Visual Studio 2022 17.11 brings new optimizations, intrinsics, features, and improvements to the MSVC backend. Check out the highlights below: 

• Performance improvements and additional functionality for all architectures: 

• • The SLP vectorizer can now recognize when vectors need to be permuted and when elements of a vector are defined by different operations. 

• • Updated corruption handling in DIA. 

• • Addressed bugs for debug records emitted during /LTCG:INCREMENTAL. 

• • /OPT:ICF is now more C++ compliant. 

• • Fixed race conditions in linker while emitting debug information. 

• • Added a new dead code elimination phase to reduce binary bloat. 

• • Added a strength reduction phase to replace expensive instructions which equivalent cheaper instructions in the lexical loop optimizer. 

• • Improved /guard:cf compilation speed for many try-catch regions. 

• • Process clang-cl OBJ files faster 

• Performance improvements and additional functionality for ARM64: 

• • Fixed pointer parameter alignment during conversion of by-value parameters to by-reference. 

• • Support popcnt instructions, __popcnt, popcnt16, and popcnt64, thanks to our friends at ARM. 

• • The vectorizer now recognizes and uses ARM64 intrinsics, when possible. 

• • Generate better instructions for VCREATE using FMOV, thanks to our friends at ARM. 

• • Add RBIT instrinics, _bitrev and _bitrev64. 

• • Support expanding memcmp when comparison length is constant, thanks to our friends at ARM. 

• • Adds support for Load-Acquire RCpc instructions v2 (FEAT_LRCPC2), /feature:rcpc2 

• • Add support for CMP, XZR, and Xm in disassembler, thanks to our friends at ARM. 

• • Generate better code for bitfield selection and assignment, thanks to our friends at ARM. 

• Improvements to ARM64EC: 

• • Considers hybrid-patchable functions during long-branch optimizations /OPT:LBR. 

• • Now emits virtual functions with __declspec(hybrid_patchable). 

• • Fixed imports with fast-forward sequences. 

• • Thunk generation in C++ files, using _Arm64XGenerateThunk(), does not require a prototype. Still requires #include <intrin.h>. 

• • Fixed pointer parameter alignment during conversion of by-value parameter to by-reference. 

• • Sped up delay loading a DLL by enhancing page protection.

• Performance improvements and additional functionality for x86 and x64: 

• • Optimized vector reduction for E-Core CPUs, thanks to our friends at Intel. 

• • Fix float conversions on x86 for standard compliance, thanks to our friends at Intel. 

• • More loops get vectorized, thanks to our friends at AMD. 

• • New functionality on x64: 

• • • Optimize FMA generation for blended code, removes /favor:ATOM restriction , thanks to our friends at Intel. 

• • • Add support for USER_MSR intrinsics, urdmsr and uwrmsr, thanks to our friends at Intel. 

• • • Add instructions in assembler and disassembler, thanks to our friends at Intel: 

• • • • Flexible Return and Event Delivery (FRED) includes ERETS and ERETU 

• • • • LKGS which is required for FRED 

• • • Enable vector loop unroller to be more aggressive, as required, thanks to our friends at AMD. 

• • • Add support for saturating add and subtract intrinsics:

• • • • _sat_add_i8 

• • • • _sat_add_i16 

• • • • _sat_add_i32 

• • • • _sat_add_i64 

• • • • _sat_add_u8 

• • • • _sat_add_u16 

• • • • _sat_add_u32 

• • • • _sat_add_u64 

• • • • _sat_sub_i8 

• • • • _sat_sub_i16 

• • • • _sat_sub_i32 

• • • • _sat_sub_i64 

• • • • _sat_sub_u8 

• • • • _sat_sub_u16 

• • • • _sat_sub_u32 

• • • • _sat_sub_u64 

Do you want to experience the new improvements in the C++ backend? Please download the latest Visual Studio 2022 and give it a try! Any feedback is welcome. We can be reached via the comments below, Developer Community, X  (@VisualC), or email at visualcpp@microsoft.com. 

Stay tuned for more information on updates to the latest Visual Studio.