would to just put RAM directly on the board. And then you'd get poorer performance, because we'd have to use more conservative memory timing numbers. You ask for expandable memory, but do you know the consequences of that? > > You know Apple is doing a OpenGL desktop already and linux has some > wobbly experiments out, but did you know that Trolltech hired someone > specifically to make this happen modularly with X11 for KDE (and other > OSS of course)? > > It will be great if a DDR slot will allow for a 1152 MB introduction to > catch everybody's attention, even though the 1024 MB would not be a > default install and it can't help to gain any fps in any game as KDE 4 > will have other uses for that memory, especially on an older PCI based > machine. Imagine KPDF having the rendered pages of a hundred page pdf > document cached in VRAM; that would mean fast scrolling on a 200MHz PCI > machine while PDF's are considered slow even on a GHz machine. Image > loading the entire content of your 1024 MB SD photocard straight to > VRAM and scrolling through 7 megapixel photos at blazing speed while > your CPU remains mostly idle, I know there is OSS code to decode > jpeg2000 on the GPU so this is not far fetched. Imagine KDE running as > fast as the feature reduced lite window-managers on slow hardware, or > seeming even faster as obscured windows remain cached in VRAM because > it uses OpenGL window composing. You want to put over a gigabyte of memory on this thing, but you complain about the cost of the rest of the card? Think of the graphics memory as a sort of cache for textures and other images. As with all caches, you reach diminishing returns as you add more memory. You ask for lots of things, but you don't seem to be considering the design consequences. Do you want this thing to take two extra years and cost a lot more? > Discardable premium > Sorry for being blunt but you seem to have been run over by > enthousiasts with little money but lots of time and rich professionals > looking for a cool hobby. Therefore you may feel that the real backers > are willing to pay more and the bigger Linux market can only be > captured with bargain prices. But I think the sweetspot remains just > below the 200 mark. The active individuals who contacted you may not be > the lion share of the market, I think the lion share are like me; > hardware followers willing to pay a small premium to stop the > configuration hurting, especially if it looks like unlimited free > software upgrades in the future. What can we sell for $200? > That is why I think an expandable customizable fpga based card at 199 > is the sweetspot. The 199 euro would be including the 20% consumer tax > common in Europe, so in the US an out of state order could also be 199 > dollar. 10 000 units should be easy, we are followers we want what > works, what other people praise about, what they use so we can install > without typing any configuration codes. You seem to forget that the original FPGA-based design suffered from two major problems: (1) Not enough logic area in the FPGA (2) Zero profit We can't sustain a business on zero profit. We need our income to exceed our expenses so that we can invest in future R&D. > > So what about the cheaper ASIC based videocard? > It sounds a lot less revolutionary, I don't think I'm gonna replace my > 5200 with it because they both can display linux and play tuxracer. > Firstly I'm looking forward to seeing optimizations in the fpga. Having > a custom layout optimized for Theora HDTV coding sounds great for > really free mythtv although perhaps a bit unrealistic even if some-one > has already done theora fpga for themselves. Upgrading educational > Windows boxes to KDE Linux may result in a compromise fpga layout to > get the most out of such slow machines. So how can you decide what the > real buyers want set in stone anyway? Good point, so I suggest polling > the real buyers. Generate a random second part of the serial number for > each produced fpga card and let users tell you which fpga design they > prefer using in the card, you might be surprised. > We have been working a very long time to try to develop something economically viable. We want to develop a company here that can continue to produce hardware for open source in the future. As we add features and reduce the sale price, it eventually becomes impossible to sustain. The economic and technical effects of every decision must be carefully weighed. > Future proof > Completely reprogrammable graphics chip. > Ready for optimizations by original designers and dedicated hobbyists. > Ready for complete customizations; rush through special scientific or > multimedia tasks like only a custom chip can. > Simply download different chip layouts from the internet and (re)load > them without reseting the computer. > _______________ > > > Thanks for all your hard work so far, > Dennis > > P.S. No, I don't care about the RTL (because of the complete > documentation effort). But promising to release the RTL as GPL after 2, > 3 or 5 years depending on the sales sounds as a great idea to my ears, > especially as businesses have written off this equipment by then. > Thank you. _______________________________________________ Open-graphics mailing list [email protected] http://lists.duskglow.com/mailman/listinfo/open-graphics List service provided by Duskglow Consulting, LLC (www.duskglow.com)
