what about the AMPRNET Roter at ucsd if someone do ping for that address will it pass it ? or ignore it ?
It would pass it. However, that restriction has nothing to do with how packets get routed through other networks. It's all about hosts on your local network being able to speak to each other.
The subnet mask you use tells your hosts if the IP address on an outbound packet is on your local subnet and should be discovered at layer 2 with ARP, or if it's on a different network and needs to be forwarded through a router (typically a default gateway) to get to its destination. It all has to do with binary math. An IP address is made up of 32 bits and the netmask tells your hosts where the bits for defining your network stop and the bits for identifying your specific hosts begin. In your case, your /29 means the first 29 bits define your network address and the remaining 3 bits are used for your hosts. Three bits can represent eight different values. The problem is, the highest host address (where all the host bits are 1s instead of 0s) is reserved in the protocol as a broadcast address and any packet addressed to it from your local network would get sent to every other host on your local network. A host address with all 0 bits is reserved in the protocol for identifying the network itself and cannot be used to identify a host. This leaves only six usable host address which also need to be used for any routers on the network such as your default gateway.
If you have more than five hosts and one router, the next larger network you could get is a /28 which uses four host bits. Four bits can represent 16 different values which would give you 14 usable host addresses.