add swap_bytes

[TheIronBorn]

We could add some llvm.bswap implementations.

https://godbolt.org/g/jH2FCR suggests this could be better.

[TheIronBorn]

I can't find a decent way to import the shuffle intrinsics

[gnzlbg]

You just need to do like x86 does and use the llvm intrinsics directly, they are in the llvm module.

[TheIronBorn]

I'm trying use coresimd::simd_llvm::*; but getting Could not find simd_llvm in coresimd (I closed this because I messed up a rebase, I'll open another later)

[TheIronBorn]

Using what I can see as the exact same technique used in codegen/mask_reductions:

#![allow(unused)]use coresimd::simd::*;pub(crate)traitSwapBytes{fnswap_bytes(self) -> Self;}implSwapBytesforu8x2{fnswap_bytes(self) -> Self{use coresimd::simd_llvm::simd_shuffle2;let shuffled:u8x2 = unsafe{simd_shuffle2(self,self,[1,0])};
u8x2::from_bits(shuffled)}}

Gets

error[E0432]: unresolved import `coresimd::simd_llvm`
--> crates/coresimd/tests/../../../coresimd/ppsv/codegen/swap_bytes.rs:14:23
|
14 | use coresimd::simd_llvm::simd_shuffle2;
| ^^^^^^^^^ Could not find `simd_llvm` in `coresimd`

Not sure what's going on.

[gnzlbg]

If you update or send a PR it is easier for me to look at it.

[TheIronBorn]

I found a solution. The weird crate structure was throwing me. Pull coming soon

[gnzlbg]

what was the problem?

[TheIronBorn]

Macro scopes combined with importing the crate modules by path.

…

Sent from my mobile device

On Jul 6, 2018, at 10:11, gnzlbg ***@***.***> wrote: what was the problem? — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub, or mute the thread.

[TheIronBorn]

We could add some llvm.bswap implementations.

https://godbolt.org/g/jH2FCR suggests this could be better.

[TheIronBorn]

I can't find a decent way to import the shuffle intrinsics

[gnzlbg]

You just need to do like x86 does and use the llvm intrinsics directly, they are in the llvm module.

[TheIronBorn]

I'm trying use coresimd::simd_llvm::*; but getting Could not find simd_llvm in coresimd (I closed this because I messed up a rebase, I'll open another later)

[TheIronBorn]

Using what I can see as the exact same technique used in codegen/mask_reductions:

#![allow(unused)]use coresimd::simd::*;pub(crate)traitSwapBytes{fnswap_bytes(self) -> Self;}implSwapBytesforu8x2{fnswap_bytes(self) -> Self{use coresimd::simd_llvm::simd_shuffle2;let shuffled:u8x2 = unsafe{simd_shuffle2(self,self,[1,0])};
u8x2::from_bits(shuffled)}}

Gets

error[E0432]: unresolved import `coresimd::simd_llvm`
--> crates/coresimd/tests/../../../coresimd/ppsv/codegen/swap_bytes.rs:14:23
|
14 | use coresimd::simd_llvm::simd_shuffle2;
| ^^^^^^^^^ Could not find `simd_llvm` in `coresimd`

Not sure what's going on.

[gnzlbg]

If you update or send a PR it is easier for me to look at it.

[TheIronBorn]

I found a solution. The weird crate structure was throwing me. Pull coming soon

[gnzlbg]

what was the problem?

[TheIronBorn]

Macro scopes combined with importing the crate modules by path.

…

Sent from my mobile device

On Jul 6, 2018, at 10:11, gnzlbg ***@***.***> wrote: what was the problem? — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub, or mute the thread.

[TheIronBorn]

We could add some llvm.bswap implementations.

https://godbolt.org/g/jH2FCR suggests this could be better.

[TheIronBorn]

I can't find a decent way to import the shuffle intrinsics

[gnzlbg]

You just need to do like x86 does and use the llvm intrinsics directly, they are in the llvm module.

[TheIronBorn]

I'm trying use coresimd::simd_llvm::*; but getting Could not find simd_llvm in coresimd (I closed this because I messed up a rebase, I'll open another later)

[TheIronBorn]

Using what I can see as the exact same technique used in codegen/mask_reductions:

#![allow(unused)]use coresimd::simd::*;pub(crate)traitSwapBytes{fnswap_bytes(self) -> Self;}implSwapBytesforu8x2{fnswap_bytes(self) -> Self{use coresimd::simd_llvm::simd_shuffle2;let shuffled:u8x2 = unsafe{simd_shuffle2(self,self,[1,0])};
u8x2::from_bits(shuffled)}}

Gets

error[E0432]: unresolved import `coresimd::simd_llvm`
--> crates/coresimd/tests/../../../coresimd/ppsv/codegen/swap_bytes.rs:14:23
|
14 | use coresimd::simd_llvm::simd_shuffle2;
| ^^^^^^^^^ Could not find `simd_llvm` in `coresimd`

Not sure what's going on.

