With the updated TBM v1.11 and new --xintensity recommended setting (--xintensity 224), I am getting 64+ MH per card (Total 322.14 MH) and with 0 stales/rejects so far (93+ mins running).

Looking good.
The default --xintesity on nvidia is 256 in v1.11 but 224 procude less stale shares, so we might change the default in v1.12.