AMD AOCC 3.0 Is Here To Help Squeeze A Bit Extra Performance Out Of Zen 3
As is usually the case with new compilers/optimizations, in some workloads the new version can pay off big time... In select workloads, AOCC 3.0 was delivering much greater performance than AOCC 2.3.
The follow-up benchmarks involving GCC and upstream Clang will look at what this means for the performance relative to those compilers and if some of these gains are attributed to the shift from Clang 11 to Clang 12 base or from the AMD Zen 3 specific tuning.
The Liquid-DSP software-defined radio digital signal processing library seemed to enjoy being built under AOCC 3.0 as another example.