If the cache is too small then you're likely to see pages flushed from cache (to disk) and then re-read from disk as needed; net result is that a too-small cache would see an appreciable volume of physical disk reads from the tempdb devices (ie, re-reading pages into cache); these disk reads should also show up in sp_sysmon as reads from your tempdb devices.
Keep in mind that if your tempdb devices have good service times (especially if pages are cached by the OS and/or SAN), some volume of re-reads from disk may be ok from a performance perspective, ymmv.
The primary objective is to size tempdb large enough to eliminate the vast majority of physical re-reads for the majority of your users.
---------------
As for wash sizing, I'd be watching for stalls in the pool (eg, dirty pages in the wash area aren't flushed to disk and marked as 'clean' by the time they reach the end of the queue thus causing some delays as the cache has to be scoured for 'clean' pages and/or a delay while waiting for disk writes to complete - so that 'clean' pages can be made accessible for new object allocation requiremetns).
Wash sizing, and possible stalls, applies to all pools ... including pools assigned to the log. Stalls can be found in monCachePool or sp_sysmon (cache stats:pool turnover:buffers grabbed dirty)