Open-PC2 brainstorming
From Wiki.open-pc.com
Contents |
Ideas for the next version of the Open-PC
hardware wishes — aka «Do us the favor of choosing the best Linux-compatible hardware»
Piccolino81: Integration of WLAN (because this seems to be quite difficult with many chipsets not fully compatible to Linux)
Dingo: Probably difficult but better graphics support is a must. Like native support for 1080i video as in Via hardware now available.
Glococo: Integrated LowCost/LowCompsumtion 3D card for run at least compiz smoothly. (Ubuntu)
TMarques:
Capable hardware with open drivers would be either AMD or the new Intel LGA 1156 processors with integrated GPU. Choosing AMD would provide a good start to get Coreboot on it. The openPC needs a digital video output or the ability to add a graphics card through PCI-Express. ASRock isn't a reliable hardware manufacturer IHMO, Gigabyte is worth the extra €€€.
Branding
Markus: A version with Green-Hardware - Focused on "Green", not on "Speed".
ebo: I agree with Markus, but there also needs to be some education regarding how much power one really needs for common applications of email, document preparation, surfing the net, compared to running particle simulations and high end gaming.
Online support is key. Partner with a strong Cloud services company (or build your own) and sell package as a "subscription to the world" instead of a "just hardware." As part of the subscription, include a series of link libraries preset for target audiences, for example education or vocational trades, home business, entertainment etc.
Include link library to screencasts and help for system usage and troubleshooting. Move to differentiate Open PC Help from all other market systems.
TMarques: Better looking wallpaper and sticker on the case? Something that fits the KDE default theme.
Coreboot
ansobegues: Coreboot instead of privative Bios and Diversification please!, low energy (and lowcost) Pc and high performance Pc would be enough.
merethan: Ditching the proprietary BIOS would be nice. Those are still descendants of the assembler-written, reverse-engineered IBM BIOS that shipped with the first IBM-PC clones. They are still 16bit code, with the exception of new stuff like TPM chip support, USB support and ACPI and alike. This constant switching of modes (64bit <=> 32bit <=> 16bit) wastes precious CPU cycles. But that is not the worst part. The ridiculously long boot time is what affects users the most. Coreboot on the contrary, when configured properly, boots it's payload just moments after you turn on your machine. It doesn't corrupt RAM (today's OSes don't use anything below 64kb because Phoenix BIOSes are well known for corrupting it) and above all, it's free and open source. Written in C, it can be ported to other architectures as well. Enough reason to invest some time into this BIOS thing. It really gives us a unique selling point.
ebo: I concur. ditch the proprietary BIOS.
Distro preferences
Use Linux Mint 8 KDE
BrunolinuxF1: Use a fully free software GNU/Linux distribution like gNewSense.
ebo: not exactly a distro preference, but related -- have a collection of EASILY accessible .config files and configuration help similar to http://kernel-seeds.org.
ebo: some real-time kernel support.
konman: openSuSE KDE or Ubuntu Gnome if you want to go with one default distribution. I'd rather specify the os of my choosing at checkout.
Ideas for the not so near future
EBo: access to limited number of GPIO lines for real-time device control.
merethan: What I personally look for is alternative architectures. This is unlikely to be feasible on the short term, but somewhere in the future we could, for example, produce the first ARM desktop since some time. Even more in the future, perhaps opencores.com has an usable chip design. They already do good with their S1 Core, which is a reduced version of the SUN OpenSPARC T1.
Enough ideas to produce something somewhat unique. Open-PC is a cool project. Everything starts small. Today we do somewhat common things when designing a computer (off the shelf hardware etc), perhaps we can do even more cool things in the future. /merethan
aNordal: FPGA card as graphic, audio and/or maybe even network interface. The idea is that since we (the open community) never get hardware exactly as we want it (not to mention their drivers), we need reprogrammable hardware. Also, the sheer all-purpose data crunching power of an FPGA is a nice complement to sequential processors (CPU, GPU, etc) on a personal computer. Imagine the possibilities, let's enjoy open hardware! On the downside, I think this is way into the future. Just google "FPGA PCI", and you will find that boards made to interoperate with PCs are merely toys, while serious FPGAs are standalone computers.
Vincent: I strongly agree with aNordal. I support the idea of a programmable HW (e.g. FPGA). I am for it and willing to help. How shell we begin? what is the best already implemented example of this kind?
TMarques: You could look into Simply RISC's S1 core. It's RISC but should be fast even synthesized.