[gnzlbg]

If you update or send a PR it is easier for me to look at it.

[TheIronBorn]

I found a solution. The weird crate structure was throwing me. Pull coming soon

[gnzlbg]

what was the problem?

[TheIronBorn]

Macro scopes combined with importing the crate modules by path.

…

Sent from my mobile device

On Jul 6, 2018, at 10:11, gnzlbg ***@***.***> wrote: what was the problem? — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub, or mute the thread.

[TheIronBorn]

We could add some llvm.bswap implementations.

https://godbolt.org/g/jH2FCR suggests this could be better.

[TheIronBorn]

I can't find a decent way to import the shuffle intrinsics

[gnzlbg]

You just need to do like x86 does and use the llvm intrinsics directly, they are in the llvm module.

[TheIronBorn]

I'm trying use coresimd::simd_llvm::*; but getting Could not find simd_llvm in coresimd (I closed this because I messed up a rebase, I'll open another later)

[TheIronBorn]

Using what I can see as the exact same technique used in codegen/mask_reductions:

#![allow(unused)]use coresimd::simd::*;pub(crate)traitSwapBytes{fnswap_bytes(self) -> Self;}implSwapBytesforu8x2{fnswap_bytes(self) -> Self{use coresimd::simd_llvm::simd_shuffle2;let shuffled:u8x2 = unsafe{simd_shuffle2(self,self,[1,0])};
u8x2::from_bits(shuffled)}}

Gets

error[E0432]: unresolved import `coresimd::simd_llvm`
--> crates/coresimd/tests/../../../coresimd/ppsv/codegen/swap_bytes.rs:14:23
|
14 | use coresimd::simd_llvm::simd_shuffle2;
| ^^^^^^^^^ Could not find `simd_llvm` in `coresimd`

Not sure what's going on.

[gnzlbg]

If you update or send a PR it is easier for me to look at it.

[TheIronBorn]

I found a solution. The weird crate structure was throwing me. Pull coming soon

[gnzlbg]

what was the problem?

[TheIronBorn]

Macro scopes combined with importing the crate modules by path.

…

Sent from my mobile device

On Jul 6, 2018, at 10:11, gnzlbg ***@***.***> wrote: what was the problem? — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub, or mute the thread.

[TheIronBorn]

We could add some llvm.bswap implementations.

https://godbolt.org/g/jH2FCR suggests this could be better.

[TheIronBorn]

I can't find a decent way to import the shuffle intrinsics

[gnzlbg]

You just need to do like x86 does and use the llvm intrinsics directly, they are in the llvm module.

[TheIronBorn]

I'm trying use coresimd::simd_llvm::*; but getting Could not find simd_llvm in coresimd (I closed this because I messed up a rebase, I'll open another later)

[TheIronBorn]

Using what I can see as the exact same technique used in codegen/mask_reductions:

#![allow(unused)]use coresimd::simd::*;pub(crate)traitSwapBytes{fnswap_bytes(self) -> Self;}implSwapBytesforu8x2{fnswap_bytes(self) -> Self{use coresimd::simd_llvm::simd_shuffle2;let shuffled:u8x2 = unsafe{simd_shuffle2(self,self,[1,0])};
u8x2::from_bits(shuffled)}}

Gets

error[E0432]: unresolved import `coresimd::simd_llvm`
--> crates/coresimd/tests/../../../coresimd/ppsv/codegen/swap_bytes.rs:14:23
|
14 | use coresimd::simd_llvm::simd_shuffle2;
| ^^^^^^^^^ Could not find `simd_llvm` in `coresimd`

Not sure what's going on.

[gnzlbg]

If you update or send a PR it is easier for me to look at it.

[TheIronBorn]

I found a solution. The weird crate structure was throwing me. Pull coming soon

[gnzlbg]

what was the problem?

[TheIronBorn]

Macro scopes combined with importing the crate modules by path.

…

Sent from my mobile device

On Jul 6, 2018, at 10:11, gnzlbg ***@***.***> wrote: what was the problem? — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub, or mute the thread.

[TheIronBorn]

We could add some llvm.bswap implementations.

https://godbolt.org/g/jH2FCR suggests this could be better.

[TheIronBorn]

I can't find a decent way to import the shuffle intrinsics

[gnzlbg]

You just need to do like x86 does and use the llvm intrinsics directly, they are in the llvm module.

