No, there is no effective way of caching units in F@H. With F@H, successive protein releases are based upon results received from current units. That is why there are deadlines.
Once a unit is issued, the clock starts, if it is not completed in time, then it has to be re-issued to someone else, because future work units will depend on its results. The clock runs even if the unit were sitting in a cache.
Projects that allow caching are dFold, Seti, D2OL and TSC. dFold caches results, rather then future work units.