Skip to content

Feat - Aarch64 FEAT_FAMINMAX #1732

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 6 commits into from
Mar 4, 2025
Merged

Feat - Aarch64 FEAT_FAMINMAX #1732

Show file tree
Hide file tree
Changes from 5 commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
ae3dcb3
feat - FEAT_FAMINMAX neon intrinsics
Jamesbarford Jan 22, 2025
09d127a
Add aarch64_unstable_target_feature and relax instruction assertion
Jamesbarford Feb 24, 2025
0ec3abf
Add faminmax intrinsics to the skip list
Jamesbarford Feb 24, 2025
087f1c4
armV7 does not need faminmax flag
Jamesbarford Feb 24, 2025
6878d88
mark FEAT_FAMINMAX intrinsics as safe
Jamesbarford Feb 28, 2025
e0be20b
add unstable for faminmax
Jamesbarford Mar 3, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
96 changes: 96 additions & 0 deletions crates/core_arch/src/aarch64/neon/generated.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -821,6 +821,102 @@ pub fn vaddvq_u64(a: uint64x2_t) -> u64 {
}
unsafe { _vaddvq_u64(a) }
}
#[doc = "Multi-vector floating-point absolute maximum"]
#[doc = "[Arm's documentation](https://developer.arm.com/architectures/instruction-sets/intrinsics/vamax_f32)"]
#[inline]
#[target_feature(enable = "neon,faminmax")]
#[cfg_attr(test, assert_instr(nop))]
#[stable(feature = "neon_intrinsics", since = "1.59.0")]
pub fn vamax_f32(a: float32x2_t, b: float32x2_t) -> float32x2_t {
unsafe extern "unadjusted" {
#[cfg_attr(
any(target_arch = "aarch64", target_arch = "arm64ec"),
link_name = "llvm.aarch64.neon.famax.v2f32"
)]
fn _vamax_f32(a: float32x2_t, b: float32x2_t) -> float32x2_t;
}
unsafe { _vamax_f32(a, b) }
}
#[doc = "Multi-vector floating-point absolute maximum"]
#[doc = "[Arm's documentation](https://developer.arm.com/architectures/instruction-sets/intrinsics/vamaxq_f32)"]
#[inline]
#[target_feature(enable = "neon,faminmax")]
#[cfg_attr(test, assert_instr(nop))]
#[stable(feature = "neon_intrinsics", since = "1.59.0")]
pub fn vamaxq_f32(a: float32x4_t, b: float32x4_t) -> float32x4_t {
unsafe extern "unadjusted" {
#[cfg_attr(
any(target_arch = "aarch64", target_arch = "arm64ec"),
link_name = "llvm.aarch64.neon.famax.v4f32"
)]
fn _vamaxq_f32(a: float32x4_t, b: float32x4_t) -> float32x4_t;
}
unsafe { _vamaxq_f32(a, b) }
}
#[doc = "Multi-vector floating-point absolute maximum"]
#[doc = "[Arm's documentation](https://developer.arm.com/architectures/instruction-sets/intrinsics/vamaxq_f64)"]
#[inline]
#[target_feature(enable = "neon,faminmax")]
#[cfg_attr(test, assert_instr(nop))]
#[stable(feature = "neon_intrinsics", since = "1.59.0")]
pub fn vamaxq_f64(a: float64x2_t, b: float64x2_t) -> float64x2_t {
unsafe extern "unadjusted" {
#[cfg_attr(
any(target_arch = "aarch64", target_arch = "arm64ec"),
link_name = "llvm.aarch64.neon.famax.v2f64"
)]
fn _vamaxq_f64(a: float64x2_t, b: float64x2_t) -> float64x2_t;
}
unsafe { _vamaxq_f64(a, b) }
}
#[doc = "Multi-vector floating-point absolute minimum"]
