I've read this message 3 or 4 times now, and I can't understand most of it. Probably english isn't your native language, but then ask someone to assist you in at least writing your issue down so others can actually decipher it. Or create a network diagram / overview and post it here.
If it's about speed, run iPerf with -P4 for parrallel mode. A single thread will not reach 10Gbps on regular X64 hardware, as no offloading is present. A dedicated ASIC like in a switch or 'real' router, can do it during breakfast. But a general purpose X86/64 CPU can't, and will usually be limited between 2 and 3.5Gbps, 4Gbps if you have some good single thread performance.
We run pfSense 2.6 CE on pretty beefy Xeon 6248R CPU's and another pair on Xeon 5118 CPU's. The 6248R does about 3.5-4 Gbps, the more common 5118 between 2.5 and 3.5. That is single connection though. If I do multiple connections I easiliy saturate bandwidth.