Optimized Compiler Builds Are Well Worth It For Intel Tiger Lake

For those wanting a straight-forward look at how the generic x86_64 (-march=x86-64 -O3) compares to the "-march=tigerlake -O3" results side-by-side, here is that breakdown:

Lastly is a look at comparing the -march=icelake-client results to -march=tigerlake for just the generational benefit of moving to the newer compiler target on the Core i7 1165G7 Ubuntu system:

As for the most beneficial instruction(s) added with Tiger Lake, I'll be doing some further compiler setting comparisons shortly. Going from Icelake client to Tiger Lake yielded enough uplift to make it worthwhile if you are building your own computationally-intensive software from source over using an older target for more broad software compatibility. Or if offering optimized binaries for different CPU targets or employing compiler features like function multi-versioning (FMV), Tiger Lake is certainly a level worth considering. I'll also have out some similar Tiger Lake benchmarks off the LLVM Clang compiler soon. All the data from today's testing can be found via OpenBenchmarking.org for those wanting to explore this latest Tiger Lake Linux benchmarking.