#[doc = "[Arm's documentation](https://developer.arm.com/architectures/instruction-sets/intrinsics/vamin_f32)"]
#[inline]
#[target_feature(enable = "neon,faminmax")]
#[cfg_attr(test, assert_instr(nop))]
#[stable(feature = "neon_intrinsics", since = "1.59.0")]
pub fn vamin_f32(a: float32x2_t, b: float32x2_t) -> float32x2_t {
unsafe extern "unadjusted" {
#[cfg_attr(
any(target_arch = "aarch64", target_arch = "arm64ec"),
link_name = "llvm.aarch64.neon.famin.v2f32"
)]
fn _vamin_f32(a: float32x2_t, b: float32x2_t) -> float32x2_t;
}
unsafe { _vamin_f32(a, b) }
}
#[doc = "Multi-vector floating-point absolute minimum"]
#[doc = "[Arm's documentation](https://developer.arm.com/architectures/instruction-sets/intrinsics/vaminq_f32)"]
#[inline]
#[target_feature(enable = "neon,faminmax")]
#[cfg_attr(test, assert_instr(nop))]
#[stable(feature = "neon_intrinsics", since = "1.59.0")]
pub fn vaminq_f32(a: float32x4_t, b: float32x4_t) -> float32x4_t {
unsafe extern "unadjusted" {
#[cfg_attr(
any(target_arch = "aarch64", target_arch = "arm64ec"),
link_name = "llvm.aarch64.neon.famin.v4f32"
)]
fn _vaminq_f32(a: float32x4_t, b: float32x4_t) -> float32x4_t;
}
unsafe { _vaminq_f32(a, b) }
}
#[doc = "Multi-vector floating-point absolute minimum"]
#[doc = "[Arm's documentation](https://developer.arm.com/architectures/instruction-sets/intrinsics/vaminq_f64)"]
#[inline]
#[target_feature(enable = "neon,faminmax")]
#[cfg_attr(test, assert_instr(nop))]
#[stable(feature = "neon_intrinsics", since = "1.59.0")]
pub fn vaminq_f64(a: float64x2_t, b: float64x2_t) -> float64x2_t {
unsafe extern "unadjusted" {
#[cfg_attr(
any(target_arch = "aarch64", target_arch = "arm64ec"),
link_name = "llvm.aarch64.neon.famin.v2f64"
)]
fn _vaminq_f64(a: float64x2_t, b: float64x2_t) -> float64x2_t;
}
unsafe { _vaminq_f64(a, b) }
}
#[doc = "Bit clear and exclusive OR"]
#[doc = "[Arm's documentation](https://developer.arm.com/architectures/instruction-sets/intrinsics/vbcaxq_s8)"]
#[inline]
Expand Down
3 changes: 2 additions & 1 deletion crates/core_arch/src/lib.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -37,7 +37,8 @@
sha512_sm_x86,
x86_amx_intrinsics,
f16,
keylocker_x86
keylocker_x86,
aarch64_unstable_target_feature
)]
#![cfg_attr(test, feature(test, abi_vectorcall, stdarch_internal))]
#![deny(clippy::missing_inline_in_public_items)]
Expand Down
6 changes: 6 additions & 0 deletions crates/intrinsic-test/missing_aarch64.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,6 +19,12 @@ vrnd32xq_f64
vrnd32zq_f64
vrnd64xq_f64
vrnd64zq_f64
vamin_f32
vaminq_f32
vaminq_f64
vamax_f32
vamaxq_f32
vamaxq_f64
# LLVM select error, and missing in Clang.
vrnd32x_f64
vrnd32z_f64
Expand Down
2 changes: 1 addition & 1 deletion crates/intrinsic-test/src/main.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -240,7 +240,7 @@ fn compile_c(
let arch_flags = if target.contains("v7") {
"-march=armv8.6-a+crypto+crc+dotprod+fp16"
} else {
"-march=armv8.6-a+crypto+sha3+crc+dotprod+fp16"
"-march=armv8.6-a+crypto+sha3+crc+dotprod+fp16+faminmax"
};