[TheIronBorn]

I'm trying use coresimd::simd_llvm::*; but getting Could not find simd_llvm in coresimd (I closed this because I messed up a rebase, I'll open another later)

[TheIronBorn]

Using what I can see as the exact same technique used in codegen/mask_reductions:

#![allow(unused)]use coresimd::simd::*;pub(crate)traitSwapBytes{fnswap_bytes(self) -> Self;}implSwapBytesforu8x2{fnswap_bytes(self) -> Self{use coresimd::simd_llvm::simd_shuffle2;let shuffled:u8x2 = unsafe{simd_shuffle2(self,self,[1,0])};
u8x2::from_bits(shuffled)}}

Gets

error[E0432]: unresolved import `coresimd::simd_llvm`
--> crates/coresimd/tests/../../../coresimd/ppsv/codegen/swap_bytes.rs:14:23
|
14 | use coresimd::simd_llvm::simd_shuffle2;
| ^^^^^^^^^ Could not find `simd_llvm` in `coresimd`

Not sure what's going on.

[gnzlbg]

If you update or send a PR it is easier for me to look at it.

[TheIronBorn]

I found a solution. The weird crate structure was throwing me. Pull coming soon

[gnzlbg]

what was the problem?

[TheIronBorn]

Macro scopes combined with importing the crate modules by path.

…

Sent from my mobile device

On Jul 6, 2018, at 10:11, gnzlbg ***@***.***> wrote: what was the problem? — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub, or mute the thread.

[TheIronBorn]

We could add some llvm.bswap implementations.

https://godbolt.org/g/jH2FCR suggests this could be better.

[TheIronBorn]

I can't find a decent way to import the shuffle intrinsics

[gnzlbg]

You just need to do like x86 does and use the llvm intrinsics directly, they are in the llvm module.

[TheIronBorn]

I'm trying use coresimd::simd_llvm::*; but getting Could not find simd_llvm in coresimd (I closed this because I messed up a rebase, I'll open another later)

[TheIronBorn]

Using what I can see as the exact same technique used in codegen/mask_reductions:

#![allow(unused)]use coresimd::simd::*;pub(crate)traitSwapBytes{fnswap_bytes(self) -> Self;}implSwapBytesforu8x2{fnswap_bytes(self) -> Self{use coresimd::simd_llvm::simd_shuffle2;let shuffled:u8x2 = unsafe{simd_shuffle2(self,self,[1,0])};
u8x2::from_bits(shuffled)}}

Gets

error[E0432]: unresolved import `coresimd::simd_llvm`
--> crates/coresimd/tests/../../../coresimd/ppsv/codegen/swap_bytes.rs:14:23
|
14 | use coresimd::simd_llvm::simd_shuffle2;
| ^^^^^^^^^ Could not find `simd_llvm` in `coresimd`

Not sure what's going on.

[gnzlbg]

If you update or send a PR it is easier for me to look at it.

[TheIronBorn]

I found a solution. The weird crate structure was throwing me. Pull coming soon

[gnzlbg]

what was the problem?

[TheIronBorn]

Macro scopes combined with importing the crate modules by path.

…

Sent from my mobile device

On Jul 6, 2018, at 10:11, gnzlbg ***@***.***> wrote: what was the problem? — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub, or mute the thread.

[TheIronBorn]

We could add some llvm.bswap implementations.

https://godbolt.org/g/jH2FCR suggests this could be better.

[TheIronBorn]

I can't find a decent way to import the shuffle intrinsics

[gnzlbg]

You just need to do like x86 does and use the llvm intrinsics directly, they are in the llvm module.

[TheIronBorn]

I'm trying use coresimd::simd_llvm::*; but getting Could not find simd_llvm in coresimd (I closed this because I messed up a rebase, I'll open another later)

[TheIronBorn]

Using what I can see as the exact same technique used in codegen/mask_reductions:

#![allow(unused)]use coresimd::simd::*;pub(crate)traitSwapBytes{fnswap_bytes(self) -> Self;}implSwapBytesforu8x2{fnswap_bytes(self) -> Self{use coresimd::simd_llvm::simd_shuffle2;let shuffled:u8x2 = unsafe{simd_shuffle2(self,self,[1,0])};
u8x2::from_bits(shuffled)}}

Gets

error[E0432]: unresolved import `coresimd::simd_llvm`
--> crates/coresimd/tests/../../../coresimd/ppsv/codegen/swap_bytes.rs:14:23
|
14 | use coresimd::simd_llvm::simd_shuffle2;
| ^^^^^^^^^ Could not find `simd_llvm` in `coresimd`

Not sure what's going on.

[gnzlbg]

If you update or send a PR it is easier for me to look at it.

[TheIronBorn]

I found a solution. The weird crate structure was throwing me. Pull coming soon

[gnzlbg]

what was the problem?

[TheIronBorn]

Macro scopes combined with importing the crate modules by path.

…

Sent from my mobile device

On Jul 6, 2018, at 10:11, gnzlbg ***@***.***> wrote: what was the problem? — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub, or mute the thread.