Update arm_intrinsics.json

crates/intrinsic-test/missing_aarch64.txt

@@ -1,8 +1,14 @@
 # Not supported by qemu (will throw illegal instruction)
-vamax_f16
-vamaxq_f16
 vamin_f16
 vaminq_f16
+vamin_f32
+vaminq_f32
+vaminq_f64
+vamax_f16
+vamaxq_f16
+vamax_f32
+vamaxq_f32
+vamaxq_f64
 vscale_f16
 vscale_f32
 vscaleq_f16
@@ -37,39 +43,6 @@ vluti2q_laneq_s8
 vluti2q_laneq_u16
 vluti2q_laneq_u8
 vluti2q_lane_u8
-
-# Not implemented in stdarch yet
-vbfdot_f32
-vbfdot_lane_f32
-vbfdot_laneq_f32
-vbfdotq_f32
-vbfdotq_lane_f32
-vbfdotq_laneq_f32
-vbfmlalbq_f32
-vbfmlalbq_lane_f32
-vbfmlalbq_laneq_f32
-vbfmlaltq_f32
-vbfmlaltq_lane_f32
-vbfmlaltq_laneq_f32
-vbfmmlaq_f32
-
-
-# Implemented in stdarch, but missing in Clang.
-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
-vrnd64x_f64
-vrnd64z_f64
 vluti4q_lane_f16_x2
 vluti4q_lane_p16_x2
 vluti4q_lane_p8
@@ -85,5 +58,27 @@ vluti4q_laneq_s8
 vluti4q_laneq_u16_x2
 vluti4q_laneq_u8
 
-# Broken in Clang
+# Not implemented in stdarch yet
+vcvtad_s32_f64
+vcvtad_u32_f64
+vcvtd_s32_f64
+vcvtd_u32_f64
+vcvtmd_s32_f64
+vcvtmd_u32_f64
+vcvtnd_s32_f64
+vcvtnd_u32_f64
+vcvtpd_s32_f64
+vcvtpd_u32_f64
+vcvts_s64_f32
+vcvts_u64_f32
+vcvtas_s64_f32
+vcvtas_u64_f32
+vcvtms_s64_f32
+vcvtms_u64_f32
+vcvtns_s64_f32
+vcvtns_u64_f32
+vcvtps_s64_f32
+vcvtps_u64_f32
+
+# Broken in Clang (fixed in https://github.com/llvm/llvm-project/pull/156029)
 vcvth_s16_f16

crates/intrinsic-test/missing_aarch64_be.txt

@@ -39,10 +39,16 @@ vusdotq_laneq_s32
 
 # Below are in common to missing_aarch64.txt
 # Not supported by qemu (will throw illegal instruction)
-vamax_f16
-vamaxq_f16
 vamin_f16
 vaminq_f16
+vamin_f32
+vaminq_f32
+vaminq_f64
+vamax_f16
+vamaxq_f16
+vamax_f32
+vamaxq_f32
+vamaxq_f64
 vscale_f16
 vscale_f32
 vscaleq_f16
@@ -77,39 +83,6 @@ vluti2q_laneq_s8
 vluti2q_laneq_u16
 vluti2q_laneq_u8
 vluti2q_lane_u8
-
-# Not implemented in stdarch yet
-vbfdot_f32
-vbfdot_lane_f32
-vbfdot_laneq_f32
-vbfdotq_f32
-vbfdotq_lane_f32
-vbfdotq_laneq_f32
-vbfmlalbq_f32
-vbfmlalbq_lane_f32
-vbfmlalbq_laneq_f32
-vbfmlaltq_f32
-vbfmlaltq_lane_f32
-vbfmlaltq_laneq_f32
-vbfmmlaq_f32
-
-
-# Implemented in stdarch, but missing in Clang.
-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
-vrnd64x_f64
-vrnd64z_f64
 vluti4q_lane_f16_x2
 vluti4q_lane_p16_x2
 vluti4q_lane_p8
@@ -125,5 +98,27 @@ vluti4q_laneq_s8
 vluti4q_laneq_u16_x2
 vluti4q_laneq_u8
 
