They used the Apple's Xcode compiler for the M1 and Intel's ICC compiler for the Alder Lake. How is that "shenanigans"? And just in case you didn't now, LLVM is not a compiler but a compiler framework (and, BTW, ICC also uses LLVM now). Apple has every opportunity to optimize the Xcode compiler (and particularly their back end for the M1), so it's only fair that Intel is given the same opportunity.
Hey Rigby
I was aware (already) that both Intels latest ICC and Apples shipped XCode are based on the LLVM compiler and toolchain frameworks - I do appreciate that you checked that I understood that. In this day and age on the internet it’s really important that we are all on the same page
However, from the tone of your response to me, I get a sense that very detailed specificity is important to you - and I hope that in the spirit of seeking that specificity, you won’t be insulted when I share that you are wrong in the assertion that “LLVM is not a compiler” !
From the LLVM project (highlighted screenshot also provided below for your consideration:
The LLVM Project is a collection of modular and reusable compiler and toolchain technologies. Despite its name, LLVM has little to do with traditional virtual machines. The name "LLVM" itself is not an acronym; it is the full name of the project.
I mentioned Intel’s LLVM backed ICC (and referenced it to ICC) because the version that Intel ships does not generate similar binary output to opensource CLang LLVM projects. Intel even mentions that their version of the LLVM backed ICC includes optimizations and enhancements that do not make their way upstream to the LLVM project itself (again see screenshots below).
Please do not take my word for it, please have a read of Intels’ own language on this very matter…
The benefits of adopting LLVM are numerous. I will offer advice for upgrading from our classic compilers to our LLVM-based compilers. We are committed to making this as seamless as possible while yielding numerous benefits for developers who use the Intel compilers. Benefits of adopting LLVM...
community.intel.com
I’ve attached and highlighted interesting commentary from Intel themselves on this very matter (as well as screencaps of benchmark differences that Intel advertises as it relates to ICC on Spec tests in comparison with open source CLang-LLVM generated binaries (again here, it’s hard to compare because of default behavior of ICC and open source projects

even on Intel!).
For the purpose of the discussion here, the Intel C Compiler defaults to -O2 with autovectorization, loop unrolling etc…
(see again the attached documentation screenshots for your interest on ICC and default optimization levels) - default is -O2 or FAST.
This level of optimization in ICC would be the equivalent of an XCode project being
explicitly run with -O3 for autovectorization, loop unrolling etc…
The default projects in XCode are -debug (for fast recompilation, debug etc…) and -release (this trades smaller binary file sizes for release).
Why I am sharing this?
Without much more information provided by Intel and beyond what we
already know are the significant differences in default optimization level philosophy by Intel and Apple - it is VERY hard to truly say know whether we are seeing a test where every effort was made to make the test as apples:apples or not. I’m looking forward to independent third parties to disect and reproduce (as I was also anxiously awaiting anandtechs breakdown of M1 Max on release).
Most importantly to what I am interested in these days - I’ll anxiously await a benchmark where laptop alderlake exports an H.265 40% faster than M1 Max or compiles a large .NET Core or Angular project faster than M1 Max …. while on battery ….
Hope some of what I shared here was helpful or at least makes for an interesting conversation talking point
Have a really great day.