i just watched "be cool" - it's a silly film, with a lot going on and some nice touches you have to watch out for. just the kind of thing i like. reminds me of "the player" - self-deprecating of hollywood and yet clever, funny and complex at the same time.
anyway - i haven't written for a while because i've been concentrating on putting a 100% free software compatible ARM netbook deal together. the chain goes something like this:
* lots of low-cost decent computer designs get made by manufacturers, such as the pegatron ARM netbook over a year ago, the lenovo skylight and many more.
* retailers look at them, love the lower price but then freak out because the machines can't run windows.
* individuals call the manufacturers and get blown off because the manufacturers won't speak to anyone who isn't buying 50,000 units. the reason for that is that the cost of producing the plastic case alone is $100,000 for the moulding; the engineering time and other materials is let's say about another $150,000. thus, if you're selling machines at $120 a time and only making 5%, you _need_ to sell at least 41,600 machines just to break even.
so, there's this huge gap between the manufacturers and the people who actually understand that ARM or MIPS processors are perfectly capable of running a decent OS - free software folks.
in those "free software folks" i would include anyone who's been running ubuntu on their laptop for more than eight months and has decided that they really don't need to dual-boot their intel laptop any more, because they just don't use anything on the windows partition any more - it's just taking up space.
so their next laptop or desktop machine might as well be one of these ARM or MIPS machines, because, well, sure it's a brave move, but that $150 desktop price or that $200 netbook price suure looks compelling...
... but wait, where are the online stores selling ARM netbooks? and what retailers are selling them?
absolutely none.
so - here's what i'm doing: i'm setting up an open-pandora-esque survey, which in fledgeling form there's a pre-run here, discussions underway.
if you've seen the openpandora, its success isn't so much the actual machine; the success is the fact that it was made on-demand, after the online games console shop owners decided that they'd had enough of the mass-market console makers not making something that they really wanted, asked if anyone would be interested in buying something really good _if_ it was available, and then financing its construction purely through pre-orders and deposits!
in the "intel / amd" realm, we see a lot of machines, there is a massive range and plenty of choice - except regarding the processor type. except... the processor is a non-integrated monkey-eating power hog that requires support infrastructure, massive communications bandwidth to the peripherals (video, SATA) i mean it's crazy. have you _seen_ the prices on PCI-e chips, PCI-e to PCI chips, SATA controllers and ... yes, you know the prices on Video cards - it's madness.
one integrated CPU such as the freescale iMX515 or the samsung S5PC110 can do all that: 3D OpenGL ES 2.0 graphics; ATA-6 IDE controller; DDR2 memory support; High-speed USB2; 2 LCD panels; HDMI Digital Video out; 10/100 Ethernet - the list of integrated connectivity just goes on and on - and it's on one chip, costing between $18 and $25 depending on clock speed.
so i've had enough, and i'm creating a survey site, with two phases:
* phase 1 - the "what do you want and can't get and really want to pay for" phase. this will be the key bit: collect information about what people want, keep them informed about what other people also want, so they can group together and get better bulk purchasing power.
* phase 2 is the "ok the specs that the most people want and are happy with the final cost, we now go look for it or we just get it made" phase.
my partner, marie, she doesn't do computers, she just uses them for the communications aspect. i was talking to her this morning, and we kinda put two and two together. we realised that whilst free software has been successful by continuing to ride on the back of the reduction in component pricing thanks to windows being such a resource-hog, free software won't be truly successful until it breaks free of that.
in other words, it's time that computers were designed which had free software in mind from the outset, rather than being bought and having the proprietary OS removed.
and with the cost of the components that are _actually_ needed now being so low, it's the perfect moment.
answer: make a survey, find out what people want, then make the damn thing.
update - SATA
someone on the gpx32 forum asked about having an SATA controller. so i looked up the prices, and there don't exist memory-addressable SATA controllers. so, i looked up memory-addressable PCI-e PHY chips: $7 from TI. then, i looked up PCIe-to-PCI bridge chips: $9 from TI. then i looked up PCI-to-SATA PHY chips: couldn't find any from TI, looked around, still couldn't find any, found something called the SIL3134 and then my number-sense kicked in and went "$16 for just the ...." and went "fuckit".
then, after michelle on arm-netbooks said "why not do PATA using a general-purpose microcontroller, or use a TI DSP with SATA on it?" i went "hmmm..." and looked up the latest ARM9 CPUs that TI have: there's an AM1808 for just under $9 in 1k volumes, runs at 375mhz, and has a 3gbit/sec SATA interface _and_ an HPI (Host Parallel Interface) where you can hook it up and do DMA transfers between the host (at the host's clock rate) and the AM1808.
now _that_ would be wicked, especially because the AM1808 has high-speed USB-OTG, 10/100 ethernet, two more SD/MMC cards, on-board RAM (128k) and on-board ROM (32k). you could write a tiny OS and a matching linux driver to control the peripherals.
i haven't found _anything_ comparable in price that has such a stonking level of integrated modern peripherals, at such a low price-point. TI is really onto a winner, here.