Now it is clearly explained that "x86-64-vnni256" requires full
support of AVX512-VNNI, but only 256-bit operands are used.
closes https://github.com/official-stockfish/Stockfish/pull/4427
No functional change
Matthew Lai (matthewlai)
Matthew Sullivan (Matt14916)
Max A. (Disservin)
Matthew Lai (matthewlai)
Matthew Sullivan (Matt14916)
Max A. (Disservin)
+Maxim Masiutin (maximmasiutin)
Maxim Molchanov (Maxim)
Michael An (man)
Michael Byrne (MichaelB7)
Maxim Molchanov (Maxim)
Michael An (man)
Michael Byrne (MichaelB7)
# avx2 = yes/no --- -mavx2 --- Use Intel Advanced Vector Extensions 2
# avxvnni = yes/no --- -mavxvnni --- Use Intel Vector Neural Network Instructions AVX
# avx512 = yes/no --- -mavx512bw --- Use Intel Advanced Vector Extensions 512
# avx2 = yes/no --- -mavx2 --- Use Intel Advanced Vector Extensions 2
# avxvnni = yes/no --- -mavxvnni --- Use Intel Vector Neural Network Instructions AVX
# avx512 = yes/no --- -mavx512bw --- Use Intel Advanced Vector Extensions 512
-# vnni256 = yes/no --- -mavx512vnni --- Use Intel Vector Neural Network Instructions 256
+# vnni256 = yes/no --- -mavx256vnni --- Use Intel Vector Neural Network Instructions 512 with 256bit operands
# vnni512 = yes/no --- -mavx512vnni --- Use Intel Vector Neural Network Instructions 512
# neon = yes/no --- -DUSE_NEON --- Use ARM SIMD architecture
# dotprod = yes/no --- -DUSE_NEON_DOTPROD --- Use ARM advanced SIMD Int8 dot product instructions
# vnni512 = yes/no --- -mavx512vnni --- Use Intel Vector Neural Network Instructions 512
# neon = yes/no --- -DUSE_NEON --- Use ARM SIMD architecture
# dotprod = yes/no --- -DUSE_NEON_DOTPROD --- Use ARM advanced SIMD Int8 dot product instructions
@echo ""
@echo "Supported archs:"
@echo ""
@echo ""
@echo "Supported archs:"
@echo ""
- @echo "x86-64-vnni512 > x86 64-bit with vnni support 512bit wide"
- @echo "x86-64-vnni256 > x86 64-bit with vnni support 256bit wide"
+ @echo "x86-64-vnni512 > x86 64-bit with vnni 512bit support"
+ @echo "x86-64-vnni256 > x86 64-bit with vnni 512bit support, limit operands to 256bit wide"
@echo "x86-64-avx512 > x86 64-bit with avx512 support"
@echo "x86-64-avx512 > x86 64-bit with avx512 support"
- @echo "x86-64-avxvnni > x86 64-bit with avxvnni support"
+ @echo "x86-64-avxvnni > x86 64-bit with vnni 256bit support"
@echo "x86-64-bmi2 > x86 64-bit with bmi2 support"
@echo "x86-64-avx2 > x86 64-bit with avx2 support"
@echo "x86-64-sse41-popcnt > x86 64-bit with sse41 and popcnt support"
@echo "x86-64-bmi2 > x86 64-bit with bmi2 support"
@echo "x86-64-avx2 > x86 64-bit with avx2 support"
@echo "x86-64-sse41-popcnt > x86 64-bit with sse41 and popcnt support"