Stat caps do seem to be based on weapon DPS, which IMO makes garuda's second best kinda just regardless for all DOW. DPS seems to impact WS damage while weapon damage is the factor in auto-attack. The higher the DPS the higher the stat cap. This is partially what makes relics so strong--they're all both the highest DPS and highest DMG weapon of their class. This means you can get gains of further and further stat increases.

Paladin is kinda special though, since its enmity modifier abilities are based specifically heavy on one stat (mind), rather than both. So I would say a warrior tank should chase capping both STR and VIT while a pld tank will see minimal threat gains with str, while capping MND and stacking enmity will see large gains. Curtana of course compounds this by perma-solving any accuracy problems and giving you much more freedom to pursue str and mind in slots that don't compete with enmity as well as rewarding further increases by raising the cap.

The primary issue with Garuda weapons is having the gear to back it up. It may well be the case that for someone who can only reach say 330/300 with some attack power melds or what ahve you will find more use in a Skirmish axe. But if you can hit the cap for Garuda (350/315) then it will win for all tanking situations. As a side note, its not particularly hard to do this either. (My war is 350/322 and 100 enmity with DL boots and some solid body and glove melds). Also, any accuracy on the weapon gives you a ring slot back. I wear no accuracy rings or materia and simply eat salt codd buffs. I have no issues on any of the game's current bosses (as both pld and war) including Garuda and Raven.


Also Judge. I used my garuda's gaze to test mind cap. I could only hit the 308, 312, adn 315 points. I spent several hours farming darksteel currency on the group to the SW of the stronghold. My average damage increased sligtly from 308 to 312, but did not increase from 312 to 315 (used flat blade and phalanx). I personally think the cap for a 40.00 dps weapon is 310 for secondary stat, as this agrees with separate testing done by a monk in my shell.