After using MMX intrinsics the FPU must be
cleared by using _mm_empty() before interfacing
with any x87 code.
This commit makes the simd_test macro automatically do that
when one of the features enabled is "mmx".
This commit:
* renames `coresimd` to `core_arch` and `stdsimd` to `std_detect`
* `std_detect` does no longer depend on `core_arch` - it is a freestanding
`no_std` library that only depends on `core` - it is renamed to `std_detect`
* moves the top-level coresimd and stdsimd directories into the appropriate
crates/... directories - this simplifies creating crate.io releases of these crates
* moves the top-level `coresimd` and `stdsimd` sub-directories into their
corresponding crates in `crates/{core_arch, std_detect}`.
This intrinsic isn't actually specified by Intel, but it's something
gated with CPUID and can otherwise be a useful thing to have when
building primitives!
There exists an `AtomicU128` type in the standard library but it's only
exposed currently (and it's unstable) when a platform fully supports
128-bit atomics. The x86_64 architecture does not support it *unless*
the `cmpxchg16b` instruction is available, and it isn't always available!
This commit is also a proposal for how we can include support for
128-bit atomics in the standard library on relevant platforms. I'm
thinking that we'll expose this one low-level intrinsic in
`std::arch::x86_64`, and then if desired a crate on crates.io can build
`AtomicU128` from this API.
In any case this is all unstable regardless!
This commit implements automatic verification of implement ARM/AArch64
intrinsics. Or it's at least a start! This downloads a snapshot of ARM's
[online documentation][docs] and implements necessary logic to parse
that and use it to verify all the intrinsics. Almost everything
checked out A-OK but a few minor tweaks were needed to the neon
intrinsics and the crc ones needed some renaming.
[docs]: https://developer.arm.com/technologies/neon/intrinsics
This commit updates stdsimd's codegen to match Clang's for the
`_mm512_abs_epi32` intrinsic (and masked versions) which doesn't use any
LLVM intrinsic calls, but rather raw SIMD operations.
These are built on top of the new `simd_select_bitmask` intrinsic
introduced recently to the compiler!
The MinGW builds already don't work and the MSVC ones are now timing out
and have been slow otherise, so they're being removed to reduce spurious
failures on Travis.
We historically have run single-threaded verbose tests because we were
faulting all over the place due to bugs in rustc itself, primarily
around calling conventions and passing values around. Those bugs have
all since been fixed so we should be clear to run multithreaded tests
quietly on CI nowadays!
Closes#621
* Update representation of `v128`
* Rename everything with new naming convention of underscores and no
modules/impls
* Remove no longer necessary `wasm_simd128` feature
* Remove `#[target_feature]` attributes (use `#[cfg]` instead)
* Update `assert_instr` tests
* Update some implementations as LLVM has evolved
* Allow some more esoteric syntax in `#[assert_instr]`
* Adjust the safety of APIs where appropriate
* Remove macros in favor of hand-coded implementations
* Comment out the tests for now as there's no known runtime for these
yet
This commit switches CI to running `rustfmt` on the stable compiler (as
rustfmt is stable now!). Additionally it deletes `rustfmt.toml` to
ensure we're following the same style as the rest of the ecosystem.
This commit renames the atomic intrinsics for wasm and tweaks them to
match the new naming convention of the memory intrinsics. Apart from
naming the `wake` intrinsic was renamed to `atomic_notify` (reflecting
the upstream rename) as well as updating to take/return unsigned
arguments, also reflecting the upstream spec changes.
