It might help people with CC 1.1 cards if tasks were allocated on a performance basis. So for example, if a Toni task was more likely to fail on a given card, there would be less chance of receiving it.
Again I would encourage GPUGrid to Trailblaze rather than follow the beaten path.