On Wed, Aug 10, 2011 at 6:54 PM, Phil Endecott <[email protected]> wrote: > Hi Luke,
wotcha mr endecott > This sort of question should always be considered in the context of the lead > time, i.e. "in 9 months time when TI and NVidia and Freescale (etc.) have > announced their next generation chips and boards, would you still be > interested > in a board with an 800 MHz Coretex A9, 512 MB RAM etc.?". hell no... unless it was either a) available in 4-6 weeks b) was part of a kit that could be upgraded c) was price-reduced massively to reflect the lower amount of RAM and CPU speed. a) _could_ happen b) could also happen (see links below for examples) c) ain't gonna happen, fact. links to stuff wot is modularly upgradeably spiffing: * DM3730 http://www.kitarm.com/news/330-kitarm-release-tablet-solution-with-omap3-dm3730-cortex-a8-dsp-dual-core.html * Samsung Enyxos http://www.hardkernel.com/renewal_2011/products/prdt_info.php?g_code=G129705564426 * Samsung S5PC110 http://www.hardkernel.com/renewal_2011/products/prdt_info.php?g_code=G129690436790 * etc > Looking at the various things I have done over the last few years: > > - I have had a couple of "NFS server"-like projects, where respectable network > and disk I/O speeds have been most important. (I can't agree with Gordan > Bobic's claim that SD cards are good enough for most applications.) Your > proposed board doesn't quite suit that because it doesn't have gigabit > ethernet. whoops :) > - I have done a couple of "display" applications, where video output in > various > formats and respectable software support for that has been most important. > Today, almost everyone (including this AML chip with Mali graphics) is in the > same boat: there is good hardware and good drivers, but the drivers are all > non-free. If someone wants to stand out in that market they know what they > have > to do. yes. stop censoring people on forum.arm.com who point this out, for a start. (yes, ARM employees who are on this list: your colleagues responsible for forum.arm.com are censoring people who call for the MediaTek-licensed MALI hardware to be openly documented just like AMD / ATI are doing). > - I have done a couple of commercial projects where a fast interface to some > custom (e.g. FPGA) logic was most important. I think PCIe is probably the > only > realistic choice for that today. The AML chip doesn't help for this. it doesn't. so.. let's hope AML's next version has that... > So to be honest, I would not expect a board based on that chip to be much use > to > me. But I know I'm far from typical of anything... well, the point is to find something that covers the needs of the people who actually will write the software. duh. marvell tried (and are behind the times, restrict things to NDAs and so on). TI bless 'em keep trying (but focus on mass-volume hand-helds, which don't obviously have PCI-e or gigabit ethernet!) Freescale try (but always seem to be 18 months behind) Samsung go for mass-market and won't speak to mere mortals... it really is quite ridiculous that the very people who could help these bloody companies to lessen the absolutely critical burden of development (software) are excluded because these CPU manufacturers aren't creating SoCs that are of interest to us! madness... l. -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected] Archive: http://lists.debian.org/capweedytfeb3ximghw3hfcbfhiwg44a88rgmrdboynmhtjv...@mail.gmail.com
