compatibility between AMD and Router OS

Hi Fellows , there is any incompatibility between AMD processer and Router OS?

Thanks

Anibal

The spec sheet says

advanced 4th generation (core frequency 100MHz or more), 5th generation (Intel Pentium, Cyrix 6X86, > AMD K5 or comparable> ) or newer uniprocessor Intel IA-32 (i386) compatible (multiple processors are not supported)

and I have run RouterOS on an AMD system some time ago (test equipment) with no problems…

Thanks Christian

me to..
one server is duron(with defect mmx part-does not show correctly pictures in windows), other celeron 300-500 (lx and bx)
:slight_smile:

I’d like to see 64bit cpu support soon, as even Via will be releasing a 64bit low-power processor for their mITX platforms first half of next year.

you can run the RouterOS on a AMD 64, but it will not use the extra features and power. a Router doesn’t need this kind of power, we have seen over 2 Gb troughput on Xeons, I can’t imagine what could one do with all the power of a 64 bit system (on a router). Also this would require a complete rewrite of parts of software so this is not likely to happen anytime soon.

I can think of some uses of routers with multiple-Gigabit throughput.

My company has some customers who want to build city-wide mesh networks with wireless (FSO/MW) 100Mbps spokes and Gigabit ethernet backbones/rings or meshes. The network needs to instantly self-heal if any wireless links break (or are blocked).
The cost point of ethernet is much more attractive than traditional telecom Sonet/SDH design, and the concept of “self healing distributed mesh” attractive (a mesh of true p-p links, not the WiFi mesh concept).
Some of the needs for high traffic are professional CCTV at up to 5Mbps per camera, broadcast television streaming, E1-over-ethernet (2Mbps), broadband wireless base stations (up to 70Mbps per sector) and/or corporate tunneled LAN-LAN connections and so on. In some areas there is very high traffic - for example at a CCTV control room all the video is aggregated, then displayed and recorded - in a moderate/big network that is more than 1Gbps. (very different than rural broadband, where you don’t need such traffic).
Now we know we can use L3 routing switches to do some of this, or at a pinch, L2 switches with rapid spanning tree to self-heal. But as a “distributed mesh” it doesn’t scale well.
What would be really nice would be a compact MT solution, with 2 or 3 Gigabit interfaces plus several 10/100 interfaces to put at the major node points. If the CPU is low-enough-power to make the box outdoor-capable, so much the better.

Any thoughts?

2gbps is nice, but when you are doing shaping, mangling, inspecting for p2p packets, then the power is nedeed. I’m solving some reports from our customers with about 500 clients, that their routers are on the top with cpu utilization… Are there some hints for optimalization of cpu time consumption?

2.9 will be more improved in this field, and it will take up less resources for queuing etc.