This is not unusual, particularly with certain older model CPUs, such as Clovertown, where you might approach full threaded performance of which the memory system is capable even without optimization. It's more likely to occur with vectorization, which you have prevented by setting -O1. If you improve the single thread performance, as you should, with optimization, you are unlikely to get a proportional gain in threaded performance.
As long as threading still improves performance, you have a reasonable case to claim to have met expectations.
Combining optimization with multi-threading does make the effect of good affinity settings (KMP_AFFINITY environment variable) more visible, along with organizing data so as to make affinity more effective.
