I am having some issues with an x86 system running 1200 simple queues, 100 firewall rules, an 100 NAT rules.
It is a 8 core i7 with 16GB of ram with 6 gigE cards. Is there a premade routerboard capable of handling this load as opposed to an x86 system that people use? We will be growing to 2000-3000 simple queues in the future and we need something that will work well.
The specific issue we are having is that during high load the queues start to act up and only give 20% of the set values even though our fiber is not maxed at all and pings go through the roof.
We are running 5.6 on the router because it is an x86 system and I'm worried upgrading the router may be fatal for the system as there are reported bugs for different x86 systems. Could this be related to the firmware as opposed to hardware? If there is a routerboard that is capable of handling this we would look to swap it out but I think we'll need to keep a high end x86 from our experience.
Any input would be great.

You should probably have a look at the CCR specs and compare then to your x86 solution. e.g. :

http://routerboard.com/CCR1036-12G-4S

Change the kind of your queue type.
The default type of simple queue is 'default-small' (Kind: pfifo), set it (Kind ) to SFQ (preturb=5, allot=1514)
I think this will solve your problem.

Simple Queue on anything less than version 6 can create up to three queues each! You have a massive number of rules! You need to stop using simple queues and move to queue trees. They are not that much harder to learn since you seem to have a good grasp of your network usage.