Ivy Bridge isn't an issue for AVX used in newer versions, but agreed, the p95 should be version 26.6 small fft for best gaming temp comparison baseline.
They haven't done anything decent to IBT in forever, I guess they figure it ain't broke...
Problem with IBT is its inaccuracy at what they deem as a torture test for the cpu, then throw in linpack (uses goodly amounts of ram to various degrees) so you don't get a clean 100% cpu load, you get anything from @80% to 100%, with ram usage, and temps rise and fall, so you can't even say 'my cpu runs at x °C, because that x°C could be anything from 80% to 120% of the cpu.
IBT is great for 1 thing only, pass/fail on cpu cooling. Kinda sucks at letting you know how much it passed or failed by.