+# Not implemented in stdarch yet
+vcvtad_s32_f64
+vcvtad_u32_f64
+vcvtd_s32_f64
+vcvtd_u32_f64
+vcvtmd_s32_f64
+vcvtmd_u32_f64
+vcvtnd_s32_f64
+vcvtnd_u32_f64
+vcvtpd_s32_f64
+vcvtpd_u32_f64
+vcvts_s64_f32
+vcvts_u64_f32
+vcvtas_s64_f32
+vcvtas_u64_f32
+vcvtms_s64_f32
+vcvtms_u64_f32
+vcvtns_s64_f32
+vcvtns_u64_f32
+vcvtps_s64_f32
+vcvtps_u64_f32
+
 # Broken in Clang
 vcvth_s16_f16

crates/intrinsic-test/missing_arm.txt

@@ -1,18 +1,3 @@
-# Not implemented in stdarch yet
-vbfdot_f32
-vbfdot_lane_f32
-vbfdot_laneq_f32
-vbfdotq_f32
-vbfdotq_lane_f32
-vbfdotq_laneq_f32
-vbfmlalbq_f32
-vbfmlalbq_lane_f32
-vbfmlalbq_laneq_f32
-vbfmlaltq_f32
-vbfmlaltq_lane_f32
-vbfmlaltq_laneq_f32
-vbfmmlaq_f32
-
 # Implemented in Clang and stdarch for A64 only even though CSV claims A32 support
 vaddq_p64
 vbsl_p64
@@ -134,18 +119,6 @@ vcvth_f16_u32
 vcvth_u32_f16
 vcvth_n_f16_u32
 vcvth_n_u32_f16
-vcvtah_s32_f16
-vcvtah_u32_f16
-vcvtmh_s32_f16
-vcvtmh_u32_f16
-vcvtpq_s16_f16
-vcvtpq_u16_f16
-vcvtp_s16_f16
-vcvtp_u16_f16
-vcvtph_s32_f16
-vcvtph_u32_f16
-vcvtnh_u32_f16
-vcvtnh_s32_f16
 vfmlsl_low_f16
 vfmlslq_low_f16
 vfmlsl_high_f16
@@ -318,3 +291,25 @@ vcvta_s16_f16
 vcvta_u16_f16
 vceqz_f16
 vceqzq_f16
+vcvtah_s32_f16
+vcvtah_u32_f16
+vcvtmh_s32_f16
+vcvtmh_u32_f16
+vcvtpq_s16_f16
+vcvtpq_u16_f16
+vcvtp_s16_f16
+vcvtp_u16_f16
+vcvtph_s32_f16
+vcvtph_u32_f16
+vcvtnh_u32_f16
+vcvtnh_s32_f16
+
+# Missing in Clang
+vusdot_laneq_s32
+vsudot_laneq_s32
+vusdotq_laneq_s32
+vsudotq_laneq_s32
+vdot_laneq_s32
+vdot_laneq_u32
+vdotq_laneq_s32
+vdotq_laneq_u32

crates/intrinsic-test/src/arm/compile.rs

@@ -15,7 +15,7 @@ pub fn build_cpp_compilation(config: &ProcessedCli) -> Option<CppCompilation> {
         .add_extra_flags(["-ffp-contract=off", "-Wno-narrowing"]);
 
     if !config.target.contains("v7") {
-        command = command.add_arch_flags(["faminmax", "lut", "sha3"]);
+        command = command.add_arch_flags(["faminmax", "lut", "sha3", "fp8"]);
     }
 
     if !cpp_compiler.contains("clang") {

crates/intrinsic-test/src/arm/config.rs

@@ -58,7 +58,9 @@ pub const PLATFORM_RUST_CFGS: &str = r#"
 #![cfg_attr(any(target_arch = "aarch64", target_arch = "arm64ec"), feature(stdarch_neon_i8mm))]
 #![cfg_attr(any(target_arch = "aarch64", target_arch = "arm64ec"), feature(stdarch_neon_sm4))]
 #![cfg_attr(any(target_arch = "aarch64", target_arch = "arm64ec"), feature(stdarch_neon_ftts))]
-#![cfg_attr(any(target_arch = "aarch64", target_arch = "arm64ec"), feature(stdarch_aarch64_jscvt))]
+#![cfg_attr(any(target_arch = "aarch64", target_arch = "arm64ec"), feature(stdarch_neon_feat_lut))]
+#![cfg_attr(any(target_arch = "aarch64", target_arch = "arm64ec"), feature(stdarch_neon_fp8))]
+#![cfg_attr(any(target_arch = "aarch64", target_arch = "arm64ec"), feature(faminmax))]
 #![feature(fmt_helpers_for_derive)]
 #![feature(stdarch_neon_f16)]