I've been there a while noting that some of my routers interfaces is presenting rx drop, but this is never presented any abnormality in the network, but yesterday I got a router to have several rx drops, so I was unable to solve the problem when I took the negotiation auto interface , it still happens the rx drops, anyone had any similar problem?

I'm guessing packets are being received faster than the device/router can process them. 337 Mbps / 50,000+ pps is nothing to sneeze at, especially if you have firewall rules. Also do you have any queing going on?

What type of device/processor is it?

If the RX dropped is a very small percentage of the total packets received (i.e. you have 1 million dropped packets out of 1 trillion received), you probably don't need to worry about it, though 0 would be nice

Otherwise, you can try to simplify or optimize firewall rules, or look at faster or multi-core processors/busses. (along with newer routeros versions that utilize multiple cores better, if you aren't already). You could also consider moving the firewall rules elsewhere (assuming this is the issue).

If you have gigabit in, and 100Bt outs, it may just be getting backed up occasionally (you could modify queues on the outbound interfaces to hold more if this is the case). (Although, as I think about it, that would probably show up as a tx drop on the outbound interface instead of a rx drop on the inbound interface).

my router and a Dell Power Edge 2x quad core x86 Opteron 1.7 ghz,
3 Giga ethernet intel network. I have only one firewall rule for NAT. and no control
bandwidth on this interface.
processing it is at most 15% of use, 1700 free memory.

If there is an instantaneous demand for more than 100% (or 400% depending on how you look at it) cpu it could be a momentary problem that wouldn't necessarily show up as high cpu usage in general (perhaps the interface was operating at 400,000 packets a second, for 100th of a second).

Is routeros configured to actually use all the CPUs? (/system resource pr, results in cpu-count: 4?)

I suspect the problem might go away once routeros gets the linux kernel that has the steering in it, if this is the reason.

Admittedly, I'm mostly guessing right now.

In the wild-guess department, perhaps some packets are being received with a vlan tag or a broadcast address that the router is not prepared to deal with, but I'm not sure if those would show up as rx-drops.

What percentage of the total is the rx drops working out to be?