First, this isn't an apples-to-apples comparison. The 1600 is an ADC, not a switch, and this model does not include hardware acceleration. The switch is just that - an L2/3 switch with no need to process each packet in software.
The cause of the reduced throughput is probably latency inserted by the LTM. The 1600 doesn't have an ePVA, so all traffic is handled in software. Under ideal conditions this will delay each packet in both directions by ~20-29 ms latency. The switch has no such requirement and is most likely forwarding each packet at nearly line-rate.
When testing throughput, overcome latency with concurrency - increase the number of traffic streams. I expect that, unless the CPU reaches peak usage, you'll see full throughput with only a few concurrent streams.
Nimbostratus
