From 59fee2083663565459cefa676ae449a5533b10cf Mon Sep 17 00:00:00 2001 From: Alex Crichton Date: Mon, 23 Sep 2019 14:27:23 -0700 Subject: [PATCH] Remove long-outdated `QUESTIONS.md` This is quite an old file now at this point! --- library/stdarch/QUESTIONS.md | 31 ------------------------------- 1 file changed, 31 deletions(-) delete mode 100644 library/stdarch/QUESTIONS.md diff --git a/library/stdarch/QUESTIONS.md b/library/stdarch/QUESTIONS.md deleted file mode 100644 index bc04ea051ca3..000000000000 --- a/library/stdarch/QUESTIONS.md +++ /dev/null @@ -1,31 +0,0 @@ -This program emits `psllw` instead of `pslliw` instruction. `pslliw` should be -usable here since `imm8` is a constant: - -```rust -extern crate stdarch; - -use std::env; -use stdarch as s; - -fn main() { - let imm8: i32 = env::args().nth(1).unwrap().parse().unwrap(); - let a = s::i16x8::new(imm8 as i16, 0, 0, 0, 0, 0, 0, 0); - println!("{:?}", s::_mm_slli_epi16(a, 4)); -} -``` - ---- - -LLVM types the `count` parameter in `_mm_sll_epi16` as `i16x8`, but the right -type is probably `i64x2`. - ---- - -What is the deal with the ucomi f64 comparison functions in SSE2? Clang's -headers indicate some specific behavior with NAN, but I can't seem to reproduce -it. Intel's official docs are very vague. - ---- - -`_mm256_blendv_pd` takes a mask parameter with type `f64x4`, but the -documentation seems to indicate that it is a bit vector. What's going on?