Crate vectoreyes
source · [−]Expand description
VectorEyes is a (almost entirely) safe wrapper library around vectorized operations.
Backends
VectorEyes chooses what backend to execute vector operations with at compile-time.
AVX2
CPUs that support the AVX, AVX2, SSE4.1, AES, SSE4.2, and PCLMULQDQ features will
use the AVX2 backend.
In addition, we have embedded specific latency numbers for:
skylakeskylake-avx512cascadelakeznver1znver2znver3
As a result, vectoreyes will be more efficient on these platforms. You can add specific
latency numbers for more targets in avx2.py.
Scalar
At the moment, this is the only alternative to the AVX2 backend. It is not particularly
optimized.
Cargo Configuration
Native CPU Setup
Compile on the machine that you’ll be running your code on, and add the
following to your .cargo/config file:
[build]
rustflags = ["-C", "target-cpu=native", "--cfg=vectoreyes-target-cpu-native"]
rustdocflags = ["-C", "target-cpu=native", "--cfg=vectoreyes-target-cpu-native"]Specific CPU Selection
If you want to compile for some specific CPU
[build]
rustflags = ["-C", "target-cpu=TARGET", "--cfg=vectoreyes-target-cpu=\"TARGET\""]
rustdocflags = ["-C", "target-cpu=TARGET", "--cfg=vectoreyes-target-cpu=\"TARGET\""]Maximal Compatibility
If you do not put any of the above in your .cargo/config file, vectoreyes will always use its
scalar backend, which does not use vector instructions.
NOTE: many functions are currently missing from this library. Please consult the Intel documentation to see if a non-implemented intrinsic would more directly accomplish your goal, and we can add it!
Modules
Structs
[i8; 16] as a vector.[i8; 32] as a vector.[i16; 8] as a vector.[i16; 16] as a vector.[i32; 4] as a vector.[i32; 8] as a vector.[i64; 2] as a vector.[i64; 4] as a vector.[u8; 16] as a vector.[u8; 32] as a vector.[u16; 8] as a vector.[u16; 16] as a vector.[u32; 4] as a vector.[u32; 8] as a vector.[u64; 2] as a vector.[u64; 4] as a vector.Enums
Constants
Traits
[T; Self::Lanes].