The open-source GCC Compiler saw a new submission today for support, but it is not for Intel's Sapphire Rapids Xeon CPUs. It is for two generations from Sapphire Rapids, also known as Granite Rapids.
Intel submits Granite Rapids Xeon CPU support to the GCC Compiler one year before the full release
In January 2023, it is anticipated that Intel will launch its newest Xeon Scalable Saphhire Rapids (January 10th, to be exact). We will see Sapphire Rapids, which Emerald Rapids will follow. After Emerald Rapids is the release of the Sapphire Rapids series, so it is beneficial for Intel to work on the GCC Compiler this early for future support. This will allow for instructions for the processor and other implementations to be added over the next year.
The new support is in the way of a patch for the future Intel Granite Rapids Xeon CPUs, adding the "-march=graniterapids" code to the compiler. The enablement will target varying extensions listed below:
- MOVBE
- MMX
- POPCNT
- CX16
- SAHF
- FXSR
- XSAVE
- PCLMUL
- FSGSBASE
- RDRND
- F16C
- BMI/BMI2
- LZCNT
- FMA
- HLE
- RDSEED
- ADCX
- PREFETCHW
- AES
- CLFLUSHOPT
- XSAVEC/XSAVES
- SGX
- PKU
- SHA
- GFNI
- VAES
- VPCLMULQDQ
- RDPID
- PCONFIG
- WBNOINVD
- CLWB
- MOVDIRI/MOVDIR64B
- ENQCMD
- CLDEMOTE
- PTWRITE
- WAITPKG
- SERIALIZE
- TSXLDTRK
- UINTR
- PREFETCHI
- SSE/SSE2/SSE3/SSSE3/SSE4.1/SSE4.2
- AMX-BF16/AMX-TILE/AMX-INT8/AMX-FP16
- AVX/AVX2/AVX512F/AVX512VL/AVX512BW/AVX512DQ/AVX512CD/AVX512VBMI/AVX512IFMA/AVX512BITALG/AVX512VP2INTERSECT/AVX-VNNI/AVX512FP16/AVX512BF16/AVX512VNNI/AVX512VPOPCNTDQ/AVX512VBMI2
A large majority of this enablement into the GCC Compiler follows the recent Grand Ridge patchwork and Sierra Forest enablement merging into the Intel AMX-FP16 and Intel CMPccXADD support, which will be in its thirteenth version, reports Michael Larabel of the website Phoronix.
It is unknown if more instruction sets will be added over the next month for GCC, but it is still quite early for full implementation, so we will continue to follow more additions as they develop.
Preparing future architectures is standard for Intel but not AMD. Zen 4 implements were just added to the upcoming GCC version last month. AMD's future implementations into GCC compiler and other open-source abilities are still around the Zen 3/4 integrations.
Developers interested in the specifics of the programming instructions can download the "Intel Architecture Instruction Set Extensions Programming Reference," which will download a PDF of the materials to explain each extension in more detail in a 230-page document.
News Sources: Phoronix, GCC Compiler request
