Intel Is Working On A Linux Networking Stack For Small Systems

[doom_Oo7]

Well, you could want to put an entire Linux in your BIOS chip I have yet to see one that is > 4mb.

[0xBADCODE]

So quark suxx more than small ARMs and MIPSes?

Normally even small low power ARM or MIPS based device can cope with full network stack. Seriously, even watches and pocket devices are now at least 32Mb RAM or more. More than enough to cope with usual TCP/IP stack and much more than that. Interestingly, it looks like when x86 being optimized for power consumption, is suxx so hard that it losing to ARM and MIPS, even in terms of performance.

And even $15-20 (this is complete module! Do you hear me, Intel?) Atheros 9xxx module would come with at least 32Mb RAM. LWIP haves at least some point (TCP/IP for boot loaders like u-boot). This idiocy from Intel just absurd.

[chithanh]

The competition for Quark are not those devices. I think Intel targets the market which is currently occupied by Cortex M0-M4, and wants to use Linux on their chips.

[LtWorf]

If you're making 2MB/4MB machines in this day and age, you're doing something wrong.

Wirleless sensor networks can be deployed in forest or volcanoes and you don't want to go changing batteries every day, but you'd rather have them run for months.

And RAM runs on batteries, the more RAM, the more power it requires.

Check Telosb nodes.

[wizard69]

Huh.

Their reference design "galileo" specs 256 MB DRAM.
Given the cost of RAM these days, anything less than that and the vendor is just being SO absurdly cheap that they aren't worth dealing with.

... not that it isn't a good idea to chop off unnecessary cruft, but this is getting slightly ridiculous.

https://communities.intel.com/docs/DOC-21835

This is the first thing that came to my mind. It looks like Intel is trying to compete against microcontrollers with Quark instead of ARM SoC descended for the cell phone world. I've seen nothing rational about Quarks place in the market.

[wizard69]

There seems to be lot developers interested in minimal networking stack, so maybe after some changes it pass

The only way I see that happening is if it delivers the same functionality as the current driver. While improvements are always possible, I'm just not convinced of the value of this project. Nor am I convinced of the value of these minimalist Linux installs. If you want compact there are all sorts of solutions in the embedded world.

[wizard69]

I wouldn't bet on it. Some quotes from the mailing list

Making 2MB RAM machines today makes no sense at all.

I'd qualify that more to say it makes no sense to run Linux on a machine with 2MB of RAM. If this is an authorized project at Intel, I see it as a sign that they have lost their minds and are massively out of touch with the rest of the world.

[wizard69]

The competition for Quark are not those devices. I think Intel targets the market which is currently occupied by Cortex M0-M4, and wants to use Linux on their chips.

Most rational people would not use Linux on those Cortex chips. If they used any sort of operating system at all it would be a compact embedded OS most likely with realtime capabilities. The reality is that the Linux kernel itself is way to large for these sorts of machines.

[doom_Oo7]

Most rational people would not use Linux on those Cortex chips. If they used any sort of operating system at all it would be a compact embedded OS most likely with realtime capabilities. The reality is that the Linux kernel itself is way to large for these sorts of machines.

Well of course if you put every and any driver and FS and scheduler... Fortunately make menuconfig is here!

[markc]

A bit late to comment but if a network enabled lightbulb or door latch could work with linux in 4Mb of ram, rather than some proprietary realtime OS, then I would be more than happy for Intel devs to pursue this and I hope Andi succeeds in getting patches into the mainline kernel.