So Fritz wasnt relevant because too old, and now that s 7 Zip that is not that good after all.
I guess that whatever bench doesnt produce the results that you re imagining as being the good ones cant be a real, a true, a reliable bench...
An integer version of 3D Particle Movement would suit perhaps.?.
I'm not saying its not relevant simply that it only shows one aspect of integer performance. Any single benchmark is insufficient. Doesn't matter what it is.
7-zip is good. Its good at testing memory latency (which is why its RAM sensitive) and operation latency. It is not a good benchmark of throughput or general integer performance.
As far as a better benchmark goes, I have already said that x264/265 is better. Obviously it is not the 'one benchmark to rule them all' but its better.
3D-PM is meh. Looks like it has a number of problems.
There s no such thing as FLOPS in a task using integer code only, for this reason in 7 ZIP the metric is in MIPS.
Whoops mistyped.