let intrinsic_name = &intrinsic.name;
Expand Down
40 changes: 40 additions & 0 deletions crates/stdarch-gen-arm/spec/neon/aarch64.spec.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13973,3 +13973,43 @@ intrinsics:
- - r
- a
- FnCall: ["vdup{neon_type[1].N}", [{FnCall: [simd_extract!, [b, 'LANE as u32']]}]]

- name: "vamax{neon_type.no}"
doc: "Multi-vector floating-point absolute maximum"
arguments: ["a: {neon_type}", "b: {neon_type}"]
return_type: "{neon_type}"
attr:
- FnCall: [target_feature, ['enable = "neon,faminmax"']]
- FnCall: [cfg_attr, [test, {FnCall: [assert_instr, [nop]]}]]
- FnCall: [stable, ['feature = "neon_intrinsics"', 'since = "1.59.0"']]
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

New intrinsics should always be added as unstable and have a tracking issue.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I've added the issue here; rust-lang/rust#137933

safety: safe
types:
- float32x2_t
- float32x4_t
- float64x2_t
compose:
- LLVMLink:
name: "_vamax{neon_type.no}"
links:
- link: "llvm.aarch64.neon.famax.{neon_type}"
arch: aarch64,arm64ec

- name: "vamin{neon_type.no}"
doc: "Multi-vector floating-point absolute minimum"
arguments: ["a: {neon_type}", "b: {neon_type}"]
return_type: "{neon_type}"
attr:
- FnCall: [target_feature, ['enable = "neon,faminmax"']]
- FnCall: [cfg_attr, [test, {FnCall: [assert_instr, [nop]]}]]
- FnCall: [stable, ['feature = "neon_intrinsics"', 'since = "1.59.0"']]
safety: safe
types:
- float32x2_t
- float32x4_t
- float64x2_t
compose:
- LLVMLink:
name: "_vamin{neon_type.no}"
links:
- link: "llvm.aarch64.neon.famin.{neon_type}"
arch: aarch64,arm64ec
162 changes: 162 additions & 0 deletions intrinsics_data/arm_intrinsics.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -118681,5 +118681,167 @@
"ZIP2"
]
]
},
{
"SIMD_ISA": "Neon",
"name": "vamin_f32",
"arguments": [
"float32x2_t a",
"float32x2_t b"
],
"return_type": {
"value": "float32x2_t"
},
"Arguments_Preparation": {
"a": {
"register": "Vn.2S"
},
"b": {
"register": "Vm.2S"
}
},
"Architectures": [
"A64"
],
"instructions": [
[
"FAMIN"
]
]
},
{
"SIMD_ISA": "Neon",
"name": "vaminq_f32",
"arguments": [
"float32x4_t a",
"float32x4_t b"
],
"return_type": {
"value": "float32x4_t"
},
"Arguments_Preparation": {
"a": {
"register": "Vn.4S"
},
"b": {
"register": "Vm.4S"
}
},
"Architectures": [
"A64"
],
"instructions": [
[
"FAMIN"
]
]
},
{
"SIMD_ISA": "Neon",
"name": "vaminq_f64",
"arguments": [
"float64x2_t a",
"float64x2_t b"
],
"return_type": {
"value": "float64x2_t"
},
"Arguments_Preparation": {
"a": {
"register": "Vn.2D"
},
"b": {
"register": "Vm.2D"
}
},
"Architectures": [
"A64"
],
"instructions": [
[
"FAMIN"
]
]
},
{
"SIMD_ISA": "Neon",
"name": "vamax_f32",
"arguments": [
"float32x2_t a",
"float32x2_t b"
],
"return_type": {
"value": "float32x2_t"
},
"Arguments_Preparation": {
"a": {
"register": "Vn.2S"
},
"b": {
"register": "Vm.2S"
}
},
"Architectures": [
"A64"
],
"instructions": [
[
"FAMAX"
]
]
},
{
"SIMD_ISA": "Neon",
"name": "vamaxq_f32",
"arguments": [
"float32x4_t a",
"float32x4_t b"
],
"return_type": {
"value": "float32x4_t"
},
"Arguments_Preparation": {
"a": {
"register": "Vn.4S"
},
"b": {
"register": "Vm.4S"
}
},
"Architectures": [
"A64"
],
"instructions": [
[
"FAMAX"
]
]
},
{
"SIMD_ISA": "Neon",
"name": "vamaxq_f64",
"arguments": [
"float64x2_t a",
"float64x2_t b"
],
"return_type": {
"value": "float64x2_t"
},
"Arguments_Preparation": {
"a": {
"register": "Vn.2D"
},
"b": {
"register": "Vm.2D"
}
},
"Architectures": [
"A64"
],
"instructions": [
[
"FAMAX"
]
]
}
]