This new version takes inspiration from the C implementation of the benchmark, but instead of explicitly using SIMD operations which can't be done on stable, it instead arranges everything the same way and leaves the actual vectorization up to LLVM.
In addition to the ~20% speed gains (see below), this PR also adds some general niceties which showcase the language a little bit: a `Vec3` type to cut down on `(x, y, z)` repetition, using `while let` instead of `loop-if-break`, iterator adapters instead of for loops etc.
Here are the times in seconds of 10 runs each on my workstation:
```
before: 6.254, 6.260, 6.263, 6.264, 6.265, 6.267, 6.334, 6.341, 6.441, 6.509
before-min: 6.254
before-median: 6.266
before-max: 6.509
after: 4.823, 4.824, 4.826, 4.827, 4.837, 4.839, 4.881, 4.959, 4.990, 5.377
after-min: 4.823
after-median: 4.838
after-max: 5.377
gcc: 4.674, 4.676, 4.680, 4.682, 4.695, 4.696, 4.701, 4.708, 4.794, 5.297
gcc-min: 4.674
gcc-median: 4.696
gcc-max: 5.297
```
On my i7 laptop the speed up is less impressive, from ~5.4s to ~4.7s, but still significant. On my Vultr VPS the numbers look closer to the workstation results. Surprisingly my laptop beats both office workstation and VPS...
It's not very common to store `Wrapping` values, but I kept wrapping and unwrapping a hash value when I taking it out of a struct to do operations on it. I couldn't store the hash as `Wrapping<u64>` because I wanted to be able to `#[derive(Default)]` for the struct.
At any rate, it feels to me that `Wrapping<T>` should implement pretty much everything `T` does. I left out `#[derive(Hash)]` since I'd be hard pressed to find a use case and wanted to avoid the extra generated code, but maybe I should add that too?
These common traits were left off originally by accident from these smart
pointers, and a past attempt (#26008) to add them was later reverted (#26160)
due to unexpected breakge (#26096) occurring. The specific breakage in worry is
the meaning of this return value changed:
let a: Box<Option<T>> = ...;
a.as_ref()
Currently this returns `Option<&T>` but after this change it will return
`&Option<T>` because the `AsRef::as_ref` method shares the same name as
`Option::as_ref`. A [crater report][crater] of this change, however, has shown
that the fallout of this change is quite minimal. These trait implementations
are "the right impls to add" to these smart pointers and would enable various
generalizations such as those in #27197.
[crater]: https://gist.github.com/anonymous/0ba4c3512b07641c0f99
This commit is a breaking change for the above reasons mentioned, and the
mitigation strategies look like any of:
Option::as_ref(&a)
a.as_ref().as_ref()
(*a).as_ref()
The diff can hopefully speak for itself. Regardless: this chapter of the book contained a sentence where "the" was mistakenly repeated twice. In this same section, there was a comma separating two sentences where a period should have been. This PR fixes both issues.
In #28864, @aarzee submitted some whitespace fixes. I r+'d it. But
@retp998 noticed[1] that this file is explicitly a test of this kind of
whitespace, and so I shouldn't have changed it. This restores the old
line endings.
1: https://github.com/rust-lang/rust/pull/28864#discussion_r41332279
This is to address issue #28803 by improving some of the references to closures, to explain what they are more clearly, while hopefully still being concise.
r? @steveklabnik
This commit contains some of the changes proposed by a rustfmt invocation,
chosen based on the fairly non-deterministic metric of how much I liked the
change. I expect we will run rustfmt on this crate again later, probably
accepting more of its changes. For now, this is already an improvement over
the status-quo.
This commit contains some of the changes proposed by a rustfmt invocation,
chosen based on the fairly non-deterministic metric of how much I liked the
change. I expect we will run rustfmt on this crate again later, probably
accepting more of its changes. For now, this is already an improvement over
the status-quo.
Currently the explain command requires full erorr codes, complete with
the leading zeros and the E at the beginning. This commit changes that,
if you don't supply a full erorr code then the error number is padded
out to the required size and the E is added to the beginning.
This means that where previously you would need to write E0001, you can
now write 0001, 001, 01 or jsut 1 to refer to the same error.
In #28864, @aarzee submitted some whitespace fixes. I r+'d it. But
@retp998 noticed[1] that this file is explicitly a test of this kind of
whitespace, and so I shouldn't have changed it. This restores the old
line endings.
1: https://github.com/rust-lang/rust/pull/28864#discussion_r41332279
This turned up as part of #3170. When constructing an `undef` value to
return in the error case, we were trying to get the element type of the
Rust-level value being indexed instead of the underlying array; when
indexing a slice, that's not an array and the LLVM assertion failure
reflects this.
The regression test is a lightly altered copy of `const-array-oob.rs`.
